ACIX-III AQUA: EVALUATION OF ATMOSPHERIC CORRECTION PROCESSORS FOR HYPERSPECTRAL SATELLITE OVER INLAND AND COASTAL WATERS

ACIX-III AQUA
EVALUATION OF ATMOSPHERIC
CORRECTION PROCESSORS FOR
HYPERSPECTRAL SATELLITE OVER
INLAND AND COASTAL WATERS
A compa a i e analysis o a mosphe ic
co ec ion me hods applied o PRISMA and
EnMAP using in si u e lec ance da a
CNR-IREA, Milan, No embe 2025
ISBN 979-12-985355-2-7
ACIX-III Aqua epo
2
Au ho lis and a ilia ions
Claudia Gia dino1, Nima Pahle an2, Alice Fabb e o1,3, Lodo ica Panizza1, And ea Pelleg ino1,4, Ryan
Vande meulen5, Ma co Gianine o6, S e an Ad iaensen7, Jo am Ag en7, Hend ik Be ne 8, Liesbe h De
Keukelae e7, T is an Ha mel9, Thomas Heege8, Ande s Knudby9, Ka in Schenk8, F ançois S einme z10,
Sindy S e ckx7, Quin en Vanhellemon 12, Yulun Wu10, Fede ica B aga13, Vi o io E. B ando14, Ma iano
B esciani1, Ana Doglio i15, Susanne K a ze 16, Sal a o e Mangano1, Daniel Ode ma 17, Gian Ma co
Sca pa13, Thomas Sch oede 18, Mo ime We he 17, E o e Lopin o19, Ke in Alonso20, Noelle C eme 20,
Geo gia Doxani20, Fe an Gascon20
As id B ache 21,22, Ma iana A. Soppa21, A o amalala Najo o Rad ianalisoa21
and addi ionally all o he coau ho s om Soppa e al. (2024) who a e no lis ed abo e:
Maximilian B ell23, Sabine Chab illa 24, Leona do M. A. Al a ado25, Pe e Gege26, S e an Pla ne 24, Ian
Somlai-Schweige 24, Nicole Pinnel24, Simone Collela14, Die e Vans eenwegen25, Maximilian
Langhein ich26, Emiliano Ca mona26, Ma in Bachmann26, Miguel Pa o26, Sebas ian Fische 27
1 Ins i u e o Elec omagne ic Sensing o he En i onmen (IREA), CNR, Milan, I aly.
2 NASA Godda d Space Fligh Cen e , G eenbel , MD, USA.
3 Depa men o Remo e Sensing, Ta u Obse a o y, Uni e si y o Ta u, Tõ a e e, Es onia
4 Depa men o Enginee ing, Uni e si y o Sapienza, Rome, I aly.
5 NOAA Fishe ies, Sil e Sp ing, MD, USA.
6 Poli ecnico di Milano, Milan, I aly.
7 Flemish Ins i u e o Technological Resea ch (VITO), Mol, Belgium.
8 EOMAP GmbH & Co. KG, See eld, Ge many.
9 Ea h Obse a ion Uni , Magellium, Ramon ille-Sain -Agne, F ance.
10 Uni e si y o O awa, O awa, Canada.
11 HYGEOS, Lille, F ance.
12 Royal Belgian Ins i u e o Na u al Sciences (RBINS), B ussels, Belgium.
13 Ins i u e o Ma ine Sciences (ISMAR), CNR, Venice, I aly.
14 Ins i u e o Ma ine Sciences (ISMAR), CNR, Rome, I aly.
15 Ins i u o de As onomía y Física del Espacio (IAFE), Buenos Ai es, A gen ina.
16 Depa men o Ecology, En i onmen and Plan Sciences (DEEP), S ockholm Uni e si y, S ockholm,
Sweden.
17 Swiss Fede al Ins i u e o Aqua ic Science and Technology (EAWAG), Dübendo , Swi ze land.
18 Commonweal h Scien i ic and Indus ial Resea ch O ganisa ion (CSIRO), B isbane, Aus alia.
19 I alian Space Agency (ASI), Rome, I aly.
20 Eu opean Space Resea ch Ins i u e, Eu opean Space Agency (ESA), F asca i, I aly.
21 Al ed Wegene Ins i u e Helmhol z Cen e o Pola and Ma ine Resea ch, 27570 B eme ha en,
Ge many
22 Ins i u e o En i onmen al Physics, Uni e si y o B emen, O o-Hahn-Allee 1, 28359 B emen,
Ge many
23 GFZ Ge man Resea ch Cen e o Geosciences, Po sdam, Ge many21 CSIRO, B isbane, Aus alia
24 Ge man Ae ospace Cen e (DLR), Remo e Sensing Technology Ins i u e, Weßling, Ge many
25 Flande s Ma ine Ins i u e (VLIZ), Oos ende, Belgium
26 Ea h Obse a ion Cen e (EOC), Ge man Ae ospace Cen e (DLR), Weßling, Ge many
27 Ge man Ae ospace Cen e (DLR), Bonn, Ge many
ACIX-III Aqua epo
3
TABLE OF CONTENTS
1 ABSTRACT .............................................................................................................................. 10
2 KEY POINTS ............................................................................................................................ 11
3 INTRODUCTION ...................................................................................................................... 12
3.1 REPORT AIMS ................................................................................................................. 15
4 MATERIALS AND METHODS ..................................................................................................... 17
4.1 IN SITU DATA .................................................................................................................. 17
4.1.1 AERONET-OC NETWORK ............................................................................................ 17
4.1.2 COMMUNITY VALIDATION DATABASE ......................................................................... 18
4.2 ATMOSPHERIC CORRECTION MODELS ........................................................................... 18
4.2.1 ACOLITE ..................................................................................................................... 19
4.2.2 hGRS .......................................................................................................................... 20
4.2.3 POLYMER ................................................................................................................... 21
4.2.4 iCOR .......................................................................................................................... 21
4.2.5 MIP ............................................................................................................................ 22
4.2.6 PACO-WASI ................................................................................................................ 22
4.2.7 ACOLITE-T-Ma .......................................................................................................... 23
5 PRISMA DATA ANALYSIS ......................................................................................................... 25
5.1 FEATURES OF THE PRISMA DATASET USED IN THE EXERCISE .......................................... 25
5.2 MATCH-UPS ANALYSIS ................................................................................................... 26
5.2.1 DATASET .................................................................................................................... 27
5.2.2 PERFORMANCE ASSESSMENT .................................................................................... 27
5.2.3 STATISTICS ................................................................................................................. 28
5.2.4 QUALITY WATER INDEX POLYNOMIAL ......................................................................... 29
5.3 RESULTS ........................................................................................................................ 31
5.3.1 OVER THE SITES ......................................................................................................... 31
5.3.2 OVER THE SPECTRAL BANDS ...................................................................................... 37
5.3.3 OPTICAL WATER TYPES ............................................................................................... 50
5.4 INTERPRETING THE QWIP OUTPUT ................................................................................. 55
6 EnMAP DATA ANALYSES .......................................................................................................... 61
6.1 DATASET AND MATCH-UP ANALYSIS ............................................................................... 61
6.2 IN SITU DATA .................................................................................................................. 61
6.3 EnMAP DATA .................................................................................................................. 63
6.4 MATCH-UP ANALYSIS ..................................................................................................... 63
ACIX-III Aqua epo
4
6.5 PERFORMANCE ASSESSMENT ........................................................................................ 65
6.5.1 MULTISPECTRAL PERFORMANCE ............................................................................... 65
6.5.2 HYPERSPECTRAL PERFORMANCE .............................................................................. 72
7 CONCLUSIONS ...................................................................................................................... 79
8 ACKNOWLEDGMENTS ............................................................................................................ 81
9 REFERENCES ......................................................................................................................... 82
10 PRISMA DATA REPOSITORY ................................................................................................ 90
11 PRISMA GALLERY ............................................................................................................... 91
ACIX-III Aqua epo
5
LIST OF FIGURES
Figu e 1 Da ase o he globally dis ibu ed si es and numbe o a ailable ma ch-ups be ween PRISMA and ield
da a. The colou o he do s indica es he numbe o a ailable ma ch-ups anging om a minimum o 1 o a
maximum o 17 o e e y s udy a ea. The Loca ion o AERONET-OC a e indica ed wi h ci cles only, he black do s
indica e CVD si es; Rio de la Pla a and AAOT (Venice) a e bo h AERONET-OC and CVD. The bo om- igh ba plo
shows he equency dis ibu ion o R s on he y-axis, wi h he alues in s ⁻¹ epo ed on he x-axis. This is shown
o bo h mul ispec al (AERONET-OC, ligh blue) and hype spec al (CVD, g een) in si u da a collec ion. The
da ke blue a ea shows he o e lap be ween he wo dis ibu ions.................................................................. 27
Figu e 2 Re e ence OWTs o he di e en AC models pe o mance assessmen [79]. ...................................... 28
Figu e 3 Spec al compa isons be ween PRISMA L2C and in si u da a in he 21 globally dis ibu ed AERONET-OC
si es. The a iabili y ac oss he da ase in he mean spec a o PRISMA L2C is displayed as blue cu es, wi h he
shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e
equi alen ly shown in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ... 32
Figu e 4 Spec al compa isons be ween ACOLITE and in si u da a in he 21 globally dis ibu ed AERONET-OC
si es. The a iabili y ac oss he da ase in he mean spec a o ACOLITE is displayed as blue cu es, wi h he
shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e
equi alen ly shown in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ... 33
Figu e 5 Spec al compa isons be ween hGRS and in si u da a in he 21 globally dis ibu ed AERONET-OC si es.
The a iabili y ac oss he da ase in he mean spec a o hGRS is displayed as blue cu es, wi h he shaded blue
a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly
shown in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ...................... 34
Figu e 6 Spec al compa isons be ween POLYMER and in si u da a in he 21 globally dis ibu ed AERONET-OC
si es. The a iabili y ac oss he da ase in he mean spec a o POLYMER is displayed as blue cu es, wi h he
shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e
equi alen ly shown in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ... 34
Figu e 7 Spec al compa isons be ween iCOR and in si u da a in he 21 globally dis ibu ed AERONET-OC si es.
The a iabili y ac oss he da ase in he mean spec a o PRISMA iCOR is displayed as blue cu es, wi h he
shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e
equi alen ly shown in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ... 35
Figu e 8 Spec al compa isons be ween MIP and in si u da a in he 21 globally dis ibu ed AERONET-OC si es.
The a iabili y ac oss he da ase in he mean spec a o MIP is displayed as blue cu es, wi h he shaded blue
a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly
shown in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ...................... 36
Figu e 9 Spec al compa isons be ween ACOLITE-T-Ma and in si u da a in he 21 globally dis ibu ed AERONET-
OC si es. The a iabili y ac oss he da ase in he mean spec a o ACOLITE-T-Ma is displayed as blue cu es,
wi h he shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u
da a a e equi alen ly shown in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he
images. ....................................................................................................................................................... 36
Figu e 10 O e all pe o mance o AC p ocesso s using he AERONET-OC ma ch-ups wi h all he sa elli e da a
combined. The numbe o ma chups pe p ocesso and pe band is epo ed in he sca e plo s (N) along wi h
he s a is ics. The black lines e e o he 1:1 line. .......................................................................................... 38
Figu e 11 Pe o mance assessmen s as de e mined by he Median Symme ic Accu acy (ϵ) and Bias (β;) o all
he ma ch-ups combined. The dashed line co esponds o a 30% h eshold [18]. ............................................ 39
Figu e 12 O e all pe o mance o AC p ocesso s using he AERONET-OC ma ch-ups wi h all he sa elli e da a
combined (equal size samples). The numbe o ma chups pe p ocesso and pe band is epo ed in he
sca e plo s (N) along wi h he s a is ics. The black lines e e o he 1:1 line. .................................................. 40
Figu e 13 Pe o mance assessmen s as de e mined by he Median Symme ic Accu acy (ϵ) and Bias (β;) o all
he ma ch-ups combined (equal size samples). The dashed line co esponds o a 30% h eshold [18]. ............. 41

ACIX-III Aqua epo
6
Figu e 14 Spec al compa isons be ween PRISMA L2C and hype spec al in si u da a in he 8 si es o he CVD.
The a iabili y ac oss he da ase in he mean spec a o PRISMA L2C is displayed as blue cu es, wi h he
shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e
equi alen ly shown in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ... 42
Figu e 15 Spec al compa isons be ween ACOLITE and hype spec al in si u da a in he 8 si es o he CVD. The
a iabili y ac oss he da ase in he mean spec a o ACOLITE is displayed as blue cu es, wi h he shaded blue
a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly
shown in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ...................... 43
Figu e 16 Spec al compa isons be ween hGRS and hype spec al in si u da a in he 8 si es o he CVD. The
a iabili y ac oss he da ase in he mean spec a o hGRS is displayed as blue cu es, wi h he shaded blue a ea
ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown
in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ................................ 44
Figu e 17 Spec al compa isons be ween POLYMER and hype spec al in si u da a in he 8 si es o he CVD. The
a iabili y ac oss he da ase in he mean spec a o POLYMER is displayed as blue cu es, wi h he shaded blue
a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly
shown in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ...................... 44
Figu e 18 Spec al compa isons be ween iCOR and hype spec al in si u da a in he 8 si es o he CVD. The
a iabili y ac oss he da ase in he mean spec a o iCOR is displayed as blue cu es, wi h he shaded blue a ea
ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown
in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ................................ 45
Figu e 19 Spec al compa isons be ween MIP and hype spec al in si u da a in he 8 si es o he CVD. The
a iabili y ac oss he da ase in he mean spec a o MIP is displayed as blue cu es, wi h he shaded blue a ea
ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown
in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ................................ 46
Figu e 20 Spec al compa isons be ween ACOLITE-T-Ma and hype spec al in si u da a in he 8 si es o he
CVD. The a iabili y ac oss he da ase in he mean spec a o ACOLITE-T-Ma is displayed as blue cu es, wi h
he shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a
a e equi alen ly shown in o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images.
................................................................................................................................................................... 46
Figu e 21 O e all pe o mance o AC p ocesso s using he CVD ma ch-ups wi h all he sa elli e da a combined.
The numbe o ma chups pe p ocesso and pe band is epo ed in he sca e plo s (N) along wi h he s a is ics.
The black lines e e o he 1:1 line. ............................................................................................................... 48
Figu e 22 Pe o mance assessmen s as de e mined by he Median Symme ic Accu acy (ϵ) and Bias (β) o all he
ma ch-ups combined. The dashed line co esponds o a 30% h eshold. ........................................................ 48
Figu e 23 O e all pe o mance o AC p ocesso s using he CVD ma ch-ups wi h all he sa elli e da a combined
(equal size samples). The numbe o ma chups pe p ocesso and pe band is epo ed in he sca e plo s (N)
along wi h he s a is ics. The black lines e e o he 1:1 line. .......................................................................... 49
Figu e 24 Pe o mance assessmen s as de e mined by he Median Symme ic Accu acy (ϵ) and Bias (β) o all he
ma ch-ups combined (equal size samples). The dashed line co esponds o a 30% h eshold [18]. .................. 50
Figu e 25 Pe o mance assessmen s as de e mined by he spec al Median Symme ic Accu acy (ε) and Bias (β)
o he AC me hods. The dashed-do ed line co esponds o he 30% accu acy h eshold sugges ed by GCOS
and adop ed also by ACIX-Aqua [18]. ............................................................................................................ 50
Figu e 26 Mean alues o PRISMA-de i ed R s spec a o each AC me hod, by OWT. Fo in si u da a, AERONET-
OC and CVD ha e been agg ega ed and plo ed as mean alues wi h s anda d de ia ions, om 443 o 667 nm. N
p o ides he numbe o a e aged in si u spec a pe OWT. ............................................................................. 52
Figu e 27 Agg ega ion o pai wise in e -compa isons (hea maps) ob ained om he MdSA (%) epo ed in each
cell. P ocesso s wi h ligh e colou s (whi e/yellow) a e likely o gene a e high quali y o a gi en OWT and band.
................................................................................................................................................................... 54
ACIX-III Aqua epo
7
Figu e 28 An example o QWIP analysis ou pu o a scene a A iake Towe AERONET-OC si e on May 11, 2020,
p ocessed using he hGRS AC app oach. (A) Map o AVW, (B) Map o he QWIP sco e, (C) sca e plo o QWIP
analysis, and (D) equency dis ibu ion o inc emen al QWIP sco es. ............................................................. 57
Figu e 29 An example o QWIP analysis ou pu o a scene a A iake Towe AERONET-OC si e on May 11, 2020,
p ocessed using he hGRS AC app oach. These plo s display he mean no malised spec al shapes mee ing
a ious c i e ia (A) |QWIP| < 0.2, (B) 0.2 > |QWIP| > 0.4, (C) QWIP > 0.4, (D) QWIP < -0.4, and (E) ou o ange wa e
ype. ........................................................................................................................................................... 58
Figu e 30 Resul s ob ained o a Lake T asimeno scene conside ing all i e AC models (A: ACOLITE, B: hGRS, C:
POLYMER, D: iCOR, E: MIP) and T-Ma p oduc (F). On he le i is shown he QWIP sco e map and on he igh
he sca e plo s o he QWIP analysis. ........................................................................................................... 60
Figu e 31 Da ase o he globally dis ibu ed si es and numbe o a ailable ma ch-ups be ween EnMAP and ield
da a. The colou o he do s indica es he numbe o a ailable ma ch-ups anging om a minimum o 1 o a
maximum o 13 o e e y s udy a ea. The Loca ion o AERONET-OC a e indica ed wi h ci cles only, he black do s
indica e CVD si es. AAOT (Venice) is bo h Hype spec al (CVD si es) and AERONET-OC. ................................. 62
Figu e 32 F equency dis ibu ion o e lec ance le els he hype spec al (CVD) and mul ispec al in si u
measu emen s (all wa eleng hs). ................................................................................................................. 65
Figu e 33 Spec al compa isons be ween EnMAP L2A wa e p oduc s (MIP) and in si u da a in he 11 globally
dis ibu ed AERONET-OC si es. The a iabili y ac oss he da ase in he mean spec a o MIP is displayed as blue
cu es, wi h he shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he
in si u da a a e equi alen ly shown in ed. SA ep esen s he Spec al Angle and N ep esen s he numbe o he
images. ....................................................................................................................................................... 66
Figu e 34 Spec al compa isons be ween ACOLITE and in si u da a in he 9 globally dis ibu ed AERONET-OC
si es. The a iabili y ac oss he da ase in he mean spec a o ACOLITE is displayed as blue cu es, wi h he
shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e
equi alen ly shown in ed. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ........ 67
Figu e 35 Spec al compa isons be ween POLYMER and in si u da a in he 10 globally dis ibu ed AERONET-OC
si es. The a iabili y ac oss he da ase in he mean spec a o POLYMER is displayed as blue cu es, wi h he
shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e
equi alen ly shown in ed. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ........ 68
Figu e 36 Spec al compa isons be ween PACO-WASI and in si u da a in he 12 globally dis ibu ed AERONET-
OC si es. The a iabili y ac oss he da ase in he mean spec a o PACO-WASI is displayed as blue cu es, wi h
he shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a
a e equi alen ly shown in ed. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ... 69
Figu e 37 O e all pe o mance o AC p ocesso s using he AERONET-OC ma ch-ups wi h all he sa elli e da a
combined. The numbe o ma chups pe p ocesso and pe band is epo ed in Table 11 along wi h he s a is ics.
