Ge ing The Snow Communi y Ready o The NASA-ISRO SAR Mission
Shadi O eisgha an, NASA L-Band InSAR SWE Wo king G oup, NISAR P ojec Team
Je P opulsion Labo a o y
Cali o nia Ins i u e o Technology
2024 NASA Snow Communi y Mee ing
Boulde , CO
Augus 14, 2024
Copy igh 2023 Cali o nia Ins i u e o Technology. Go e nmen sponso ship acknowledged.
SWE Re ie al Using Ze o baseline epea -pass in e e ome y
•2017 Decadal Su ey calls o Snow Wa e Equi alen (SWE) measu emen wi h 3cm accu acy,
using ac i e senso ( ada ) as pa o he DS Ea h Explo e mission ca ego y.
•Leinss e al, TGRS 2015 (Fi s Expe imen al demo) : Towe based epea -pass in e e ome y
a X-band
•success ul wi h e y as (e e y 10 minu es) e isi ime
•Ruiz e al, TGRS 2022: Mul i channel (1-18 GHz) SodSAR owe ada in no he n Finland.
• L-, S- C and X-bands o e open and clea a ea.
•all possible pai s wi h 12 days empo al baseline (NISAR eplica)
•P o ides obus ness agains deco ela ion and can sol e o los phase cycles on high equency
bands.
Gune iussen e al., TGRS 2001
SWE om scale (mm)
SWE om scale (mm)SWE om scale (mm)
RMSE
(mm)
L
-
6.63
S
-
12.63
C
-
21.75
L+S
4.21
L+C
4.42
All
3.97
•SnowEX campaign was ocused on UAVSAR epea -pass in e e ome y.
•win e 2020: Biweekly sp ead o e di e en US si es (6 so ies o e 13 si es, ac oss 5 s a es)
•win e 2021: Weekly wi h ewe egional si es and mo e equen empo al sampling (10 so ies o e 6 si es, ac oss 4
s a es)
•Local expe ienced ield eams deployed on he da e o each o e pass (Sou ce o alida ion)
•Ai bo ne LIDAR o e subdomain a a subse si es (G ea sou ce o alida ion)
•coo dina ed wi h Sen inel-1 eam o 6 days e isi o some o he SnowEx si es
Repea -pass In e e ome y; SnowEx UAVSAR Time Se ies
•Eli Deeb, HP Ma shall, Jack Ta icone o ganized mon hly zoom
mee ings o InSAR SWE e ie al since 2021
(30+ people, 12 s uden s/ ecen g ads)
•In pe son mee ings a di e en con e ences (WSC, NISAR Wo kshop,
IGARSS)
•Some ecen InSAR SWE/dep h publica ions using SnowEx da a:
•Zacha y Hoppinen, Shadi O eisgha an, Hans-Pe e Ma shall, Ross Mowe , Kelly Elde ,
and Ca ie Vuyo ich, Snow wa e equi alen e ie al o e Idaho – Pa 2: Using L-
band UAVSAR epea -pass in e e ome y, The C yosphe e, 18, 575–
592, h ps://doi.o g/10.5194/ c-18-575-2024, 2024
•Shadi O eisgha an, Robe Zinke, Zacha y Hoppinen, and Hans Pe e Ma shall, Snow
wa e equi alen e ie al o e Idaho – Pa 1: Using Sen inel-1 epea -pass
in e e ome y, The C yosphe e, 18, 559–574, h ps://doi.o g/10.5194/ c-18-559-
2024, 2024
•Jack Ta icone, Ryan W. Webb, Hans-Pe e Ma shall, Anne W. Nolin, and F anz J.
Meye , Es ima ing snow accumula ion and abla ion wi h L-band in e e ome ic
syn he ic ape u e ada (InSAR), The C yosphe e, 17, 1997–