The black lines e e o he 1:1 line. ............................................................................................................... 70
Figu e 38 Pe o mance assessmen s as de e mined by he Median Symme ic Accu acy (ϵ) and median
symme ic Bias (β) o all he ma ch-ups combined. The dashed line co esponds o a 30% h eshold [18]. ...... 72
Figu e 39 Spec al compa isons be ween EnMAP L2A s anda d p oduc s (MIP) and hype spec al in si u da a in
he 6 si es o he CVD. The a iabili y ac oss he da ase in he mean spec a o MIP is displayed as blue cu es,
wi h he shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u
da a a e equi alen ly shown in ed. SA ep esen s he Spec al Angle and N ep esen s he numbe o he
images. ....................................................................................................................................................... 74
Figu e 40 Spec al compa isons be ween ACOLITE and hype spec al in si u da a in he 5 si es o he CVD. The
a iabili y ac oss he da ase in he mean spec a o ACOLITE is displayed as blue cu es, wi h he shaded blue
a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly
shown in ed. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ........................... 75
Figu e 41 Spec al compa isons be ween POLYMER and hype spec al in si u da a in he 5 si es o he CVD. The
a iabili y ac oss he da ase in he mean spec a o POLYMER is displayed as blue cu es, wi h he shaded blue
ACIX-III Aqua epo
8
a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly
shown in ed. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ........................... 76
Figu e 42 Spec al compa isons be ween PACO-WASI and hype spec al in si u da a in he 6 si es o he CVD.
The a iabili y ac oss he da ase in he mean spec a o PACO-WASI is displayed as blue cu es, wi h he
shaded blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e
equi alen ly shown in ed. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images. ........ 77
Figu e 43 Spec a o ϵ, β, RMSE and MdAPE o MIP, Polyme , ACOLITE and PACO-WASI o coinciden
hype spec al ma ch-ups du ing he ope a ional mission phase (N=7). The g ey shaded a ea ep esen s he
equi ed unce ain y (RMSE) ou side o s ong a mosphe ic abso p ion egions as de ined by he EnMAP G ound
Segmen and o AOT a 550 nm lowe han 0.4 [22]. ...................................................................................... 78
ACIX-III Aqua epo
9
LIST OF TABLES
Table 1 A sec ion o aceabili y ma ix in he ‘Inland and Coas al’ wa e ma ke a ea as de ined by he CHIME
MRD (c . Table 4-1) showing link be ween use needs and equi emen s, de ined on he basis o high-le el EU
policies and he CHIME mission equi emen s............................................................................................... 13
Table 2 In si u da a used in he s udy, pe cen age o loca ion in inland and coas al wa e s ( he % a e compu ed
wi h espec o he numbe o si es o each o he ins umen s). * highligh s in si u da a only used o ma chups
in he EnMAP da a e alua ion. ...................................................................................................................... 17
Table 3 Table 3 Main cha ac e is ics o he AC models and T-Ma model. No ably ACOLITE, POLYMER and MIP
a e applied o bo h PRISMA and EnMAP; hGRS, iCOR and T-Ma only o PRISMA; PACO-WASI only o EnMAP. .. 18
Table 4 Band se ings in he di e en da ase s: he nominal band used in he ma ch-up analysis and he
co esponding band se ings o AERONET-OC and PRISMA. ........................................................................... 26
Table 5 Concep ual desc ip ion o OWTs [79]. ............................................................................................... 28
Table 6 Band names ..................................................................................................................................... 37
Table 7 AC models pe o mance by wa eleng h and OWT (including also he T-Ma model). ............................ 54
Table 8 A e age SA esul s o each OWT class. N indica es he size o each class. .......................................... 55
Table 9 Lis o s udy si es, loca ion, ype o in si u measu emen s and numbe o images (C: Commissioning, O:
Ope a ional mission phase, LQ: Low Quali y). ............................................................................................... 61
Table 10 Band se ings in he di e en da ase s: he nominal band used in he ma ch-up analysis and he
co esponding band se ings o AERONET-OC and EnMAP. The i s pa o he able e e s o he band se ing
used o he en i e da ase , while he second pa e e s o he addi ional bands conside ed o he CVD da ase
only. ............................................................................................................................................................ 64
Table 11 Summa y o he i e s a is ical me ics o he EnMAP ma chup da a shown in Figu e 37. N ep esen s
he o al numbe o samples o each band. .................................................................................................. 71
Table 12 S a is ics o MIP, Polyme , ACOLITE and PACO-WASI o wo wa eleng h anges using he common
hype spec al da ase (bo h mission phases). ............................................................................................... 78
ACIX-III Aqua epo
16
mission is much younge han PRISMA, less ma chups wi h in si u da a a e a ailable and only ou
di e en AC me hods a e conside ed in his epo . The e alua ion o he da a by in si u da a akes
ad an age mos ly o he same si es as used o he PRISMA e alua ion. The da a, me hods, esul s
and discussion o he EnMAP AC e alua ion is based on he s udy by [22] which ocuses on 2022-2023
ma chups only and was complemen ed by he ecen ly de eloped a mosphe ic co ec ion me hod
PACO-WASI.
In addi ion o he p esen ed Sec ion 2, he epo is composed by Sec ion 3, which p esen s he in si u
and sa elli e da a and he AC models applied o bo h PRISMA and EnMAP. In Sec ion 4 and Sec ion 5
he epo p esen s he esul s o PRISMA and EnMAP sepa a ely, o he sake o cla i y. Sec ion 6 is
based on he conclusions and combining he esul s o PRISMA and EnMAP.

ACIX-III Aqua epo
17
4 MATERIALS AND METHODS
4.1 IN SITU DATA
In o de o alida e sa elli e mission da a, i is essen ial o ha e a la ge amoun o in si u da a,
collec ed a he same ime as he sa elli e o e passes [18]. This sec ion in oduces he in si u da a
used o ca y ou his wo k: global ne wo ks, ixed adiome e da a and ieldwo k da a, as p esen ed in
Table 2. The da a a e p esen ed simila ly o hose p o ided in ACIX-Aqua, o pu poses o con inui y
ac oss he wo exe cises.
Table 2 In si u da a used in he s udy, pe cen age o loca ion in inland and coas al wa e s ( he % a e compu ed wi h espec
o he numbe o si es o each o he ins umen s). * highligh s in si u da a only used o ma chups in he EnMAP da a
e alua ion.
Ne wo k
In si u
ins umen
Example o
ins umen use
Cha ac e is ics
Loca ion and
ep esen a i eness o he
sys em
AERONET-OC
CIMEL CE-
318T
[23]
Fixed adiome e ,
mul ispec al da a
85% coas al wa e s 15%
inland wa e s
COMMUNITY
VALIDATION DATABASE
(CVD)
WISPS a ion
[11]
Fixed adiome e ,
hype spec al da a
100% inland wa e s
PANTHYR
[24]
Fixed adiome e ,
hype spec al da a
100% coas al wa e s
THETIS
[25]
Fixed adiome e ,
hype spec al da a
100% inland wa e s
HYPSTAR
[26]
Fixed adiome e ,
hype spec al da a
67% coas al wa e s
34% inland wa e s
WISP-3
[4]
Ad-hoc campaigns,
hype spec al da a
75% inland wa e s
25% coas al wa e
Hype -P o*
[27]
Ad-hoc campaigns,
hype spec al da a
100% oligo ophic wa e
Sa lan ic
Hype OCR*
[28]
Fixed adiome e ,
hype spec al da a
100% coas al wa e
OOSS*
[22,29]
Ad-hoc campaigns,
hype spec al da a
100% inland wa e s
4.1.1 AERONET-OC NETWORK
Se e al p o ocols can be ollowed o alida ion ac i i ies, he mos common o which is he global
mul ispec al ne wo k AERONET-OC [23,30]. The ne wo k co e s o e 40 aqua ic si es wo ldwide,
each wi h di e en op ical cha ac e is ics. The main ad an ages o he ne wo k a e au onomous
ope a ion and da a a ailabili y wi hin hou s o collec ion, high quali y and consis ency o da a due o
s anda dised collec ion me hodology, annual ins umen calib a ions and da a e-p ocessing
echniques, and open access p oduc s h ough a speci ic da a policy. AERONET-OC p o ides
measu emen s o no malised wa e adiance (LWN) measu ed by modi ied CIMEL (Pa is, F ance) CE-
318T sola pho ome e s ins alled on ixed o sho e pla o ms, mul ispec al da a in he spec al ange
400-1020 nm (~10 nm bandwid h). P e iously, AERONET-OC suppo ed he adiome ic
cha ac e isa ion o Landsa -8 OLI and Sen inel-2 MSI o aqua ic applica ions, wi h he p e ious ACIX-
ACIX-III Aqua epo
18
Aqua exe cise [18]; in his epo i is used o he assessmen o PRISMA and EnMAP hype spec al
da a.
4.1.2 COMMUNITY VALIDATION DATABASE
The global AERONET-OC ne wo k aluably p o ides a global co e age o wa e si es (mos ly ma ine,
wi h lakes a e s ill unde ep esen ed) in a mul i-spec al se ing, while he a ailabili y o equi alen
da a in a hype spec al band se ing emains a he limi ed. As o ACIX-Aqua, in his wo k, he da ase
collec ed by he CVD was hence exploi ed along wi h he AERONET/OC. The CVD is he esul o a
collabo a ion o di e en scien i ic eams ha p o ided in si u hype spec al da a o alida e he
di e en AC models. In pa icula , he ollowing ixed posi ion au onomous adiome e s deployed o
collec ei he abo e o unde wa e measu emen s we e used in he s udy: WISPS a ion [11], Hyps a
[26,31], Pan hy [24,32], Hype OCR [28] and The is [25]. Fu he mo e, CVD includes also da a
collec ed by he ollowing hand-held spec o adiome e s used du ing ieldwo ks: Hype -P o [27],
OOSS [29] and WISP-3 [4,33]. All da a we e supplied by p o ide s in physical uni s o abo e wa e R s
(s -1).
4.2 ATMOSPHERIC CORRECTION MODELS
This sec ion p esen s he cha ac e is ics o he di e en AC models. A summa y o he main ea u es
is gi en in Table 3.
Table 3 Table 3 Main cha ac e is ics o he AC models and T-Ma model. No ably ACOLITE, POLYMER and MIP a e applied o
bo h PRISMA and EnMAP; hGRS, iCOR and T-Ma only o PRISMA; PACO-WASI only o EnMAP.
AC model
ACOLITE
hGRS
POLYMER
iCOR
MIP
PACO-WASI
T-Ma (adjacency-
e ec co ec ion
only)
Gaseous
O2, O3 (OMI)
O2, O3, O4,
NO2, CO,
CO2, CH4
( om CAMS)
O3 and
NO2
O2, O3,
NO2, CO,
CO2, CH4
(O3
clima ology)
O2, O3,
CO, CO2,
NO, NO2,
CH4, SO2
H2O, O3, CO2, CO,
CH4, N2O, O2, NH3,
NO, NO2, So2, HNO3
MERRA2
Wa e apo
NCEP
ECMWF/CAMS
No
ECMWF
(OLI) [34]
No
Yes [34]
MERRA2
Sun-glin
Fi o esiduals a
ρ c(1609) & ρ c(2200)
Op imal
es ima ion
om SWIR
bands
T ea ed as
bulk signal
Sub ac ion
o minimum
Implici
[35]
No
Sky-glin
[36]
h ough
adia i e
ans e LUT
No
[37]
[35]
No
No
Adjacency
e ec s
No
No
No
SIMEC [38]
[39]
Yes [40]
[41]
Ae osol
Da k a ge app oach
(a ea-based)
Op imal
es ima ion
wi h OPAC
models
Polynomial
i ing (pe -
pixel)
Da k a ge
and AOT
mul i-
pa ame e
in e sion
Pixel-wise
coupled
e ie al
Da k Dense
Vege a ion and da k
a ge app oach (a ea
/ scene based)
MERRA2
ACIX-III Aqua epo
19
(a ea-
based)
Rayleigh LUT
6SV [42]
Coupled
Rayleigh-
ae osol di use
adiance LUT
SOS [43]
MODTRAN
5.0 [44]
Con ained
in ully
coupled
model
MODTRAN 5.4 [44]
No applicable
Geome y
Scene cen e o OLI
and 5-km g ids o
MSI
Pe -pixel
Pe -pixel
Pe -pixel
Pe -pixel
Pe -pixel
Scene cen e
Ae osol
model
Con inen al/ma i ime
ae osols [42] [42]
OPAC
No
MODTRAN
u al
models [44]
modi ied
She le
Fenn
MODTRAN u al
models [44]
Linea ly mixed
con inen al/ma i ime
ae osols [42]
Cloud
masking
ρ (1609) > 0.0215
Th esholds on
se e al
spec al
no malized
indexes and
bands a ound
2200 nm
ρ (865) >
0.2
Cloud mask
laye s a e
p o ided
[45]
Decision
T ee
based on
spec al
and
spec al
indice
h esholds
Th esholds on
se e al spec al
no malized indexes
No applicable
Ou pu g id
cell size (m)
10
30
10/20/60
60
30
Same as L1 da a
Same as inpu
Assump ions
on bio-op ical
condi ions
No
No
Yes [46]
No
No
Yes [47]
No
Ve sion
20231023.0
-
-
3.2
2.3.798
EnMAP 01.05.02
(PACO), 7 (WASI)
2.2.2
Open sou ce
access
Yes (ACOLITE)
No (Unde
cons uc ion)
Yes
(POLYMER
)
No
No
No (PACO)
Yes (WASI)
Yes (T-Ma )
O ganiza ions
RBINS
Magellium
HYGEOS
VITO
EOMAP
DLR
Uni e si y o O awa
Re e ences
[48,49]
[50]
[51,52]
[45]
[39]
[47,53,54]
[41]
4.2.1 ACOLITE
ACOLITE, de eloped a Royal Belgian Ins i u e o Na u al Sciences, is a so wa e de eloped o he AC
o sa elli e images o aqua ic applica ions. I is appliable o images acqui ed by nume ous sa elli e
senso s, among which Landsa 8/9 [55], Sen inel-2 MSI [48,56,57] and Sen inel-3 OLCI [58] and o
da a acqui ed by se e al hype spec al missions, including PRISMA [24] and EnMAP [22]. ACOLITE is
a ailable on Gi Hub
2
and in his s udy he e sion 20231023.0 has been used, wi hou and wi h he
op ion o glin co ec ion. ACOLITE is based on he Da k Spec um [48,57] in which mul iple da k
a ge s in he scene o subscene a e chosen o cons uc a da k spec um. This is hen used o
es ima e he AOT a 550 and a mosphe ic pa h e lec ance acco ding o he bes - i ing ae osol
model. The AOT and ae osol ype we e imposed o be ixed o e he 30 × 30 km PRISMA acquisi ion.
ACOLITE used he Con inen al and Ma i ime ae osol models om 6SV. Fo ACOLITE DSF p ocessing
bo h a PRISMA L1 image and he ma ching L2C da a a e equi ed as inpu s. A un ime, band speci ic
2
h ps://gi hub.com/acoli e/acoli e
ACIX-III Aqua epo
20
Gaussian ela i e spec al esponse unc ions a e gene a ed o bo h VNIR and SWIR de ec o s, using
he cen al wa eleng h and FWHM in o ma ion p o ided in he HDF5 me ada a. Geome y in o ma ion
is cu en ly no included in he L1 da a, and hence he pe -pixel sun and iew geome y (i.e. sun and
iew zeni h and ela i e azimu h angles) is ex ac ed om he ma ching L2C ile. R s is ou pu by
di iding he co ec ed wa e e lec ance by π. He ea e we will e e o R s ob ained om he ACOLITE
AC o PRISMA L1 as ACOLITE DSF. In addi ion o he DSF p ocedu e, ACOLITE has an op ional image-
based sun glin co ec ion [48]. This co ec ion is pe o med by es ima ing he in e ace e lec ance
signal in he SWIR om he a e age ρs be ween 1500 and 2400 nm - i.e. assuming ze o wa e lea ing
adiance in his spec al ange. To ex end his a e age SWIR obse a ion owa ds he VNIR, he
in e ace e lec ance is modelled wi h OSOAA [59] o he cu en scene a e age sun-senso
geome y, he es ima ed ae osol model and AOT o a high (20 m s-1) wind speed. Remo e sensing
e lec ance is ou pu by di iding he ai –wa e in e ace e lec ance co ec ed wa e e lec ance by π.
4.2.2 hGRS
The p ocesso hGRS is a hype spec al adap a ion o he p e ious Glin Remo al o Sen inel-2-like
(GRS) algo i hm which was de eloped o handle p ope co ec ion o he sunglin signal [50] and was
e alua ed h ough he p e ious in e compa ison exe cise ACIX-Aqua [18]. The new hype spec al
algo i hm hGRS p o ides a wo-s ep algo i hm wi h (i) a co ec ion o he gaseous abso p ion
ollowed by (ii) a coupled co ec ion o he a mosphe ic e ec (wi h ae osol es ima ion) and he
e lec ion o he sun on he wa e su ace (sunglin ). No e ha his co ec ion o sunglin is o u mos
impo ance o high-spa ial esolu ion and nadi - iewing senso s such as PRISMA o which he
sunglin signal migh be se e ely p onounced. The p oposed implemen a ion o he gaseous
abso p ion is based on he LibRadT an package. The hype spec al a mosphe ic ansmi ance was
p ecompu ed a 0.1-nm spec al esolu ion o a ious concen a ions o each abso bing gas (O3, O2,
wa e apo , CH4, CO2…). Based on hose p e-compu ed ansmi ances he inal gaseous
ansmi ance is compu ed om he gas concen a ion alues aken om he CAMS da abase and
hen con olu ed wi h he spec al esponse unc ions o he PRISMA senso . As o he ae osol
componen o he signal, he adia i e modeling is based on he models om he Op ical P ope ies o
Ae osol and Clouds (OPAC), e sion 4. Those ae osol models a e buil upon he OPAC da abase [60]
encompassing spec al complex e ac i e index, size dis ibu ion, hyg oscopici y changes and non-
sphe ical pa icles. Ae osol models a e gene a ed based on a mix u e o pu e single componen s. One
ad an age o hose models is o conside he spec al a ia ion o he complex e ac i e index o he
indi idual ae osols. he sca e ing ma ix based on Mie heo y o homogeneous sphe e, T-ma ix o
ellipsoidal/sphe oidal pa icles; hose compu a ions we e ex ended o la ge pa icle sizes wi h he
“Imp o ed Geome ic Op ics Me hod” [61]. The compu a ions we e pe o med h ough he MOPSMAP
package (Modeled op ical p ope ies o ensembles o ae osol pa icles) based on p e-compu ed look-
up ables [62]. These ae osol op ical p ope ies LUT we e used o gene a e he in insic e lec ance o
he a mosphe e and he o al Rayleigh-ae osol ansmi ances h ough he ull ec o adia i e
ans e code OSOAA [59]. The ae osol con en , wa e apo and sunglin componen a e inally
ACIX-III Aqua epo
21
e ie ed om he op imal es ima ion using he Le enbe g-Ma qua d app oach p o iding he inal
in o ma ion o AC and associa ed unce ain y pixel-wise.
4.2.3 POLYMER
The POLYMER p ocesso , de eloped by HYGEOS, is an AC algo i hm pu posely de eloped o
e ie ing he ocean colou when he obse a ion is con amina ed by he sun glin [51,52]. I has been
de eloped o MERIS on En isa and is cu en ly applicable o images acqui ed by o he senso s:
MODIS Aqua, SeaWiFS, VIIRS, Sen inel-3 OLCI, Sen inel-2 MSI, PRISMA, HICO and EnMAP. POLYMER
is a ailable on Gi Hub
3
and in his s udy he e sion 4.17be a2 has been used. The POLYMER
algo i hm uses he image bands in he whole spec al ange om he blue o he nea in a ed, o
decouple he a mosphe ic and su ace componen s o he signal om he wa e e lec ance. The
bands nea s ong a mosphe ic abso p ion ea u es a e il e ed ou . The algo i hm elies on wo
models, an analy ical model o he a mosphe e which uses a second-o de polynomial o mula ion
wi h espec o he wa eleng h, and a wa e e lec ance model based on wo pa ame e s, one o he
chlo ophyll concen a ion and one o he backsca e ing coe icien . The la e is based on he bio-
op ical model o [46]. Wa e -lea ing e lec ance is u he no malised o he nominal wa eleng h and
o he wa e -lea ing e lec ance bidi ec ional e ec s (BRDF), such ha he adiome ic ou pu o
Polyme is he ully no malised wa e -lea ing e lec ance spec um co ec ed o bidi ec ional e ec s.
De ailed in o ma ion on he p e-co ec ion (e.g., es ima ion o he gaseous ansmi ance, Rayleigh
sca e ing, and p e-glin co ec ion), spec al ma ching, and he a mosphe ic model was p esen ed in
[51,52]; in o ma ion on he wa e -lea ing e lec ance model can be ound in [46] and i s modi ica ions,
wa e -lea ing e lec ance no malisa ion, and BRDF co ec ion we e desc ibed in [52]. R s is ou pu by
di iding he co ec ed no malised wa e -lea ing e lec ance by π.
4.2.4 iCOR
iCOR (image COR ec ion o a mosphe ic e ec s) is a MODTRAN-5-based AC ool designed o
p ocess sa elli e da a collec ed o e coas al, inland, o ansi ional wa e s and land [45]. Basic iCOR
e sions o Landsa -8, Sen inel-2, and OLCI-Sen inel-3 a e publicly a ailable as ee plugins o he
SNAP oolbox. Fo ACIX-III-Aqua, an ad anced e sion o iCOR has been de eloped o enable he
co ec ion o hype spec al PRISMA da a. S a ing om PRISMA TOA da a, iCOR iden i ies land, wa e ,
and cloud pixels h ough a senso -speci ic h esholding app oach. The AOT alues a e hen e ie ed
om he image using iCOR’s image-based AOT e ie al me hod. When cloud co e is oo ex ensi e,
o he scene lacks su icien spec al a iabili y o accu a e AOT e ie al, an au oma ic allback
mechanism uses Nea -Real Time (NRT) AOT alues om he Cope nicus A mosphe ic Moni o ing
Se ice (CAMS). Wa e apo e ie al is pe o med di ec ly om PRISMA image y using he
A mosphe ic P e-co ec ed Di e en ial Abso p ion (APDA) app oach [34]. Addi ionally, SIMEC
adjacency co ec ion [38] is applied o minimize adjacency e ec s, especially nea land-wa e
3
h ps://gi hub.com/hygeos/polyme

ACIX-III Aqua epo
22
bounda ies. Following AOT e ie al and adjacency co ec ion, spec al bands a e a mosphe ically
co ec ed using p e-calcula ed MODTRAN-5 look-up ables (LUTs). These LUTs inco po a e (1) sola
and iewing geome y, (2) e ain al i ude om he GLOBE DEM, (3) he e ie ed AOT alues, (4) wa e
apo concen a ions, and (5) mon hly ozone clima ology da a om he To al Ozone Mapping
Spec ome e (TOMS). Finally, o wa e pixels, a F esnel e lec ance co ec ion is applied, while land
pixels a e modelled as Lambe ian su aces. To add ess unce ain ies in F esnel co ec ion ela ed o
glin and haze, he minimum wa e -lea ing e lec ance in selec ed NIR-SWIR bands is calcula ed o
each wa e pixel. This alue, i g ea e han ze o, is sub ac ed om he e ie ed wa e -lea ing
e lec ance alues ac oss all spec al bands.
4.2.5 MIP
MIP is an ope a ional p ocessing chain ha is in p inciple senso -independen . I is op imized o mos
accu a ely p edic in wa e p ope ies and uses a coupled a mosphe e-wa e model s o ed in a
lookup- able. An in e sion algo i hm sea ches o he mos likely se o model pa ame e s ha explain
he obse ed senso adiances. Those pa ame e s include su ace al i ude, geome y, ae osol
p ope ies, wa e ing edien concen a ions and wa e su ace oughness. Excep o he adjacency
co ec ion, he pa ame e s a e e ie ed pixel-wise. Re lec ance p oduc s a e no di ec ly used in he
e ie al p ocess, bu a e a side p oduc ha can be econs uc ed using he e ie ed a mosphe ic
and wa e su ace p ope ies and p opaga ing hose h ough he adia i e ans e simula ion. To
p oduce wa e e lec ance da a, he chain c ea es a pixelwise classi ica ion in o land/wa e /cloud. I
emo es he adjacency e ec ha land pixels ha e on wa e and hen e ie es all ele an pa ame e s
o each wa e pixel. F om hose pa ame e s he e lec ance p oduc s can be c ea ed using he
me hod men ioned abo e.
4.2.6 PACO-WASI
Fo he EnMAP sa elli e da a only, a combina ion o wo so wa e ools, PACO and WASI, was applied.
The EnMAP Le el-2A da a can be downloaded as wa e p oduc , p ocessed using MIP, as land
p oduc , p ocessed using PACO, o as a combined p oduc . PACO-WASI e e s o he pos p ocessing
o he land p oduc wi h he so wa e WASI o con e su ace e lec ance o emo e sensing
e lec ance. PACO (Py hon-Based A mosphe ic Co ec ion, [53]) is he Py hon e sion o he so wa e
package ATCOR [40,63] which is op imized o he a mosphe ic co ec ion o mul i- and hype spec al
sa elli e and ae ial images o e land su aces. I makes use o p ecompu ed lookup ables, gene a ed
wi h Mod an 5.4, and accoun s o g ound e ec s such as e ain ele a ion and adjacency e ec s.
The esul ing Le el-2A p oduc is Hemisphe ical Di ec ional Re lec ance (HDR) a bo om o
a mosphe e, commonly called su ace e lec ance. WASI (Wa e Colo Simula o ) is a so wa e
designed o he simula ion and da a analysis o spec al measu emen s in and abo e wa e . I can be
downloaded om [54]. O iginally de eloped o ield spec ome e s [64], WASI has since been
adap ed o p ocess mul i- and hype spec al image da a o e wa e om any senso [47]. The inpu
ACIX-III Aqua epo
23
da a mus be a mosphe ically co ec ed and can be in uni s o e lec ance o adiance. The simula ion
o emo e sensing e lec ance is based on he analy ical models o [65] o op ically deep and shallow
wa e s, which had been de eloped using adia i e ans e simula ions wi h Hyd oligh [66]. The
e lec ions a he wa e su ace a e simula ed using he model o [67] which decomposes he sky
adiance e lec ed in senso di ec ion in o h ee spec ally di e en componen s o igina ing om he
sun (sun glin ), sca e ing a molecules (Rayleigh glin ) and sca e ing a pa icles (ae osol glin ). The
wa eleng h-dependen glin componen s a e pa ame e ized in e ms o ae osol op ical hickness,
Angs öm exponen o ae osol sca e ing, wa e apo and ozone scale heigh using he i adiance
model o [68]. All model pa ame e s a e expe imen ally accessible and all iles speci ying he op ical
p ope ies o he componen s in he wa e and he a mosphe e can be exchanged easily, hence WASI
allows he simula ion o measu emen s o a wide ange o en i onmen al condi ions. Da a analysis
applies in e se modelling (IM) o spec al measu emen s, and in case o image da a, i addi ionally
employs a neu al ne wo k (NN) ha is ained image-speci ically using he IM esul s om a
ep esen a i e subse o se e al hund ed pixels. The lexibili y o WASI allows a egional ine- uning o
he simula ions and o da a p ocessing, which can p o ide mo e accu a e esul s compa ed o global
algo i hms, ye he op imiza ion o IM o he ac ual condi ions equi es expe knowledge. Since he
same physical p ocesses a e esponsible o pa h adiance and sky glin , IM canno dis inguish
be ween pa h adiance e o s om a mosphe ic co ec ion and sky glin , which has he ad an age o
he de i a ion o emo e sensing e lec ance ha bo h e ec s a e co ec ed oge he . The glin
co ec ion algo i hm is desc ibed in [35]. In he combined PACO-WASI wo k low, PACO is i s applied
o co ec he a mosphe e, ollowed by WASI, which p ocesses he su ace e lec ance ou pu o
PACO o co ec o sun glin , sky glin and pa h adiance e o s. I is impo an o no e ha PACO-
WASI is no ye an ope a ional wo k low as he s ep o in e se modelling equi es manual ine- uning
o WASI o each indi idual scene by an expe ienced use . This uning p ocess in ol es i e a i e
adjus men s o he IM se ings – including he selec ion o i pa ame e s, ini ial alues o i
pa ame e s, and spec al ange used o da a p ocessing – un il wo c i e ia a e ul illed: (1) he
modelled spec a i he su ace e lec ance spec a as close as possible, (2) he noise caused by
spec al ambigui ies is minimized. Spec al ambigui ies, which a e a common p oblem in op ically
complex wa e s [69], a e iden i ied using co ela ion plo s o pixels ha ha e been p ocessed using
bo h IM and NN app oaches; hese diagnos ic plo s a e gene a ed au oma ically by WASI. An
au oma iza ion o hese ine- uning s eps is cu en ly unde de elopmen .
4.2.7 ACOLITE-T-Ma
As men ioned ea lie , in addi ion o he i e AC models, he co ec ion o he adjacency e ec using T-
Ma combined wi h ACOLITE was included. In pa icula , in his exe cise, a p e-p ocessing s ep using
T-Ma [41,70] was applied o co ec o he adjacency e ec a he TOA le el, ollowed by ACOLITE
AC ( e sion: 20231023) wi h he same se ing as he ACOLITE-only implemen a ion. T-Ma adjus s he
TOA e lec ance o each spec al band o sa is y he homogeneous-su ace assump ion, i.e. pixels
a e co ec ed o he TOA e lec ance hey would exhibi i each pixel we e su ounded by pixels o
ACIX-III Aqua epo
24
iden ical e lec ance. The co ec ion p ocess begins by calcula ing he a mosphe ic poin sp ead
unc ion (PSF) speci ic o an image’s sola -senso geome y, ollowed by con ol ing he image wi h
he PSF as he ke nel. The di e ence be ween he o iginal and con ol ed images is scaled, based on
adia i e ans e simula ions, by a ac o ha app oxima es he a io o upwa d di use o upwa d
di ec ansmi ance. The scaled di e ence is hen sub ac ed om he o iginal image, ollowed by a
ma ix mul iplica ion ha accoun s o a ia ions in su ace-le el i adiance due o su ace
he e ogenei y using a scene-speci ic look-up able, o inalize he adjacency e ec co ec ion [41].
When compu ing he PSFs and scaling ac o s in T-Ma , he a mosphe ic composi ion, including
ozone, wa e apo , and ae osol p ope ies om he GMAO MERRA2 eanalysis da a [71], is used o
cha ac e ize he op ical p ope ies o he a mosphe e a a ious al i udes ia he 6S adia i e ans e
model [72].
ACIX-III Aqua epo
25
5 PRISMA DATA ANALYSIS
PRISMA (PRecu so e Ipe Spe ale della Missione Applica i a) is an ad anced Ea h obse a ion
sa elli e unded and ope a ed by he I alian Space Agency (ASI). Launched on 22nd Ma ch 2019
aboa d he Vega ocke , PRISMA o bi s in a low Ea h Sun-synch onous o bi a an al i ude o
app oxima ely 614.8 km [24,73–77]. I is designed o e isi he same a ea e e y 29 days, hough his
pe iod can be educed o less han one week using o -nadi poin ing capabili ies. This lexibili y
allows PRISMA o cap u e speci ic a eas mo e equen ly, making i sui able o di e se en i onmen al
moni o ing applica ions. The PRISMA payload is a combina ion o wo key ins umen s: a
hype spec al senso wi h a G ound Sampling Dis ance (GSD) o 30 me e s, and a panch oma ic (PAN)
came a wi h a GSD o 5 me e s. The hype spec al senso cap u es da a in 238 con iguous spec al
bands anging om 400 o 2500 nm, enabling p ecise de ec ion o ea u es ac oss he isible, nea -
in a ed (VNIR), and sho -wa e in a ed (SWIR) egions. This imaging spec ome e c ea es a
hype da a cube, allowing o de ailed analysis o spec al and spa ial ea u es in a wide ange o
en i onmen al con ex s. Meanwhile, he PAN came a p o ides high- esolu ion con ex o he
hype spec al da a, aiding in applica ions ha equi e ine -scale mapping. PRISMA's sys em can
acqui e, downlinking, and a chi ing up o 200,000 km² o hype spec al and panch oma ic images
daily, co e ing a eas be ween 70°S and 70°N in la i ude and om 180°W o 180°E in longi ude. In a
single pass, i can cap u e images up o 1000 km apa by adjus ing i s iewing angles. This lexibili y
enhances i s empo al esolu ion, which is c ucial o acking dynamic en i onmen al changes o e
ime. PRISMA p oduc s ange om Le el 0 (L0) o Le el 2 (L2): L0 ep esen s o ma ed da a p oduc s,
Le el 1 (L1) ep esen s adiance da a adiome ically co ec ed and calib a ed in physical uni s, and
L2 ep esen s he esul o he con e sion o Top-o -A mosphe e (TOA) spec al adiance
measu emen s in o Bo om-o -A mosphe e (BOA) emo e sensing e lec ance measu emen s. L2
da a a e di ided in o:
● L2B (geoloca ed g ound spec al adiance p oduc );
● L2C (geoloca ed a -su ace e lec ance p oduc );
● L2D ( he geocoded e sion o L2C p oduc s).
The AC p ocess is desc ibed in sec ion 2.2.1 PRISMA mission is ecognized o i s po en ial in a ious
domains, including ag icul u e, o es y, mining, and aqua ic emo e sensing. I s da a is pa icula ly
aluable o algo i hm de elopmen and inno a i e en i onmen al moni o ing echniques.
5.1 FEATURES OF THE PRISMA DATASET USED IN THE EXERCISE
The da a used in he exe cise a e he PRISMA L1 da a ( e sion 4.1-0) and PRISMA L2C da a ( e sion
02.05). The quali y o L1 da a was assessed in [73] by compa ison wi h TOA adiances simula ed wi h
a adia i e ans e code, om in si u measu emen s, ob ained om a se ies o ixed-posi ion
au onomous adiome e s. The esul s showed ha PRISMA p o ides TOA adiances wi h he same
ampli ude and shape as hose simula ed in si u wi h sligh ly la ge unce ain ies a sho e
ACIX-III Aqua epo
32
● In he Gus a Dalen Towe and I be Ligh house si es, he spec al signa u es a e simila : peak
R s (< 0.005 s -1) a ound 550 nm, which hen ends o dec ease o ze o.
● In he Gala a pla o m si e, he spec um is associa ed wi h op ically complex wa e s, wi h R s
maxima ypically obse ed in he 490-560 nm ange, p ope ies common o he Sec ion 7 si e).
● In he Lucinda si e (in he opical coas al wa e s o he G ea Ba ie Ree lagoon nea he
He be Ri e es ua y), he spec um shows a peak jus be o e 550 nm and a s anda d
de ia ion o ± 0.005 s -1.
PRISMA L2C
In all cases, an o e es ima ion o he R s PRISMA L2C spec a compa ed o he in si u da a can be
obse ed; be e ag eemen be ween sa elli e da a and in si u da a is seen in mo e eu ophic wa e s,
whe e he wa e -lea ing signal is highe . Thus, he lowe SA alue (8.92°) was iden i ied in he case o
he Bahia Blanca si e. In mo e clea wa e s whe e he e is a low signal, PRISMA's pe o mance is o en
sub-op imal, as has also been epo ed in o he s udies [24]: mean SA ~ 18°. Fo he Lake E ie si e, he
R s PRISMA spec a p esen s he highes s anda d de ia ion (±0.015 s -1). In Lake Okeechobee, whe e
p onounced abso p ion and e lec ion peaks a e obse ed, he e is a low le el o ag eemen
(SA=32.86°). The un ealis ic spec al a iabili y a some si es (e.g., Casablanca, Gala a pla o m,
Socheongcho, USC SeaP ism) should also be no ed. This non-physical a iabili y indica es he
p esence o image a e ac s (e.g., sca e ed ligh ), pa icula ly in he blue pa o he spec um and a
si es wi h low-magni ude spec a (i.e. blue o o ganic- ich wa e s). The SA anges om 8.92° (Bahia
Blanca si e) o 47.49° (Palg unden si e).
Figu e 3 Spec al compa isons be ween PRISMA L2C and in si u da a in he 21 globally dis ibu ed AERONET-OC si es. The
a iabili y ac oss he da ase in he mean spec a o PRISMA L2C is displayed as blue cu es, wi h he shaded blue a ea
ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown in o ange. SA
ep esen s he Spec al Angle and N ep esen s he numbe o he images.

ACIX-III Aqua epo
33
ACOLITE
A compa ison be ween ACOLITE and in si u da a shows a good i be ween he da a. Pa icula ly in he
case o he Bahia Blanca (SA=2.93°), Casablanca pla o m (SA=8.21°), Lake E ie (SA=10.68°) and
Zeeb ugge (SA=9.37°) si es. The highes s anda d de ia ion alues we e eco ded in he cases o
LISCO (±0.002 s -1), San Ma co pla o m (±0.007 s -1) and Sou h G eenbay (±0.01 s -1). The SA anges
om 2.93° (Bahia Blanca si e) o 36.98° (I be Ligh house si e).
Figu e 4 Spec al compa isons be ween ACOLITE and in si u da a in he 21 globally dis ibu ed AERONET-OC si es. The
a iabili y ac oss he da ase in he mean spec a o ACOLITE is displayed as blue cu es, wi h he shaded blue a ea
ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown in o ange.
SA ep esen s he Spec al Angle and N ep esen s he numbe o he images.
hGRS
He e, he p oduc s ob ained by applying hGRS model o sa elli e da a; SA was lowe han 20° o he
62% o he s udy a eas. The e is high conco dance in he NIR egion, and some di e ence wi h in si u
da a in he blue egion. The highes ag eemen was eached in Bahia Blanca (SA=5.1°), ollowed by
Zeeb ugge and Casablanca (bo h wi h SA=8.9°). Leas ag eemen was no ed in Kemigawa (SA=46.1°),
while a good pe o mance can be no ed in he case o he Sou h G eenbay si e, whe e a hype -
eu ophic condi ion (supe nu ien ichness) is p esen . In San Ma co pla o m and Lake E ie a high
s anda d de ia ion was egis e ed (±0.015 s -1 and ±0.008 s -1 espec i ely).
ACIX-III Aqua epo
34
Figu e 5 Spec al compa isons be ween hGRS and in si u da a in he 21 globally dis ibu ed AERONET-OC si es. The
a iabili y ac oss he da ase in he mean spec a o hGRS is displayed as blue cu es, wi h he shaded blue a ea
ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown in o ange. SA
ep esen s he Spec al Angle and N ep esen s he numbe o he images.
POLYMER
In he case o POLYMER, he e is a good le el o ag eemen wi h in si u measu emen s, pa icula ly in
he case o he A iake Towe , Casablanca pla o m, Lake E ie, Lucinda and Rio de la Pla a si es, wi h
mean SA = 3.5°. Only in a ew cases is he blue spec um somewha noisy and he e is an
o e es ima ion o he R s spec um compa ed o he in si u da a. In he case o Lake Okeechobee, he
esul o he compa ison is weak. The SA a ies om 2.54° (Casablanca si e) o 73.33° (Lake
Okeechobee si e).
Figu e 6 Spec al compa isons be ween POLYMER and in si u da a in he 21 globally dis ibu ed AERONET-OC si es. The
a iabili y ac oss he da ase in he mean spec a o POLYMER is displayed as blue cu es, wi h he shaded blue a ea
ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown in o ange. SA
ep esen s he Spec al Angle and N ep esen s he numbe o he images.
ACIX-III Aqua epo
35
iCOR
The R s spec a de i ed om he co ec ion wi h iCOR showed a good ag eemen o all he s udy
a eas wi h in si u da a. The highes ag eemen was eached in Zeeb ugge (SA=4.87°), ollowed by
Bahia Blanca (SA=6.88°), and A iake Towe (SA=7.34°), while lowe conco dance was ound in
Palg unden (SA=31.01°). In Lake E ie, LISCO and Sou h G eenbay he e lec ance peak, due o he
eu ophic condi ions o he wa e , can be no iced (0.007 s -1, 0.003 s -1 and 0.012 s -1 espec i ely). A
s anda d de ia ion o abou ±0.010 s -1 was egis e ed in San Ma co Pla o m and Lake Okeechobee.
Figu e 7 Spec al compa isons be ween iCOR and in si u da a in he 21 globally dis ibu ed AERONET-OC si es. The a iabili y
ac oss he da ase in he mean spec a o PRISMA iCOR is displayed as blue cu es, wi h he shaded blue a ea ep esen ing
he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown in o ange. SA ep esen s
he Spec al Angle and N ep esen s he numbe o he images.
MIP
The R s spec a de i ed om he co ec ion wi h MIP showed a good ag eemen o all he s udy a eas
wi h in si u da a (SA lowe han 20° o he 75% o he s udy a eas), pa icula ly in he blue egion,
al hough his a ea, oge he wi h he NIR spec al egion p esen ed some unusual peaks (a ound 700
nm). The spec al ea u e a 760 nm is p obably due o he oxygen abso p ion. The SA anges om
5.02° (Zeeb ugge si e) o 33.01° (Palg unden si e). In San Ma co pla o m high s anda d de ia ion was
egis e ed (±0.009 s -1).
ACIX-III Aqua epo
36
Figu e 8 Spec al compa isons be ween MIP and in si u da a in he 21 globally dis ibu ed AERONET-OC si es. The a iabili y
ac oss he da ase in he mean spec a o MIP is displayed as blue cu es, wi h he shaded blue a ea ep esen ing he
s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown in o ange. SA ep esen s he
Spec al Angle and N ep esen s he numbe o he images.
ACOLITE-T-Ma
Simila esul s ( espec o ACOLITE) ob ained om he ACOLITE-T-Ma model, he e is a sligh
imp o emen a si es close o he coas , i.e. hose ha may be mo e subjec o he adjacency e ec .
In ac , he imp o emen is mos no iceable in he cases o LISCO and Sou h G eenbay (dis ance om
he coas o 1.6 and 0.5 nau ical miles), SA imp o ed by abou 20%. The SA anges om 2.81° (Bahia
Blanca si e) o 32.00° (I be Ligh house si e).
Figu e 9 Spec al compa isons be ween ACOLITE-T-Ma and in si u da a in he 21 globally dis ibu ed AERONET-OC si es.
The a iabili y ac oss he da ase in he mean spec a o ACOLITE-T-Ma is displayed as blue cu es, wi h he shaded blue
a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown in
o ange. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images.
ACIX-III Aqua epo
37
5.3.2 OVER THE SPECTRAL BANDS
A e pe o ming he spec al compa ison, sa elli e and in si u da a om all 21 si es we e agg ega ed
by AC model and esampled o he spec al con igu a ion o AERONET-OC, selec ing he PRISMA
bands closes o he co esponding AERONET-OC bands. Table 6 shows he name assigned o each
co esponding band ha is used om now on.
Table 6 Band names
Name
b_443
b_490
b_560
b_620
b_667
b_709
Band
443 nm
490 nm
560 nm
620 nm
667 nm
709 nm
Figu e 10 shows he sca e plo s in he 6 selec ed bands oge he wi h he sample size alue (N) and
s a is ics (RMSD, MAD, MAPD, Mean bias, Slope). In gene al, mo e dispe sion was no ed in he
UV/blue egion, pa icula ly a b_443 (RMSD anged 0.0042 - 0.0128 s -1) and b_490 (RMSD anged
0.0030 - 0.0101 s -1). Good i ing a b_560 (RMSD anged 0.0023 - 0.0073 s -1) and b_620 (RMSD
anged 0.0018 - 0.0076 s -1). Sligh dispe sion in he case o he ed band a b_667 (RMSD anged
0.0014 - 0.0065 s -1). The dispe sion a b_709 may also be due o he p esence o ew samples (RMSD
anged 0.0019 - 0.0099 s -1). Fo all bands POLYMER showed he lowes RMSD alues, ollowed by MIP
(be e pe o mance a b_443). Sligh ly highe RMSD alues in he case o ACOLITE-T-Ma and hGRS
(lowes alues a b_490). The sample size (N) changes a di e en wa eleng hs due o he di e en
con igu a ions o he in si u ins umen s; o he same wa eleng h, i also changes depending on he
model due o he p esence o nega i e alues ( emo ed om he analysis) and he comple e non-
e u n in ou pu o he p oduc .

ACIX-III Aqua epo
38
Figu e 10 O e all pe o mance o AC p ocesso s using he AERONET-OC ma ch-ups wi h all he sa elli e da a combined. The
numbe o ma chups pe p ocesso and pe band is epo ed in he sca e plo s (N) along wi h he s a is ics. The black lines
e e o he 1:1 line. (Image adap ed om [79]).
Figu e 11 shows he wo s a is ics MdSA and Bias used p e iously in he ACIX-Aqua p ojec pape in
ba plo o ma . In gene al, he noisies band, whe e he highes e o alues we e eco ded, was
b_443, while a lowe e o was ound a b_560 and b_620. Lowe alues o MdSA and Bias, hence
be e ag eemen be ween he da a, we e ob ained in he case o MIP. Fu he mo e, o almos all
bands MIP showed MdSA alues sligh ly highe han he 30% h eshold. Highe unce ain ies shown in
he case o PRISMA.
ACIX-III Aqua epo
39
Figu e 11 Pe o mance assessmen s as de e mined by he Median Symme ic Accu acy (ϵ) and Bias (β;) o all he ma ch-ups
combined. The dashed line co esponds o a 30% h eshold [18]. (Image aken om Figu e 4 [79]).
To pe o m a mo e obus analysis, he s a is ical compa ison was also ca ied ou by conside ing
samples wi h equal size: o do his, he eco ds o each model we e emo ed when e en one model
had nega i e alues o did no p o ide he ou pu . In his case, POLYMER showed again he lowe
mean bias, ollowed by MIP ( om b_443 o b_667) and hGRS a b_709. Sligh ly highe alues in he
case o ACOLITE-T-Ma and iCOR; while he i s one pe o med pa icula ly well a b_620, b_667 and
b_709, iCOR showed be e esul s a b_443 and b_490. O e all, he e o alues epo ed in he ba
plo (Figu e 13) a e sligh ly dec eased and MIP has MdSA e o alues sligh ly unde he 30%
h eshold.
ACIX-III Aqua epo
40
Figu e 12 O e all pe o mance o AC p ocesso s using he AERONET-OC ma ch-ups wi h all he sa elli e da a combined
(equal size samples). The numbe o ma chups pe p ocesso and pe band is epo ed in he sca e plo s (N) along wi h he
s a is ics. The black lines e e o he 1:1 line.
ACIX-III Aqua epo
41
Figu e 13 Pe o mance assessmen s as de e mined by he Median Symme ic Accu acy (ϵ) and Bias (β;) o all he ma ch-ups
combined (equal size samples). The dashed line co esponds o a 30% h eshold [18].
5.3.2.1 CVD
Figu e 14 - Figu e 20 show he spec al compa isons be ween sa elli e and in si u da a o he se en
CVD si es. The selec ed s udy a eas a e ou lakes ( h ee in I aly and one in Swi ze land), Venice
(lagoon and AAOT), and Rio de La Pla a (A gen ina). The s udy a eas ha e signi ican op ical and
mo phome ic di e ences; o all o hem, in si u adiome ic da a we e a ailable o calib a ion and
alida ion pu poses. Fo Lake Ga da he analysis has been pe o med di e en ia ing be ween
op ically deep and shallow wa e s (2 m a e age dep h, close o Si mione peninsula). A b ie
cha ac e isa ion o he si es ollows:
● Lake Ga da (oligo-meso ophic): is a deep subalpine glacial lake, whe e he R s spec um is
a ound 0.007 s -1 wi h a small s anda d de ia ion; ins ead, in he shallow wa e s he R s
spec um is abou double (peak a ound 550 nm) and has a la ge s anda d de ia ion.
● Lake Gene e (oligo-meso ophic): is a deep subalpine glacial lake, he R s spec um is a ound
0.006 s -1 wi h a s anda d de ia ion o abou 0.004 s -1.
● Rio de la Pla a (eu ophic): is a basin es ua y wi h a ypical spec al signa u e o b own wa e ,
he R s peak is abou 0.030 s -1 (a ound 600 nm).
● Lake T asimeno (meso-eu ophic): is a shallow ec onic lake, wi h he R s peak a ound 0.0025
s -1 in he g een egion, abso p ion and e lec ance peaks (a ound 670-700 nm) ela ed o he
p esence o Chl-a a e no iceable om he spec al signa u e.
● Lake Va ese (meso-eu ophic): is a shallow glacial lake, p esen ing a spec al signa u e simila
o Lake T asimeno wi h sligh ly lowe R s peak alues.
● Venice (AAOT) (oligo-meso ophic): is a ansi ional egion be ween open sea and coas al
wa e s, he R s peak is abou 0.007 s -1 (a ound 500 nm), he spec al signa u e ends oughly
o ze o a e he e lec ance peak.
● Venice Lagoon (meso-eu ophic): is a shallow coas al en i onmen , he R s peak is abou
0.020 s -1, wi h high s anda d de ia ion (abou 0.010 s -1).
ACIX-III Aqua epo
48
Figu e 21 O e all pe o mance o AC p ocesso s using he CVD ma ch-ups wi h all he sa elli e da a combined. The numbe
o ma chups pe p ocesso and pe band is epo ed in he sca e plo s (N) along wi h he s a is ics. The black lines e e o
he 1:1 line. (Image adap ed om [79])
Figu e 22 Pe o mance assessmen s as de e mined by he Median Symme ic Accu acy (ϵ) and Bias (β) o all he ma ch-ups
combined. The dashed line co esponds o a 30% h eshold
5
.
The s a is ical compa ison was epea ed wi h he same modali ies conside ing samples wi h equal
size, as did o AERONET-OC. The esul s can be ound in Figu e 23 and Figu e 24. As in he p e ious
analysis, POLYMER showed he lowes MAPD alues in all he bands, excep o b_709 whe e MIP
(MAPD=37%) and hGRS (MAPD=42%) ag eed mo e wi h in si u da a. MIP is pe o ming well also a b_
620 and b_667 (MAPD=38%), while hGRS a b_443 and b_490 (MAPD=59% and 42% espec i ely).
5
h ps://gcos.wmo.in /en/essen ial-clima e- a iables/lakes/ec - equi emen s

ACIX-III Aqua epo
49
Figu e 23 O e all pe o mance o AC p ocesso s using he CVD ma ch-ups wi h all he sa elli e da a combined (equal size
samples). The numbe o ma chups pe p ocesso and pe band is epo ed in he sca e plo s (N) along wi h he s a is ics.
The black lines e e o he 1:1 line.
ACIX-III Aqua epo
50
Figu e 24 Pe o mance assessmen s as de e mined by he Median Symme ic Accu acy (ϵ) and Bias (β) o all he ma ch-ups
combined (equal size samples). The dashed line co esponds o a 30% h eshold [18].
Deploying hype spec al da a, which p o ides a wide ange o bands, a mo e de ailed analysis was
ca ied ou conside ing he en i e spec um (Figu e 25). Conce ning all AC me hods, highe e o s
occu ed in he blue and ed spec al egions, as p e iously demons a ed o bo h PRISMA (Pelleg ino
e al., 2023) and EnMAP [22]. PRISMA L2C has highe ϵ and β alues han he o he AC me hods;
POLYMER p o ides lowe ϵ alues in he blue band, while hGRS p o ides lowe ϵ alues in he ed-NIR
bands.
Figu e 25 Pe o mance assessmen s as de e mined by he spec al Median Symme ic Accu acy (ε) and Bias (β) o he AC
me hods. The dashed-do ed line co esponds o he 30% accu acy h eshold sugges ed by GCOS and adop ed also by
ACIX-Aqua [18]. (Image aken om Figu e 6 [79]).
5.3.3 OPTICAL WATER TYPES
The pai wise in e compa ison s a egy o in si u R s esul ed in eigh OWTs (selec ed among he 10
p oposed by [80]) ha we e hen used o classi y he co esponding PRISMA-de i ed R s spec a om
he AC me hods (Figu e 26). In he i s OWT classes (OWT 2, 3a and 3b), ep esen ing blue wa e s wi h
inc easing loads o suspended and dissol ed ma e s, he numbe o in si u R s spec a o be used in
he pai wise in e compa ison wi h PRISMA anges om 10 o OWT 2 o 57 o OWT 3b. Fo OWT 2, he
spec al shape o PRISMA-de i ed R s alues gene ally con o ms o he in si u measu emen s o all
ACIX-III Aqua epo
51
AC p ocesso s excep PRISMA L2C. In he case o OWT 3a and 3b PRISMA-de i ed R s alues exhibi
sys ema ic o e es ima ion a wa eleng hs sho e han 450 nm o all AC me hods. The mos
ep esen ed class o in si u R s was he OWT 4, wi h a highe p edominance o OWT 4b (82 spec a)
han OWT 4a (22 spec a). This class is ypical o g eenish wa e , wi h high phy oplank on biomass and
wi h low e lec ance a he sho es wa eleng hs due o abso p ion by pa icles and yellow
subs ances. Gene ally, PRISMA-de i ed R s spec a main ain he ea u es o he in si u da a collec ion
wi h some excep ions in he blue bands, mos ly om PRISMA L2C. OWT 5 is also ep esen a i e o
g een eu ophic wa e , wi h subs an ially highe phy oplank on biomass and a local R s maximum
a ound 700 nm; his is also cap u ed by PRISMA as co ec ed wi h all he AC me hods. OWT 6 is
ins ead cha ac e is ic o b igh wa e wi h high de i us concen a ions, which has a high e lec ance
de e mined by sca e ing. These wa e s a e also cap u ed by PRISMA, wi h some excep ions o MIP
and ACOLITE-T-Ma , which unde es ima e he sho es wa eleng hs. OWT 7, ypical o da k wa e
domina ed by abso p ion om high concen a ion o yellow subs ances, shows high di e gence in
spec al shape o PRISMA L2C and POLYMER, while he o he AC me hods p o ide spec a mo e
uni o m wi h he in si u spec a.
ACIX-III Aqua epo
52
Figu e 26 Mean alues o PRISMA-de i ed R s spec a o each AC me hod, by OWT. Fo in si u da a, AERONET-OC and CVD
ha e been agg ega ed and plo ed as mean alues wi h s anda d de ia ions, om 443 o 667 nm. N p o ides he numbe o
a e aged in si u spec a pe OWT. (Image aken om Figu e 7 [79]).
ACIX-III Aqua epo
53
OWTs we e hen used o build hea maps (Figu e 27) o illus a e pe band and pe OWT pe o mances
es ima ed ia win a es, as compu ed by [18]. The cells o he hea maps o each AC model ha e been
colo ed acco ding o he no malized index (No m(ε)); he AC me hods ha can gene a e he highes -
quali y p oduc s o a gi en OWT and a gi en band a e highligh ed wi h b igh e colo s. The absolu e
alue o ε (%) is also shown in each cell o comple eness.
The hea maps a e compu ed only o he i s i e o he six p e-de ined bands due o he low numbe
o samples in he band a 709 nm. The PRISMA L2C p o ides he lowes pe o mance ac oss all OWTs
and bands, while b igh e colo s gene ally eme ge o MIP, POLYMER and ACOLITE-T-Ma . A close
inspec ion o he hea maps e eals ha , depending on bo h he OWT and he band, each AC me hod
pe o med di e en ly. Fo example, a 443 nm, hGRS showed he lowes alue o ε (33%) o OWT 2,
POLYMER o OWTs 3a, 3b, 4b, 5a and 6, iCOR o OWT 4a (22%) and ACOLITE-T-Ma o OWT 7 (9%).
The Table 7 p o ides an o e iew o he me hod’s pe o mance, based on ε, o each OWT in he i e
signi ican bands.

ACIX-III Aqua epo
54
Figu e 27 Agg ega ion o pai wise in e -compa isons (hea maps) ob ained om he MdSA (%) epo ed in each cell.
P ocesso s wi h ligh e colou s (whi e/yellow) a e likely o gene a e high quali y o a gi en OWT and band. (Image aken om
Figu e 8 [79]).
Table 7 AC models pe o mance by wa eleng h and OWT (including also he T-Ma model). (Table aken om Figu e 8 [79]).
OWT
443 nm
490 nm
560 nm
620 nm
667 nm
2
hGRS
hGRS
ACOLITE
MIP
MIP
3a
POLYMER
MIP
MIP
ACOLITE-T-Ma
MIP
3b
POLYMER
hGRS
hGRS
ACOLITE-T-Ma
POLYMER
4a
iCOR
MIP
POLYMER
MIP
POLYMER
4b
POLYMER
MIP
hGRS
hGRS
MIP
5a
POLYMER
POLYMER
POLYMER
POLYMER
ACOLITE-T-Ma
ACIX-III Aqua epo
55
6
POLYMER
POLYMER
POLYMER
POLYMER
POLYMER
7
ACOLITE-T-Ma
ACOLITE-T-Ma
MIP
ACOLITE-T-Ma
MIP
Addi ionally, he ollowing Table 8 shows he SA esul s o each OWT class. The mean SA o each
class is conside ed, N indica es he class size.
Table 8 A e age SA esul s o each OWT class. N indica es he size o each class.
AC Models
2
(N=10)
3a
(N=26)
3b
(N=57)
4a
(N=22)
4b
(N=82)
5a
(N=16)
6
(N=14)
7
(N=12)
PRISMA
14.3°
19.3°
23.2°
10.0°
26.4°
22.1°
14.4°
34.0°
ACOLITE
31.4°
20.8°
18.9°
7.0°
20.1°
17.4°
21.5°
20.2°
hGRS
9.2°
13.1°
19.6°
6.5°
22.4°
26.9°
11.9°
35.5°
POLYMER
9.3°
12.0°
12.6°
10.3°
13.3°
18.5°
15.2°
74.4°
iCOR
12.6°
14.9°
24.2°
7.3°
17.1°
14.7°
11.7°
18.4°
MIP
10.3°
12.0°
14.9°
10.7°
16.1°
15.4°
18.0°
15.1°
ACOLITE-T-Ma
30.5°
19.2°
19.3°
6.9°
18.1°
15.6°
17.5°
16.6°
5.4 INTERPRETING THE QWIP OUTPUT
The in en o u ilising he QWIP in he ACIX-III Aqua was no o quan i y he compa a i e skill o AC
app oaches, bu ins ead o p o ide a scene-by-scene concep ualiza ion o spec al e lec ance
beha iou and a se o diagnos ic ools o p o ide spa ial/spec al con ex o algo i hm pe o mance.
The QWIP does no p o ide me ics ha con ey he absolu e accu acy o spec al e u ns, bu i does
con ey whe he he spec al e u ns a e ou pu ing easonable shapes ha esemble ypical in si u
e lec ance pa e ns. A known ca ea is ha QWIP is la gely insensi i e o andom noise o small
esidual a mosphe ic ea u es in he spec a. The e may be limi a ions in ega ds o he
ep esen a i eness o he da ase used o de elop he QWIP, howe e , he e appea s o be an
unde lying mechanis ic a ionale o op ical p ope ies o beha e somewha eliably (i.e. some ela i e
co a iance o abso p ion and backsca e ) in na u al wa e s, which likely explains he consis en
endency o he spec al shape o e lec ance o closely ollow he NDI (490,665) o e a b oad ange
o wa e ypes. Ne e heless, alse posi i es and nega i es may be p esen , and he spec al
e lec ance ou pu should be conside ed in he in e p e a ion o esul s. The suppo ing online
ma e ials p o ide nine igu es o each independen scene and AC app oach. Figu e 28 (A, B, C, D)
isually display esul s o he QWIP analysis, while Figu e 29 (A, B, C, D, E) p o ide ou pu o he
ex ac ed spec al e lec ance ma ching a gi en quali y con ol c i e ia de ined by QWIP sco es.
In e p e a ion guidelines, limi a ions, and ca ea s o he esul s and igu es a e lis ed below.
De ailed explana ions o each plo a e gi en as ollows:
● Figu e 28 (A) - Map o Appa en Visible Wa eleng h (AVW). The magni ude o alues
co esponds o he weigh ed ha monic mean o he e lec ance spec a. Concep ually, his
ep esen s he ela i e balance poin o he isible e lec ance spec um. Typically, a ed-
shi ed alues (>580 nm) a e cha ac e ised by high sedimen /CDOM o e y high biomass and
a blue-shi ed alues (440 –450 nm) ep esen clea , oligo ophic condi ions. No e, he
ACIX-III Aqua epo
56
colou -scale used in Figu e 29 ha o he AVW image, o help concep ualise whe e in he
image he spec a a e coming om.
● Figu e 28 (B) - Map o he QWIP sco e. The Quali y Wa e Index Polynomial i sel (Figu e 28 (C),
black line) ep esen s an empi ical ela ionship be ween a 2-band No malised Di e ence
Index (NDI) and he AVW. The di e ence be ween a spec um’s NDI and he QWIP equals i s
QWIP sco e. Spec a co esponding o lowe absolu e QWIP sco es (|QWIP| ≤ 0.2) end o
exhibi ea u es ha a e mo e closely aligned wi h in si u da a (and a e colou ed in g ayscale
he e) while highe absolu e QWIP sco es ( ed/yellow shading; |QWIP| > 0.2) a e lagged as
po en ially suspec da a. The h eshold c i e ia a e based on analysis o in si u da a, howe e ,
hese a e a bi a y guidelines o aid in concep ualiza ion and should no be conside ed
absolu e.
● Figu e 28 (C) - This sca e plo displays how each da a poin in he image compa es o he
QWIP. Di e gen alues ypically exhibi a p opo ional educ ion in spec al quali y. Using a
decision ee based on e lec ance alues, each poin is addi ionally classi ied as one o h ee
wa e ypes [82]. Type I (blue-g een), Type II (g een), and Type III (b own) wa e s ypically ha e
AVW anges o 440 –530 nm, 500 –600 nm, and 540 –600 nm, espec i ely. Spec a a e
independen ly lagged i hey all ou o a gi en wa e ype ange (see Figu e 29 (E)). While no a
uni e sal ule, spec a wi h highe sco es (QWIP > 0.4; Figu e 29 (C)) end o exhibi mo e
eg egious spec al anomalies ela i e o lowe sco es (QWIP < 0.4; Figu e 29 (D)).
● Figu e 28 (D) - A his og am showing he equency dis ibu ion o QWIP sco es, binned in |0.1|
inc emen s. The o al numbe o pixels alling in o each bin a e also shown. Spec al quali y
can a y along a con inuum o QWIP sco es, i.e. exe cise cau ion in designa ing an absolu e
pass/ ail c i e ia.
● Figu e 29 (A) – The mean no malised spec al shapes co esponding o |QWIP| ≤ 0.2, as a
unc ion o AVW. Fo example, all spec a mee ing he c i e ia o 550 nm < AVW < 551 nm &
QWIP < |0.2| a e a e aged and displayed as one ep esen a i e spec a, he spec a is colou -
coded as AVW = 550 nm, and so o h o inc emen al AVW alues anging om AVW = 440 –
600 nm. No e, all spec a a e no malised by he apezoidal in eg a ion o he a ea unde he
e lec ance cu e p io o a e aging. These QWIP c i e ia end o yield spec al e lec ance
shapes ha co espond mos closely o in si u da a, ela i e o highe absolu e QWIP sco es.
This is no always he case, hence he a ailabili y o hese igu es as a quali a i e check. These
spec a co espond o he g ey-scale a eas in Figu e 28 (B), and he colou coding o he
spec a co espond o he AVW map in Figu e 28 (A).
● Figu e 29 (B) - The mean no malised spec al shapes co esponding o an in e media e ange
o QWIP sco es (0.2 > |QWIP| > 0.4), as a unc ion o AVW. Some o hese spec a a e s ill in he
ange o he de elopmen al da ase o QWIP (i.e. hey’ e no all ‘bad’ spec a), bu some
ob ious p ocessing ailu es may s a o p esen in his ange. These a e hei own ca ego ies
ep esen ing a mix o ‘good’ and ‘bad’ spec a.
● Figu e 29 (C) - The mean no malised spec al shapes co esponding o QWIP > 0.4, as a
unc ion o AVW. While no a unila e al ule, QWIP sco es mee ing his c i e ion end o
ACIX-III Aqua epo
57
cap u e some e iden AC ailu es. O all he QWIP ou pu , his pa icula c i e ion ends o
consis en ly lag p oblema ic spec a, wi h ew Type II excep ions.
● Figu e 29 (D) - The mean no malised spec al shapes co esponding o QWIP < -0.4, as a
unc ion o AVW. While his c i e ia will o en cap u e AC ailu es, some imes o he wise good-
looking spec al shapes may mee his c i e ia i a la ge po ion o he spec um con ains
sligh ly nega i e o nea -ze o alues, i.e. be on he lookou o easonable shapes ha we e
pe haps jus o e co ec ed, usually in he a -blue o a - ed po ion o he spec um.
● Figu e 29 (E) – The mean no malised spec al shapes co esponding o spec a wi h an AVW –
wa e ype misma ch. Acco ding o [19], some imes a spec um will ha e a low QWIP sco e
(indica ing be e quali y), bu he wa e ype [82] seems o all ou o ange o he expec ed
AVW alue. Fo example, in his ins ance he e we e se e al “g een” wa e s ha we e shi ed
much oo a owa ds he blue end o he spec um compa ed o wha we ypically see wi h in
si u da a. These c i e ia end o be obus in e ms o lagging suspec spec a.
Figu e 28 An example o QWIP analysis ou pu o a scene a A iake Towe AERONET-OC si e on May 11, 2020, p ocessed
using he hGRS AC app oach. (A) Map o AVW, (B) Map o he QWIP sco e, (C) sca e plo o QWIP analysis, and (D)
equency dis ibu ion o inc emen al QWIP sco es.
ACIX-III Aqua epo
64
excluded om he MIP and Polyme e ie als in he AC in e compa ison exe cise. Fo he ela i e
compa ison be ween he ACs, i can be expec ed ha all p ocesso s a e a ec ed o a simila ex en by
he ins umen unce ain ies [18]. To check o he high empo al a iabili y o he in si u
measu emen s close o he EnMAP o e pass ime, we calcula ed he mean and s anda d de ia ion a
he ± 15 minu e window when da a we e a ailable. Fo h ee days a LJCO he in si u e lec ance
showed high empo al a iabili y, and we op ed o compa e EnMAP da a o he mean in si u
e lec ance ins ead o using he closes measu emen in ime, as o all o he in si u ma chups.
Table 10 Band se ings in he di e en da ase s: he nominal band used in he ma ch-up analysis and he co esponding band
se ings o AERONET-OC and EnMAP. The i s pa o he able e e s o he band se ing used o he en i e da ase , while he
second pa e e s o he addi ional bands conside ed o he CVD da ase only.
Nominal band
AERONET-OC band se ing
EnMAP band se ing
AERONET-OC & CVD da ase s
443 nm
443 ± 5 nm
444.7 ± 3.0 nm
490 nm
490 ± 5nm
491.8 ± 2.9 nm
560 nm
560 ± 5 nm
561.1 ± 3.2 nm
620 nm
620 ± 5 nm
622.9 ± 3.6 nm
667 nm
667 ± 5 nm
666.6 ± 3.8 nm
EnMAP e ie als we e compa ed o he in si u measu emen s ollowing he OLCI ma ch-up p o ocol
[94] and using a 3 x 3-pixel box and a ime window o ± 01:15 h. Mos o he measu emen s we e
a ailable wi hin ± 15 minu es om he o e pass (36 ou o 50); 17 du ing he o e pass. EnMAP-MIP
pixels we e excluded om analysis when hey we e masked as ci us, cloud, cloud shadow, and haze
lags. Fo Polyme , we ound ha he de aul lags did no always accu a ely co espond o in alid
ma ch-ups, so we decided no o apply any lags and in es iga e his issue u he . Fo ACOLITE we
applied he de aul ecommended lags: NIR o SWIR h eshold es , CIRRUS h eshold es , TOA
h eshold es , NEGATIVE su ace e lec ance es and EXTENT es . Mos o he pixels o he 3 x 3-pixel
box we e e ained in he end, usually 7 o 9 pixels, o he alid ma ch-ups. The EnMAP spec um
ep esen s he median o he pixel box a e he lagged and ou lie pixels we e emo ed. The numbe
o alid ma ch-ups o each AC p ocesso o hype - and mul ispec al in si u da a was 22 and 24 o
MIP, 21 and 20 o Polyme , 11 and 17 o ACOLITE wi hou and 3 and 8 o ACOLITE wi h glin
co ec ion applied, and 21 and 30 o PACO-WASI, espec i ely. The equency dis ibu ion o he R s
le els o he in si u ma chup da a a e p o ided in Figu e 32.
As o he PRISMA e alua ion, we ollowed he s a is ical me ics de ailed in sec ion 2.4.3. We also
analysed he spec al shape o EnMAP using he spec al angle mappe (SAM, θ◦) o de e mine he
simila i y be ween an EnMAP and an in si u e e ence spec um. . Spec al Angles (SA) we e
calcula ed on he mean spec a.

ACIX-III Aqua epo
65
Figu e 32 F equency dis ibu ion o e lec ance le els he hype spec al (CVD) and mul ispec al in si u measu emen s (all
wa eleng hs).
6.5 PERFORMANCE ASSESSMENT
In his sec ion, esul s om he compa ison o EnMAP s anda d L2A (MIP) p oduc s and ACOLITE,
POLYMER and PACO-WASI AC models a e p esen ed o he compa ison o AERONET-OC and o he
hype spec al ma chups. Fi s ly, he ma chups a indi idual si es a e discussed. Then he s a is ical
esul s o e all common ma chups o he ou AC models MIP, POLYMER, ACOLITE and PACO-WASI
a e p esen ed and in e compa ed. The s udy [22] addi ionally p esen s s a is ical esul s o all
ma chups o EnMAP-MIP o hype spec al and mul ispec al in si u da a which we e much mo e (22
and 24, espec i ely) hen he ones in common o all h ee AC models (11 and 14, espec i ely).
6.5.1 MULTISPECTRAL PERFORMANCE
Figu e 33 - Figu e 36 show he spec al compa isons be ween sa elli e and in si u da a o he nine
common AERONET-OC si es among he ou EnMAP AC models. Fo some si es se e al ma chups
we e a ailable, bu only one is shown as example. Fo EnMAP MIP da a, wo mo e ma chups we e
a ailable (MVCO and I be Ligh house), and also one mo e o EnMAP Polyme (MVCO). Example
spec a o hese si es a e shown as well. O e all, he R s spec a o AERONET-OC peaks sligh ly
lowe han 0.002 s -1 in he blue egion o maxima o a ound 0.025 s -1 in he g een egion. A b ie
cha ac e iza ion o he si es and hei example ma chup ollow:
● Venice (AAOT) and Socheongcho si es, whe e R s spec a a e highe in he blue egion and
end o dec ease a e abou 505 nm, ep esen he ypical shape o clea wa e spec a. R s is
e y low o Socheongcho (<0.004 s -1) and shows a he la ge s anda d de ia ion (~20%).
● Bahia Blanca, Banana Ri e , Chesapeake Bay, I be Ligh house, San Ma co pla o m, Sou h
G eenbay si e, and Kemigawa si es, whe e R s spec a a e highe in he g een egion (a ound
560 nm), ep esen he ypical shape o u bid wa e spec a. While Kemigawa shows he
ACIX-III Aqua epo
66
lowes R s alues (<0.002 s -1), he Bahia Blanca si e p esen s he peak in he g een egion
wi h he highes alue o R s (<0.025 s -1); his si e is cha ac e ised as an a ea subjec o
di use e osion and s ong idal cu en s which a e esponsible o he ypically high
suspended loads in he channel, also he s anda d de ia ion is la ge which is simila o
Chesapeake Bay.
● In he Sou h G eenbay si e, he peak a ound 710 nm is e y e iden , compa ed o he o he
cases, his could be due o he ac ha his si e is cha ac e ised by a condi ion o hype -
eu ophica ion (supe - ichness o nu ien s). A less no iceable peak a ound 680 nm is also
isible in he case o LISCO si e.
● In he Lucinda si e (in he opical coas al wa e s o he G ea Ba ie Ree lagoon nea he
He be Ri e es ua y), he spec um shows a peak jus below 550 nm and a small s anda d
de ia ion o ± 0.0005 s -1.
MIP
Excep o Kemigawa a all bands and o Lucinda in he blue, EnMAP-MIP ma ches well he AERONET-
OC bands magni udes and shape (Figu e 33, SA <20°). Fo Kemigawa he e is a all bands an
o e es ima ion by EnMAP-MIP, while a Lucinda o bands 400 o 490 nm EnMAP-MIP is lowe han
AERONET-OC. The s anda d de ia ion in R s measu ed by ce ain in si u measu emen s e lec ing
highe a iabili y in he en i onmen al condi ions is simila ly ob ained by he EnMAP-MIP e ie als
s anda d de ia ion wi hin he 3x3 ma chup pixel box.
Figu e 33 Spec al compa isons be ween EnMAP L2A wa e p oduc s (MIP) and in si u da a in he 11 globally dis ibu ed
AERONET-OC si es. The a iabili y ac oss he da ase in he mean spec a o MIP is displayed as blue cu es, wi h he shaded
ACIX-III Aqua epo
67
blue a ea ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown in
ed. SA ep esen s he Spec al Angle and N ep esen s he numbe o he images.
ACOLITE
EnMAP-ACOLITE o e es ima es R s a all si es signi ican ly (Figu e 34). De ia ions in he spec al
shape a e pa ly la ge (SA 19°-40°, Sou h G eenbay, Kemigawa, Chesapeake Bay and Banana Ri e ),
bu also a he low o he es (SA 4.5-<15°). The s anda d de ia ion in R s measu ed by ce ain in si u
measu emen s is simila o he EnMAP-ACOLITE e ie als s anda d de ia ion wi hin he 3x3 ma chup
pixel box o Chesapeake Bay while i is much highe o he h ee o he si es. O e all, he spec al
shape ag ees wi h he AERONET-OC obse a ions; howe e , excep o Lucinda, Kemigawa, and Bahía
Blanca, he de ia ions a e conside ably la ge han hose obse ed o EnMAP-MIP.
Figu e 34 Spec al compa isons be ween ACOLITE and in si u da a in he 9 globally dis ibu ed AERONET-OC si es. The
a iabili y ac oss he da ase in he mean spec a o ACOLITE is displayed as blue cu es, wi h he shaded blue a ea
ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown in ed. SA
ep esen s he Spec al Angle and N ep esen s he numbe o he images.
POLYMER
EnMAP-Polyme de ia es signi ican ly in R s a all si es, excep o Bahia Blanca and Chesapeake Bay
whe e he ag eemen is wi hin he s anda d de ia ion o AERONET-OC (Figu e 35). Mos ly he alues
a e much highe in he blue, and much lowe in he ed compa ed o AERONET-OC. De ia ion in he
spec al shape a e he la ges among he ou AC models: SA ange om ~9° o Bahia Blanca and
Lucinda, o ~15°-25° o Venice (AAOT), San Ma co Pla o m, Socheongcho and MVCO and be ween
40° o 67° o he emaining ou si es. The s anda d de ia ion in R s measu ed by ce ain in si u
ACIX-III Aqua epo
68
measu emen s o AERONET-OC is only sligh ly la ge o he EnMAP-ACOLITE e ie als s anda d
de ia ion wi hin he 3x3 ma chup pixel box.
Figu e 35 Spec al compa isons be ween POLYMER and in si u da a in he 10 globally dis ibu ed AERONET-OC si es. The
a iabili y ac oss he da ase in he mean spec a o POLYMER is displayed as blue cu es, wi h he shaded blue a ea
ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown in ed. SA
ep esen s he Spec al Angle and N ep esen s he numbe o he images.
PACO-WASI
Excep o Kemigawa and MVCO, he e lec ance de i ed by PACO-WASI ma ches well he AERONET-
OC measu emen s in magni ude and shape (Figu e 36). Fo Kemigawa, PACO-WASI o e es ima es
e lec ance in all bands s ongly, and o MVCO i o e es ima es i om 400 o 550 nm mode a ely.
Wi h R s < 0.001 s -1, Kemigawa has om all si es by a he lowes e lec ance in he blue, which
makes a mosphe ic co ec ion pa icula ly challenging. Mo e han one ma chup was a ailable o
Bahia Blanca, Chesapeake Bay, Socheongcho and Lucinda. Fo he i s h ee, he s anda d de ia ion
o he in si u measu emen s is high, e lec ing a high empo al a iabili y in he en i onmen al
condi ions. A e y simila a iabili y is ob ained om PACO-WASI. The o e lapping s anda d
de ia ions o bo h da ase s indica e a good co espondence o PACO-WASI e lec ance wi h he in
si u measu emen s.
ACIX-III Aqua epo
69
Figu e 36 Spec al compa isons be ween PACO-WASI and in si u da a in he 12 globally dis ibu ed AERONET-OC si es. The
a iabili y ac oss he da ase in he mean spec a o PACO-WASI is displayed as blue cu es, wi h he shaded blue a ea
ep esen ing he s anda d de ia ion. The mean and s anda d de ia ion o he in si u da a a e equi alen ly shown in ed. SA
ep esen s he Spec al Angle and N ep esen s he numbe o he images.
Summa y
The common mul ispec al da ase o EnMAP con ains 14 ma ch-ups wi h AERONET-OC da a om
nine s udy si es: one a AAOT, ou a Bahia Blanca, one a Banana Ri e , wo a Cheasepeake Bay,
one a Kemigawa O sho e, one a San Ma co Pla o m, one a Socheongcho, one a Sou h G eenbay
and wo a LJCO. Figu e 37 shows he o e all pe o mance o AC p ocesso s using all he AERONET-
OC ma ch-ups o each EnMAP AC model sepa a ely wi h all he EnMAP sa elli e da a combined.
Table 11 summa izes he s a is ical me ics o he AC model pe o mance o each e e ence band as
plo ed in Figu e 37.
Rega ding RMSD and MAD, he bes pe o mance, i.e. he lowes alues a he i e bands, was
ob ained o PACO-WASI, closely ollowed by MIP. The wo o he AC models show much highe
alues. Fo MAPD, MIP is bes (excep a 620 nm whe e POLYMER is sligh ly be e ), ollowed by
POLYMER and PACO-WASI (which is e en sligh ly be e han MIP a 667 nm). The mean bias is bes a
443 nm and 490 nm o MIP; a 560 nm o POLYMER and PACO-WASI; a 620 nm o PACO-WASI, and
a 667 nm o POLYMER ( e y closely ollowed by MIP). Gene ally, he mean bias is qui e high a 443
nm o all ou me hods (wo s o PACO WASI wi h 131%). While ACOLITE pe o med wo s o he
p e ious me ics, i pe o med simila ly well o PACO-WASI and e en be e han PACO-WASI a 443

ACIX-III Aqua epo
70
nm. MIP showed he weakes esul s he e. Conce ning he slope, PACO-WASI is closes o 1 o all
bands, excep a 443 nm, whe e ACOLITE is sligh ly be e . Fo all me hods, he eg ession a 443 nm
is much wo se han o all o he bands, anging om an i-co ela ion in case o POLYMER (slope = -
0.24) o no mo e han 0.75 (ACOLITE). POLYMER and MIP s ill show no con enien eg ession a 490
nm (slopes o 0.61 and 0.71, espec i ely), bu ACOLITE and PACO-WASI a e close o 1. All me hods
e eal slopes be ween 0.75 and 1.09 a he emaining bands a 560 nm, 620 nm and 667 nm.
Figu e 37 O e all pe o mance o AC p ocesso s using he AERONET-OC ma ch-ups wi h all he sa elli e da a combined. The
numbe o ma chups pe p ocesso and pe band is epo ed in Table 11 along wi h he s a is ics. The black lines e e o he
1:1 line.
ACIX-III Aqua epo
71
Table 11 Summa y o he i e s a is ical me ics o he EnMAP ma chup da a shown in Figu e 37. N ep esen s he o al
numbe o samples o each band.
Figu e 38 summa izes he esul s o he s a is ical pa ame e s ϵ and β. In e ms o Median Symme ic
Accu acy (ϵ, Figu e 38 le panel), PACO-WASI pe o ms bes , ollowed by MIP. PACO-WASI mee s he
30% h eshold [18] e y well o all bands excep 443 nm (38%), and MIP mee s he 30% h eshold well
o he g een and ed bands, while he e o is much highe a 443 nm (110%) and sligh ly abo e he
h eshold o 490 nm (32%). ACOLITE and POLYMER exhibi highe e o s ac oss all i e bands, mos
no ably o he blue bands (>95% and >75%, espec i ely). POLYMER shows highe accu acy han
ACOLITE bu has he la ges e o a 443 nm among he ou AC models.
Rega ding he Median Symme ic Bias (β, Figu e 38, igh panel), PACO-WASI shows he lowes
median symme ic Bias, which is posi i e (5-10%) o all bands excep 443 nm (-7%). MIP shows he
second lowes alues o β, which a e nega i e o all bands (la ges o 443 nm wi h 65%, lowe o he
o he bands wi h 5-15 %). POLYMER shows a highe posi i e β han MIP’s nega i e β o he blue
bands, and simila nega i e bias o he o he bands. ACOLITE ob ains highes β alues which a e
always posi i e.
One has o keep in mind ha in his in e compa ison, PACO-WASI is based on an imp o ed e sion o
he EnMAP L1 and L2A da a as compa ed o he o he h ee AC me hods which may explain some o
he be e pe o mance. Fu he mo e, p ocessing was no done au oma ically as o he o he
so wa es bu equi ed ime-consuming op imiza ion o in e sion by expe s. In addi ion, he numbe
o alid ma chups a ies among he di e en AC me hods, bu also among he di e en wa eleng hs.
ACIX-III Aqua epo
72
Compa isons only conside ing he same ma chups a e p e e ed o a anking o me hods, as has
been p o ided o EnMAP-MIP, -POLYMER and -ACOLITE in [22].
Figu e 38 Pe o mance assessmen s as de e mined by he Median Symme ic Accu acy (ϵ) and median symme ic Bias (β) o
all he ma ch-ups combined. The dashed line co esponds o a 30% h eshold [18].
6.5.2 HYPERSPECTRAL PERFORMANCE
Figu es 39, 40, 41 and 42 show examples o he spec al compa isons be ween sa elli e and in si u
da a o he ou -common hype spec al in si u da a si es (AAOT, Lake Cons ance, Lampedusa,
Lucinda) among he ou EnMAP AC models. Addi ionally, o EnMAP-MIP wo mo e ma chups we e
a ailable (Oos ende and Lake T asimeno), while o EnMAP-Polyme only ma chups we e a ailable o
Lake T asimeno and o EnMAP-ACOLITE only o Oos ende. Some o hese si es had se e al
ma chups, bu only one is shown as example. O e all, he R s spec a o hype spec al in si u peaks
sligh ly lowe han 0.007 s -1 in he blue egion o maxima o a ound 0.027 s -1 in he g een egion. A
b ie cha ac e isa ion o he si es and hei example ma chup ollows:
● Lampedusa whe e R s spec a a e highes a 400nm and con inuously dec easing o he ed,
showing e y clea (open ocean) wa e spec a, al hough he wa e dep h a he s a ion is jus
20 m. R s is qui e low (<0.007 s -1) bu no as low as ound o he AERONET-OC si e
Socheongcho (Chap e 4.2.1).
● Venice (AAOT), whe e R s spec a peak a 500-530 nm and a e a bi highe in he g een han he
blue egion, dec ease a lo in he ed, bu s ill ha e some con ibu ion a 600 o 700 nm,
showing mo e meso ophic wi h li le CDOM bu some sedimen . I shows less clea wa e
condi ions as opposed o he ma chup shown o AAOT in 4.2.1. The e is qui e a la ge s anda d
de ia ion (>20%) indica ing dynamic (mos likely suspended sedimen ) condi ions du ing
ma chup.
● Lake Cons ance: a a he deep (mean dep h 100m), subalpine glacial lake, whe e R s spec a
peak a 535 nm, a e much highe in he g een han he blue egion, indica ing signi ican CDOM
ACIX-III Aqua epo
73
loading as compa ed o sedimen s. The e is a s ong dec ease in he ed pa , bu s ill some
R s le . I indica es meso ophic wa e condi ions wi h a he low chlo ophyll-a concen a ion.
● Lucinda is in he opical coas al wa e s o he G ea Ba ie Ree lagoon nea he He be Ri e
es ua y; he spec um shows a peak a 560 nm and also a a he la ge s anda d de ia ion up o
50% indica ing highly a iable suspended loads du ing he ma chup.
● Oos ende is a coas al si e whe e R s spec a a e highe in he g een egion (a ound 560 nm),
ep esen he ypical shape o he u bid wa e spec a, bu also signi ican con ibu ion o
CDOM abso p ion. I p esen s he peak in he g een egion wi h he highes alue o R s
(<0.027 s -1); his si e is cha ac e ised as an a ea subjec o di use e osion and s ong idal
cu en s which a e esponsible o he ypically high suspended loads in he channel.
● Lake T asimeno (meso-eu ophic): is a shallow ec onic lake, wi h he R s peak a ound 0.0025
s -1 in he g een egion. The spec um shows e lec ance peaks and dips in he ange 670-700
nm ela ed o he p esence o Chl-a.
MIP
O e all, we obse e e y good ag eemen be ween EnMAP-MIP and in si u spec a o he di e en
wa e si es and e lec ance magni udes and he spec al shape is well kep (SA be ween 2.9 o 8.2°;
Figu e 39). The highes disag eemen was obse ed a Oos ende, a egion wi h ypical u bid and e y
u bid wa e s [48]. The unde es ima ion o MIP a Oos ende migh be imp o ed by he new EnMAP
p ocesso e sion (in oduced in Ma ch 2025) ha should ix he issue o he di e en wa e ypes in
he MIP p ocesso , as now all p oduc s a e e ie ed using he clea wa e op ion. Th ee low-quali y
images o LJCO we e acqui ed on da es (July 08, Augus 15, and Augus 16, 2022) when in si u
e lec ance exhibi ed high empo al a iabili y (highligh ed by he shaded g een egion). As p e iously
men ioned, on hese days, we compa ed he EnMAP da a o he mean o he 15-minu e window
(indica ed by he dashed g een line). In he ed-NIR egion, he e a e s ill a mosphe ic abso p ion
ea u es ha a e no comple ely emo ed, in pa icula he O2 band a ound 760 nm. O e all, spec al
noise, specially a sho e wa eleng hs, is obse ed in he EnMAP-MIP p oduc and i is mos ly due o
an insu icien sampling in he con olu ion o compu e wa e look-up- ables in MIP. This issue was
ecen ly ixed and will be implemen ed in he nex EnMAP p ocesso e sion. These band- o-band
spec al a ia ions a e also obse ed in he PRISMA da a by [24] and [76] and CHRIS-PROBA by [95],
possibly a esul o in e -band calib a ion issues.
ACIX-III Aqua epo
80
was sligh ly highe han o AERONET-OC; indeed, o CVD mos o he AC me hods p oduced da a
ha sa is ied he 30% accu acy h eshold ecommended by he GCOS o R s.
Addi ionally, each a mosphe ic co ec ion p ocesso showed a ying deg ees o accu acy depending
on he Op ical Wa e Type (OWT); hese da a we e in ac ep esen a i e o eigh OWTs ha
encompass blue, phy oplank on- ich, u bid and humic wa e s in ma ine and lacus ine ecosys ems.
Analysing he hea maps helped illus a e how me hod pe o mance depends on wa e p ope ies,
indica ing he AC me hod wi h he lowes ε o each combina ion o OWT and spec al band. OWT
analysis was no conduc ed o EnMAP due o he lowe numbe o samples.
To conclude, ACIX-III Aqua has p o ided a aluable amewo k o unde s anding how s a e-o - he-a
a mosphe ic co ec ion me hods pe o m when applied o PRISMA and EnMAP, which ha e been
o e ing ele an obse a ions o wa e applica ions. In pa icula , he s udy p o ides an o e iew o
he compa ison o ield- and sa elli e-de i ed R s alues o speci ic PRISMA bands o e 26 globally
dis ibu ed inland and coas al wa e si es, using a da ase spanning om 2020 o 2024. None heless,
some elemen s ha e o be conside ed o imp o ing he exe cise:
• Some o he AC me hods a e also p o iding es ima es o AOT; a compa ison o such
es ima ion wi h e e ence da a om AERONET and CVD ne wo ks (e.g., Mic o ops II) would
p o ide u he elemen s o e alua e he AC me hods.
• The QWIP analysis, p esen ed he e jus as an example, would me i u he in es iga ion o
ob ain a ision ac oss he en i e image (30 km by 30 km) a he han he di ec compa ison in a
limi ed po ion (i.e. 90 m x 90 m); his would allow o app ecia e how AC me hods pe o m in
shallow wa e s, wi h clouds e c.
• A deepe analysis should be unde aken o conside he unce ain ies (e.g., geo-loca ion)
associa ed wi h bo h in si u and sa elli e da a, and o e alua e he impac on he es ima ion o
biophysical pa ame e s, as pe o med in p e ious ACIX-Aqua s udies [18].
• Bo h PRISMA and EnMAP ha e no been speci ically designed o mee he obse a ional
equi emen s (e.g., high signal o noise a io) o de eloping aqua ic applica ions so ha he
AC esul s a e o cou se depending on he p ope ies o hese sa elli e da a.
• Fu he analysis is equi ed o unde s and how he a ia ions in AC pe o mance a e ela ed o
he inc eased hype spec al in o ma ion con en o PRISMA/EnMAP and how he esul s o his
exe cise compa e o hose ob ained wi h o he sa elli es (e.g. PACE).
The analysis is hence no ending and we do hope ha such wo k will p osecu e also o suppo he
gene a ion o imp o ed p oduc s o aqua ic applica ions om hype spec al da a collec ed by u u e
mission, such as CHIME and SBG.

ACIX-III Aqua epo
81
8 ACKNOWLEDGMENTS
This wo k would no ha e been possible wi hou access o high-quali y in si u da a. We a e deeply
g a e ul o he AERONET-OC P incipal In es iga o s: J. Ishizaka, K. A ai, H. Loisel, P. P a olongo, R.
F ouin, M. Talone, F. Mélin, B. Bulga elli, G. Zibo di, H. Higa, T. Moo e, S. Rube g, M. Wang, S. Ahmed,
A. Gile son, T. Sch oede , Y.J. Pa k, B. Jones, M. Ragan, D. Au in, H. Sosi, S. Ladne , and D. Van de
Zande. We also acknowledge he con ibu ions o he Join Resea ch Cen e (JRC) o he Eu opean
Commission, JAXA’s GCOM-C p ojec , he In eg a ed Ma ine Obse ing Sys em (IMOS)—enabled by
he Na ional Collabo a i e Resea ch In as uc u e S a egy (NCRIS)—and he HYPERNETS H2020
p ojec o p o iding hype spec al da a om AAOT and Lampedusa. Special hanks go o Mau o
Musan i and Sal a o e Mangano o hei ieldwo k in I aly; Ma a Gomes, Milad Ni oumand-Jadidi, he
Phy oop ics g oup membe s, and LUBW/ISF o hei suppo du ing he Lake Cons ance EnMAP
alida ion campaigns. We a e g a e ul o Angelo Amodio and Luigi Ag imano om Plane ek o hey
aluable suppo in handling PRISMA p oduc s. We hank Robe a Pozzi and Simone Lella om CNR-
IREA o suppo ing da a handling and s o age. We a e also g a e ul o all con ibu o s and suppo e s
o he EnMAP Cal/Val Wa e P ojec . We since ely hank Shun Bi o aluable discussions on op ical
wa e ypes. This wo k was suppo ed by: he I alian Space Agency h ough he PRISCAV P ojec (g an
no. 2019-5-HH.0) and PANDA-WATER P ojec (con ac ASI no. 2022-15-U.0); he Eu opean Space
Agency (con ac no. 4000139081/22/I-EF, HYPERNETS-POP); he Ho izon 2020 HYPERNETS p ojec
(g an ag eemen no. 775983); he Bundesminis e ium ü Wi scha und Klimaschu z (g an s
50EE1915, 50EE1923, 50EE2401); he DLR o unding he EnMAP mission; The Swiss Na ional Science
Founda ion (SNSF) unde he Lake3P p ojec (g an no. 204783); The Swedish Na ional Space Agency
(Dn . 2021-00050); he Belgian Science Policy O ice h ough he TERRASCOPE p ojec (con ac no.
CB/67/12).
ACIX-III Aqua epo
82
9 REFERENCES
1. S ong, A.E. Remo e Sensing o Algal Blooms by Ai c a and Sa elli e in Lake E ie and U ah
Lake. Remo e Sens En i on 1974, 3, 99–107, doi:10.1016/0034-4257(74)90052-2.
2. Shi, W.; Wang, M. Th ee‐dimensional Obse a ions om MODIS and CALIPSO o Ocean
Responses o Cyclone Na gis in he Gul o Ma aban. Geophys Res Le 2008, 35,
doi:10.1029/2008GL035279.
3. S ump , R.P.; Wynne, T.T.; Bake , D.B.; Fahnens iel, G.L. In e annual Va iabili y o
Cyanobac e ial Blooms in Lake E ie. PLoS One 2012, 7, e42444,
doi:10.1371/jou nal.pone.0042444.
4. B esciani, M.; Adamo, M.; De Ca olis, G.; Ma a, E.; Pasqua iello, G.; Vaičiū ė, D.; Gia dino, C.
Moni o ing Blooms and Su ace Accumula ion o Cyanobac e ia in he Cu onian Lagoon by
Combining MERIS and ASAR Da a. Remo e Sens En i on 2014, 146, 124–135,
doi:10.1016/j. se.2013.07.040.
5. Ma hews, M.W.; Ode ma , D. Imp o ed Algo i hm o Rou ine Moni o ing o Cyanobac e ia
and Eu ophica ion in Inland and Nea -Coas al Wa e s. Remo e Sens En i on 2015, 156, 374–
382, doi:10.1016/j. se.2014.10.010.
6. Tyle , A.N.; Hun e , P.D.; Spy akos, E.; G oom, S.; Cons an inescu, A.M.; Ki chen, J.
De elopmen s in Ea h Obse a ion o he Assessmen and Moni o ing o Inland, T ansi ional,
Coas al and Shel -Sea Wa e s. Science o The To al En i onmen 2016, 572, 1307–1321,
doi:10.1016/j.sci o en .2016.01.020.
7. Pahle an, N.; Scho , J.R.; F anz, B.A.; Zibo di, G.; Ma kham, B.; Bailey, S.; Schaa , C.B.;
Ond usek, M.; G eb, S.; S ai , C.M. Landsa 8 Remo e Sensing Re lec ance (R s) P oduc s:
E alua ions, In e compa isons, and Enhancemen s. Remo e Sens En i on 2017, 190, 289–301,
doi:10.1016/j. se.2016.12.030.
8. Hedley, J.D.; Roel sema, C.; B ando, V.; Gia dino, C.; Ku se , T.; Phinn, S.; Mumby, P.J.;
Ba ile o, O.; Lapo e, J.; Koe z, B. Co al Ree Applica ions o Sen inel-2: Co e age,
Cha ac e is ics, Ba hyme y and Ben hic Mapping wi h Compa ison o Landsa 8. Remo e Sens
En i on 2018, 216, 598–614, doi:10.1016/j. se.2018.07.014.
9. Pahle an, N.; Balasub amanian, S. V.; Sa ka , S.; F anz, B.A. Towa d Long-Te m Aqua ic
Science P oduc s om He i age Landsa Missions. Remo e Sens (Basel) 2018, 10, 1337,
doi:10.3390/ s10091337.
10. Cazzaniga, I.; B esciani, M.; Colombo, R.; Della Bella, V.; Padula, R.; Gia dino, C. A Compa ison
o Sen inel-3-OLCI and Sen inel-2-MSI-De i ed Chlo ophyll- a Maps o Two La ge I alian Lakes.
Remo e Sensing Le e s 2019, 10, 978–987, doi:10.1080/2150704X.2019.1634298.
11. B esciani, M.; Pina di, M.; F ee, G.; Luciani, G.; Gheb ehiwo , S.; Laanen, M.; Pe e s, S.; Della
Bella, V.; Padula, R.; Gia dino, C. The Use o Mul isou ce Op ical Senso s o S udy
Phy oplank on Spa io-Tempo al Va ia ion in a Shallow Tu bid Lake. Wa e (Basel) 2020, 12, 284,
doi:10.3390/w12010284.
ACIX-III Aqua epo
83
12. B esciani, M.; Rossini, M.; Mo abi o, G.; Ma a, E.; Pina di, M.; Coglia i, S.; Juli a, T.; Colombo,
R.; B aga, F.; Gia dino, C. Analysis o Wi hin- and be ween-Day Chlo ophyll-a Dynamics in
Man ua Supe io Lake, wi h a Con inuous Spec o adiome ic Measu emen . Ma F eshw Res
2013, 64, 303, doi:10.1071/MF12229.
13. Hes i , E.L.; B ando, V.E.; B esciani, M.; Gia dino, C.; Ma a, E.; Villa, P.; Dekke , A.G.
Measu ing F eshwa e Aqua ic Ecosys ems: The Need o a Hype spec al Global Mapping
Sa elli e Mission. Remo e Sens En i on 2015, 167, 181–195, doi:10.1016/j. se.2015.05.023.
14. Dekke , A.G.; Vos, R.J.; Pe e s, S.W.M. Compa ison o Remo e Sensing Da a, Model Resul s
and in Si u Da a o To al Suspended Ma e (TSM) in he Sou he n F isian Lakes. Science o The
To al En i onmen 2001, 268, 197–214, doi:10.1016/S0048-9697(00)00679-3.
15. Die ssen, H.M.; Ackleson, S.G.; Joyce, K.E.; Hes i , E.L.; Cas agna, A.; La ende , S.; McManus,
M.A. Li ing up o he Hype o Hype spec al Aqua ic Remo e Sensing: Science, Resou ces and
Ou look. F on En i on Sci 2021, 9, doi:10.3389/ en s.2021.649528.
16. S iebel, M.; Behl, S.; Diehl, S.; S ibo , H. Spec al Niche Complemen a i y and Ca bon
Dynamics in Pelagic Ecosys ems. Am Na 2009, 174, 141–147, doi:10.1086/599294.
17. Doxani, G.; Ve mo e, E.; Roge , J.-C.; Gascon, F.; Ad iaensen, S.; F an z, D.; Hagolle, O.;
Holls ein, A.; Ki ches, G.; Li, F.; e al. A mosphe ic Co ec ion In e -Compa ison Exe cise.
Remo e Sens (Basel) 2018, 10, 352, doi:10.3390/ s10020352.
18. Pahle an, N.; Mangin, A.; Balasub amanian, S. V.; Smi h, B.; Alikas, K.; A ai, K.; Ba bosa, C.;
Bélange , S.; Binding, C.; B esciani, M.; e al. ACIX-Aqua: A Global Assessmen o A mosphe ic
Co ec ion Me hods o Landsa -8 and Sen inel-2 o e Lakes, Ri e s, and Coas al Wa e s.
Remo e Sens En i on 2021, 258, 112366, doi:10.1016/j. se.2021.112366.
19. Die ssen, H.M.; Vande meulen, R.A.; Ba nes, B.B.; Cas agna, A.; Knaeps, E.; Vanhellemon , Q.
QWIP: A Quan i a i e Me ic o Quali y Con ol o Aqua ic Re lec ance Spec al Shape Using
he Appa en Visible Wa eleng h. F on ie s in Remo e Sensing 2022, 3,
doi:10.3389/ sen.2022.869611.
20. S o ch, T.; Honold, H.-P.; Chab illa , S.; Habe meye , M.; Tucke , P.; B ell, M.; Ohndo , A.;
Wi h, K.; Be z, M.; Kuchle , M.; e al. The EnMAP Imaging Spec oscopy Mission owa ds
Ope a ions. Remo e Sens En i on 2023, 294, 113632, doi:10.1016/j. se.2023.113632.
21. Chab illa , S.; Foe s e , S.; Segl, K.; Beamish, A.; B ell, M.; Asadzadeh, S.; Milewski, R.; Wa d,
K.J.; B osinsky, A.; Koch, K.; e al. The EnMAP Spacebo ne Imaging Spec oscopy Mission: Ini ial
Scien i ic Resul s Two Yea s a e Launch. Remo e Sens En i on 2024, 315, 114379,
doi:10.1016/j. se.2024.114379.
22. Soppa, M.A.; B ell, M.; Chab illa , S.; Al a ado, L.M.A.; Gege, P.; Pla ne , S.; Somlai-Schweige ,
I.; Sch oede , T.; S einme z, F.; Sche le , D.; e al. Full Mission E alua ion o EnMAP Wa e
Lea ing Re lec ance P oduc s Using Th ee A mosphe ic Co ec ion P ocesso s. Op Exp ess
2024, 32, 28215, doi:10.1364/OE.523813.
23. Zibo di, G.; Mélin, F.; Be hon, J.-F.; Holben, B.; Slu ske , I.; Giles, D.; D’Alimon e, D.;
Vandema k, D.; Feng, H.; Schus e , G.; e al. AERONET-OC: A Ne wo k o he Valida ion o
Ocean Colo P ima y P oduc s. J A mos Ocean Technol 2009, 26, 1634–1651,
doi:10.1175/2009JTECHO654.1.
ACIX-III Aqua epo
84
24. B aga, F.; Fabb e o, A.; Vanhellemon , Q.; B esciani, M.; Gia dino, C.; Sca pa, G.M.; Man è, G.;
Concha, J.A.; B ando, V.E. Assessmen o PRISMA Wa e Re lec ance Using Au onomous
Hype spec al Radiome y. ISPRS Jou nal o Pho og amme y and Remo e Sensing 2022, 192,
99–114, doi:10.1016/j.isp sjp s.2022.08.009.
25. Wües , A.; Bou a d, D.; Guilla d, J.; Ibelings, B.W.; La anchy, S.; Pe ga, M.; Pasche, N.
LéXPLORE: A Floa ing Labo a o y on Lake Gene a O e ing Unique Lake Resea ch
Oppo uni ies. WIREs Wa e 2021, 8, doi:10.1002/wa 2.1544.
26. González Vilas, L.; B ando, V.E.; Concha, J.A.; Goyens, C.; Doglio i, A.I.; Doxa an, D.; Dille, A.;
Van de Zande, D. Valida ion o Sa elli e Wa e P oduc s Based on HYPERNETS in Si u Da a
Using a Ma ch-up Da abase (MDB) File S uc u e. F on ie s in Remo e Sensing 2024, 5,
doi:10.3389/ sen.2024.1330317.
27. Casey, K.A.; Rousseaux, C.S.; G egg, W.W.; Boss, E.; Chase, A.P.; C aig, S.E.; Mouw, C.B.;
Reynolds, R.A.; S amski, D.; Ackleson, S.G.; e al. A Global Compila ion o in Si u Aqua ic High
Spec al Resolu ion Inhe en and Appa en Op ical P ope y Da a o Remo e Sensing
Applica ions. Ea h Sys Sci Da a 2020, 12, 1123–1139, doi:10.5194/essd-12-1123-2020.
28. IMOS AODN Lucinda Je y Coas al Obse a o y: Hype OCR Da a A ailable online:
h ps:// h edds.aodn.o g.au/ h edds/ca alog/IMOS/SRS/OC/LJCO/ca alog.h ml (accessed on
7 Ma ch 2025).
29. Klo z, P. Assessmen o he Sui abili y o Plane Scope Supe Do e Image y o Aqua ic Remo e
Sensing a Lake Junin in Cen al Pe u, Ludwig-Maximilians-Uni e si ä München, 2024.
30. Zibo di, G.; Holben, B.N.; Talone, M.; D’Alimon e, D.; Slu ske , I.; Giles, D.M.; So okin, M.G.
Ad ances in he Ocean Colo Componen o he Ae osol Robo ic Ne wo k (AERONET-OC). J
A mos Ocean Technol 2021, 38, 725–746, doi:10.1175/JTECH-D-20-0085.1.
31. Ruddick, K.G.; B ando, V.E.; Co izzi, A.; Doglio i, A.I.; Doxa an, D.; Goyens, C.; Kuusk, J.;
Vanhellemon , Q.; Vans eenwegen, D.; Bialek, A.; e al. WATERHYPERNET: A P o o ype
Ne wo k o Au oma ed in Si u Measu emen s o Hype spec al Wa e Re lec ance o Sa elli e
Valida ion and Wa e Quali y Moni o ing. F on ie s in Remo e Sensing 2024, 5,
doi:10.3389/ sen.2024.1347520.
32. Vans eenwegen, D.; Ruddick, K.; Ca ijsse, A.; Vanhellemon , Q.; Beck, M. The Pan-and-Til
Hype spec al Radiome e Sys em (PANTHYR) o Au onomous Sa elli e Valida ion
Measu emen s—P o o ype Design and Tes ing. Remo e Sens (Basel) 2019, 11, 1360,
doi:10.3390/ s11111360.
33. Homme som, A.; K a ze , S.; Laanen, M.; Ansko, I.; Ligi, M.; B esciani, M.; Gia dino, C.; Bel án-
Abaunza, J.M.; Moo e, G.; We nand, M.; e al. In e compa ison in he Field be ween he New
WISP-3 and O he Radiome e s (T iOS Ramses, ASD FieldSpec, and TACCS). J Appl Remo e
Sens 2012, 6, 063615, doi:10.1117/1.JRS.6.063615.
34. Schläp e , D.; Bo el, C.C.; Kelle , J.; I en, K.I. A mosphe ic P eco ec ed Di e en ial Abso p ion
Technique o Re ie e Columna Wa e Vapo . Remo e Sens En i on 1998, 65, 353–366,
doi:10.1016/S0034-4257(98)00044-3.
35. Gege, P.; G ö sch, P. A Spec al Model o Co ec ing Sunglin and Skyglin . In P oceedings o
he Ocean Op ics XXIII 2016; 2016.
ACIX-III Aqua epo
85
36. Go don, H.R. Ocean Colo Remo e Sensing Sys ems: Radiome ic Requi emen s.; Sla e , P.N.,
Ed.; Oc obe 12 1988; p. 151.
37. Haan, J.F.; Kokke, J.M.M. Remo e Sensing Algo i hm De elopmen : Toolki I: Ope a ionaliza ion
o A mosphe ic Co ec ion Me hods o Tidal and Inland Wa e s. Ne he lands Remo e Sensing
Boa d (BCRS) 1996.
38. S e ckx, S.; Knaeps, S.; K a ze , S.; Ruddick, K. SIMila i y En i onmen Co ec ion (SIMEC)
Applied o MERIS Da a o e Inland and Coas al Wa e s. Remo e Sens En i on 2015, 157, 96–
110, doi:10.1016/j. se.2014.06.017.
39. Kisele , V.; Bulga elli, B.; Heege, T. Senso Independen Adjacency Co ec ion Algo i hm o
Coas al and Inland Wa e Sys ems. Remo e Sens En i on 2015, 157, 85–95,
doi:10.1016/j. se.2014.07.025.
40. Rich e , R.; Schläp e , D.; Mülle , A. An Au oma ic A mosphe ic Co ec ion Algo i hm o
Visible/NIR Image y. In J Remo e Sens 2006, 27, 2077–2085,
doi:10.1080/01431160500486690.
41. Wu, Y.; Knudby, A.; Pahle an, N.; Lapen, D.; Zeng, C. Senso -Gene ic Adjacency-E ec
Co ec ion o Remo e Sensing o Coas al and Inland Wa e s. Remo e Sens En i on 2024, 315,
114433, doi:10.1016/j. se.2024.114433.
42. Ko cheno a, S.Y.; Ve mo e, E.F.; Ma a ese, R.; Klemm, J ., F.J. Valida ion o a Vec o Ve sion o
he 6S Radia i e T ans e Code o A mosphe ic Co ec ion o Sa elli e Da a Pa I: Pa h
Radiance. Appl Op 2006, 45, 6762, doi:10.1364/AO.45.006762.
43. Lenoble, J.; He man, M.; Deuzé, J.L.; La ance, B.; San e , R.; Tan é, D. A Successi e O de o
Sca e ing Code o Sol ing he Vec o Equa ion o T ans e in he Ea h’s A mosphe e wi h
Ae osols. J Quan Spec osc Radia T ans 2007, 107, 479–507, doi:10.1016/j.jqs .2007.03.010.
44. Be k, A.; Ande son, G.P.; Acha ya, P.K.; Be ns ein, L.S.; Mu a o , L.; Lee, J.; ...; She le, E.P.
MODTRAN5: 2006 Upda e. In Algo i hms and Technologies o Mul ispec al, Hype spec al,
and Ul aspec al Image y XII; 2006; Vol. 6233, pp. 508–515.
45. De Keukelae e, L.; S e ckx, S.; Ad iaensen, S.; Knaeps, E.; Reusen, I.; Gia dino, C.; B esciani,
M.; Hun e , P.; Neil, C.; Van de Zande, D.; e al. A mosphe ic Co ec ion o Landsa -8/OLI and
Sen inel-2/MSI Da a Using ICOR Algo i hm: Valida ion o Coas al and Inland Wa e s. Eu J
Remo e Sens 2018, 51, 525–542, doi:10.1080/22797254.2018.1457937.
46. Pa k, Y.-J.; Ruddick, K. Model o Remo e-Sensing Re lec ance Including Bidi ec ional E ec s o
Case 1 and Case 2 Wa e s. Appl Op 2005, 44, 1236, doi:10.1364/AO.44.001236.
47. Gege, P. WASI-2D: A So wa e Tool o Regionally Op imized Analysis o Imaging Spec ome e
Da a om Deep and Shallow Wa e s. Compu Geosci 2014, 62, 208–215,
doi:10.1016/j.cageo.2013.07.022.
48. Vanhellemon , Q. Adap a ion o he Da k Spec um Fi ing A mosphe ic Co ec ion o Aqua ic
Applica ions o he Landsa and Sen inel-2 A chi es. Remo e Sens En i on 2019, 225, 175–192,
doi:10.1016/j. se.2019.03.010.

ACIX-III Aqua epo
86
49. Vanhellemon , Q.; Ruddick, K. A mosphe ic Co ec ion o Me e-Scale Op ical Sa elli e Da a o
Inland and Coas al Wa e Applica ions. Remo e Sens En i on 2018, 216, 586–597,
doi:10.1016/j. se.2018.07.015.
50. Ha mel, T.; Chami, M.; To mos, T.; Reynaud, N.; Danis, P.-A. Sunglin Co ec ion o he Mul i-
Spec al Ins umen (MSI)-SENTINEL-2 Image y o e Inland and Sea Wa e s om SWIR Bands.
Remo e Sens En i on 2018, 204, 308–321, doi:10.1016/j. se.2017.10.022.
51. S einme z, F.; Deschamps, P.-Y.; Ramon, D. A mosphe ic Co ec ion in P esence o Sun Glin :
Applica ion o MERIS. Op Exp ess 2011, 19, 9783, doi:10.1364/OE.19.009783.
52. S einme z, F.; Ramon, D. Sen inel-2 MSI and Sen inel-3 OLCI Consis en Ocean Colou
P oduc s Using POLYMER. In P oceedings o he Remo e Sensing o he Open and Coas al
Ocean and Inland Wa e s; F ouin, R.J., Mu akami, H., Eds.; SPIE, Oc obe 30 2018; p. 13.
53. de los Reyes, R.; Langhein ich, M.; Schwind, P.; Rich e , R.; P lug, B.; Bachmann, M.; Mülle , R.;
Ca mona, E.; Zekoll, V.; Reina z, P. PACO: Py hon-Based A mosphe ic Co ec ion. Senso s
2020, 20, 1428, doi:10.3390/s20051428.
54. IOCCG (2025) A ailable online: h ps://ioccg.o g/ esou ces/so wa e/ (accessed on 20 Oc obe
2025).
55. Vanhellemon , Q.; Ruddick, K. Tu bid Wakes Associa ed wi h O sho e Wind Tu bines Obse ed
wi h Landsa 8. Remo e Sens En i on 2014, 145, 105–115, doi:10.1016/j. se.2014.01.009.
56. Vanhellemon , Q.; Ruddick, K. Acoli e o Sen inel-2: Aqua ic Applica ions o MSI Image y. In
P oceedings o he 2016 ESA Li ing Plane Symposium; P ague, Czech Republic, 2016.
57. Vanhellemon , Q. Sensi i i y Analysis o he Da k Spec um Fi ing A mosphe ic Co ec ion o
Me e- and Decame e-Scale Sa elli e Image y Using Au onomous Hype spec al Radiome y.
Op Exp ess 2020, 28, 29948, doi:10.1364/OE.397456.
58. Vanhellemon , Q.; Ruddick, K. A mosphe ic Co ec ion o Sen inel-3/OLCI Da a o Mapping o
Suspended Pa icula e Ma e and Chlo ophyll-a Concen a ion in Belgian Tu bid Coas al
Wa e s. Remo e Sens En i on 2021, 256, 112284, doi:10.1016/j. se.2021.112284.
59. Chami, M.; La ance, B.; Fougnie, B.; Chowdha y, J.; Ha mel, T.; Waque , F. OSOAA: A Vec o
Radia i e T ans e Model o Coupled A mosphe e-Ocean Sys em o a Rough Sea Su ace
Applica ion o he Es ima es o he Di ec ional Va ia ions o he Wa e Lea ing Re lec ance o
Be e P ocess Mul i-Angula Sa elli e Senso s Da a o e he Ocean. Op Exp ess 2015, 23,
27829, doi:10.1364/OE.23.027829.
60. Hess, M.; Koepke, P.; Schul , I. Op ical P ope ies o Ae osols and Clouds: The So wa e
Package OPAC. Bull Am Me eo ol Soc 1998, 79, 831–844, doi:10.1175/1520-
0477(1998)079<0831:OPOAAC>2.0.CO;2.
61. Yang, P.; Feng, Q.; Hong, G.; Ka awa , G.W.; Wiscombe, W.J.; Mishchenko, M.I.; Dubo ik, O.;
Laszlo, I.; Sokolik, I.N. Modeling o he Sca e ing and Radia i e P ope ies o Nonsphe ical
Dus -like Ae osols. J Ae osol Sci 2007, 38, 995–1014, doi:10.1016/j.jae osci.2007.07.001.
62. Gas eige , J.; Wiegne , M. MOPSMAP 1.0: A Ve sa ile Tool o he Modeling o Ae osol Op ical
P ope ies. Geosci Model De 2018, 11, 2739–2762, doi:10.5194/gmd-11-2739-2018.
ACIX-III Aqua epo
87
63. Rich e , R. A Spa ially Adap i e Fas A mosphe ic Co ec ion Algo i hm. In J Remo e Sens
1996, 17, 1201–1214, doi:10.1080/01431169608949077.
64. Gege, P. The Wa e Colo Simula o WASI: An In eg a ing So wa e Tool o Analysis and
Simula ion o Op ical in Si u Spec a. Compu Geosci 2004, 30, 523–532,
doi:10.1016/j.cageo.2004.03.005.
65. Albe , A.; Mobley, C. An Analy ical Model o Subsu ace I adiance and Remo e Sensing
Re lec ance in Deep and Shallow Case-2 Wa e s. Op Exp ess 2003, 11, 2873,
doi:10.1364/OE.11.002873.
66. Mobley, C.D.; Gen ili, B.; Go don, H.R.; Jin, Z.; Ka awa , G.W.; Mo el, A.; Reine sman, P.;
S amnes, K.; S a n, R.H. Compa ison o Nume ical Models o Compu ing Unde wa e Ligh
Fields. Appl. Op . 1993, 32.
67. Gege Analy ic Model o he Di ec and Di use Componen s o Downwelling Spec al I adiance
in Wa e . Appl Op 2012, 51, 1407, doi:10.1364/AO.51.001407.
68. G egg, W.W.; Ca de , K.L. A Simple Spec al Sola I adiance Model o Cloudless Ma i ime
A mosphe es. Limnol Oceanog 1990, 35, 1657–1675, doi:10.4319/lo.1990.35.8.1657.
69. De oin‐Pla el, M.; Chami, M. How Ambiguous Is he In e se P oblem o Ocean Colo in Coas al
Wa e s? J Geophys Res Oceans 2007, 112, doi:10.1029/2006JC003847.
70. Wu, Y.; Knudby, A.; Lapen, D. Topog aphy-Adjus ed Mon e Ca lo Simula ion o he Adjacency
E ec in Remo e Sensing o Coas al and Inland Wa e s. J Quan Spec osc Radia T ans 2023,
303, 108589, doi:10.1016/j.jqs .2023.108589.
71. Global Modeling and Assimila ion O ice MERRA-2 Ta g1_2d_chm_Nx: 2d,1-Hou ly,Time-
A e aged,Single-Le el,Assimila ion,Ca bon Monoxide and Ozone Diagnos ics V5.12.4.
G eenbel , MD, USA, Godda d Ea h Sciences Da a and In o ma ion Se ices Cen e (GES DISC)
[Da ase ] 2015.
72. Ve mo e, E.F.; Tan e, D.; Deuze, J.L.; He man, M.; Mo ce e, J.-J. Second Simula ion o he
Sa elli e Signal in he Sola Spec um, 6S: An O e iew. IEEE T ansac ions on Geoscience and
Remo e Sensing 1997, 35, 675–686, doi:10.1109/36.581987.
73. Gia dino, C.; B esciani, M.; B aga, F.; Fabb e o, A.; Ghi a di, N.; Pepe, M.; Gianine o, M.;
Colombo, R.; Coglia i, S.; Gheb ehiwo , S.; e al. Fi s E alua ion o PRISMA Le el 1 Da a o
Wa e Applica ions. Senso s 2020, 20, 4553, doi:10.3390/s20164553.
74. Ni oumand-Jadidi, M.; Bo olo, F.; B uzzone, L. Wa e Quali y Re ie al om PRISMA
Hype spec al Images: Fi s Expe ience in a Tu bid Lake and Compa ison wi h Sen inel-2.
Remo e Sens (Basel) 2020, 12, 3984, doi:10.3390/ s12233984.
75. B esciani, M.; Gia dino, C.; Fabb e o, A.; Pelleg ino, A.; Mangano, S.; F ee, G.; Pina di, M.
Applica ion o New Hype spec al Senso s in he Remo e Sensing o Aqua ic Ecosys em
Heal h: Exploi ing PRISMA and DESIS o Fou I alian Lakes. Resou ces 2022, 11, 8,
doi:10.3390/ esou ces11020008.
76. Pelleg ino, A.; Fabb e o, A.; B esciani, M.; de Lima, T.M.A.; B aga, F.; Pahle an, N.; B ando,
V.E.; K a ze , S.; Gianine o, M.; Gia dino, C. Assessing he Accu acy o PRISMA S anda d
ACIX-III Aqua epo
88
Re lec ance P oduc s in Globally Dis ibu ed Aqua ic Si es. Remo e Sens (Basel) 2023, 15,
2163, doi:10.3390/ s15082163.
77. Fabb e o, A.; B esciani, M.; Pelleg ino, A.; Alikas, K.; Pina di, M.; Mangano, S.; Padula, R.;
Gia dino, C. T acking Wa e Quali y and Mac ophy e Changes in Lake T asimeno (I aly) om
Spacebo ne Hype spec al Image y. Remo e Sens (Basel) 2024, 16, 1704,
doi:10.3390/ s16101704.
78. Bailey, S.W.; We dell, P.J. A Mul i-Senso App oach o he on-O bi Valida ion o Ocean Colo
Sa elli e Da a P oduc s. Remo e Sens En i on 2006, 102, 12–23, doi:10.1016/j. se.2006.01.015.
79. Gia dino, C.; Pahle an, N.; Fabb e o, A.; Panizza, L.; Pelleg ino, A.; Vande meulen, R.;
Gianine o, M.; Ad iaensen, S.; Ag en, J.; Be ne , H.; e al. ACIX-III Aqua: E alua ion o
A mosphe ic Co ec ion o Hype spec al PRISMA Image y o e Inland and Coas al Wa e s. In
J Remo e Sens 2025, 1–25, doi:10.1080/01431161.2025.2574517.
80. Bi, S.; Hie onymi, M. Holis ic Op ical Wa e Type Classi ica ion o Ocean, Coas al, and Inland
Wa e s. Limnol Oceanog 2024, 69, 1547–1561, doi:10.1002/lno.12606.
81. Vande meulen, R.A.; Mannino, A.; C aig, S.E.; We dell, P.J. 150 Shades o G een: Using he Full
Spec um o Remo e Sensing Re lec ance o Elucida e Colo Shi s in he Ocean. Remo e Sens
En i on 2020, 247, 111900, doi:10.1016/j. se.2020.111900.
82. Balasub amanian, S. V.; Pahle an, N.; Smi h, B.; Binding, C.; Schalles, J.; Loisel, H.; Gu lin, D.;
G eb, S.; Alikas, K.; Randla, M.; e al. Robus Algo i hm o Es ima ing To al Suspended Solids
(TSS) in Inland and Nea sho e Coas al Wa e s. Remo e Sens En i on 2020, 246, 111768,
doi:10.1016/j. se.2020.111768.
83. B ache , A.; Rand ianalisoa, A.N.; Soppa, M.A.; Chab illa , S.; Sche le , D.; S einme z, F.;
Vanhellemon , Q.; Gege, P.; ...; Vans eenwegen, D. E alua ion o ENMAP Wa e Lea ing
Re lec ance P oduc s om Th ee A mosphe ic Co ec ion Me hods by Valida ion wi h
Hype spec al and Mul ispec al in Si u Measu emen s om Inland and Coas al Wa e a ound
he Wo ld. In P oceedings o he ESA Li ing Plan Symposium; Vienna, Aus ia, 2025.
84. Goyens, C.; La igne, H.; Dille, A.; Ve ae en, H. Using Hype spec al Remo e Sensing o
Moni o Wa e Quali y in D inking Wa e Rese oi s. Remo e Sens (Basel) 2022, 14, 5607,
doi:10.3390/ s14215607.
85. Pe e s S.; Laanen M.; G oe sch P.; Ghezehegn S.; Pose K.; Homme som A.; De Reus E.; Spaias
L. WISPs a ion: A New Au onomous abo e Wa e Radiome e Sys em. In P oceedings o he
Ocean Op ics XXIV Con e ence; Dub o nik, C oa ia, 2018; pp. 7–12.
86. B ando, V.; Keen, R.; Daniel, P.; Baumeis e , A.; Ne he y, M.; Baumeis e , H.; Hawdon, A.;
Swan, G.; Mi chell, R.; Campbell, S.; e al. The Lucinda Je y Coas al Obse a o y’s Role in
Sa elli e Ocean Colou Calib a ion and Valida ion o G ea Ba ie Ree Coas al Wa e s. In
P oceedings o he OCEANS’10 IEEE SYDNEY; IEEE, May 2010; pp. 1–8.
87. Tils one, G.; Dall’Olmo, G.; Hie onymi, M.; Ruddick, K.; Beck, M.; Ligi, M.; Cos a, M.;
D’Alimon e, D.; Vellucci, V.; Vans eenwegen, D.; e al. Field In e compa ison o Radiome e
Measu emen s o Ocean Colou Valida ion. Remo e Sens (Basel) 2020, 12, 1587,
doi:10.3390/ s12101587.
ACIX-III Aqua epo
89
88. Zibo di, G.; Voss, K.J.; Johnson, B.C.; Muelle , J.L. Ocean Op ics and Biogeochemis y P o ocols
o Sa elli e Ocean Colou Senso Valida ion, Volume 3.0: P o ocols o Sa elli e Ocean Colou
Da a Valida ion: In Si u Op ical Radiome y. In e na ional Ocean Colou Coo dina ing G oup
(IOCCG) Da mou h, NS, Canada 2019.
89. Lee, Z.P.; Du, K.; Voss, K.J.; Zibo di, G.; Lubac, B.; A none, R.; Weidemann, A. An Inhe en -
Op ical-P ope y-Cen e ed App oach o Co ec he Angula E ec s in Wa e -Lea ing Radiance.
Appl Op 2011, 50, 3155, doi:10.1364/AO.50.003155.
90. Thuillie , G.; He sé, M.; Labs, D.; Foujols, T.; Pee e mans, W.; Gillo ay, D.; Simon, P.C.; Mandel,
H. The Sola Spec al I adiance om 200 o 2400 Nm as Measu ed by he SOLSPEC
Spec ome e om he A las and Eu eca Missions. Sol Phys 2003, 214, 1–22,
doi:10.1023/A:1024048429145.
91. Sche le , D.; B ell, M.; Bohn, N.; Al a ado, L.; Soppa, M.A.; Segl, K.; B ache , A.; Chab illa , S.
EnPT – an Al e na i e P e-P ocessing Chain o Hype spec al EnMAP Da a. In P oceedings o
he IGARSS 2023 - 2023 IEEE In e na ional Geoscience and Remo e Sensing Symposium; IEEE,
July 16 2023; pp. 7416–7418.
92. Jakimow, B.; Janz, A.; Thiel, F.; Okujeni, A.; Hos e , P.; an de Linden, S. EnMAP-Box: Imaging
Spec oscopy in QGIS. So wa eX 2023, 23, 101507, doi:10.1016/j.so x.2023.101507.
93. Vanhellemon , Q. Daily Me e-Scale Mapping o Wa e Tu bidi y Using CubeSa Image y. Op
Exp ess 2019, 27, A1372, doi:10.1364/OE.27.0A1372.
94. EUMETSAT Recommenda ions o Sen inel-3 OLCI Ocean Colou P oduc Valida ions in
Compa ison wi h in Si u Measu emen s – Ma chup P o ocols. .
95. La igne, H.; Ruddick, K. In e -Band Calib a ion o Hype spec al Wa e Remo e Sensing:
Demons a ion o Ch is-P oba. In P oceedings o he 2021 IEEE In e na ional Geoscience and
Remo e Sensing Symposium IGARSS; IEEE, July 11 2021; pp. 7771–7774.
96. Doxa an, D.; ElKilani, B.; Co izzi, A.; Goyens, C. Valida ion o Sa elli e-De i ed Wa e -Lea ing
Re lec ance in Con as ed F ench Coas al Wa e s Based on HYPERNETS Field Measu emen s.
F on ie s in Remo e Sensing 2024, 4, doi:10.3389/ sen.2023.1290110.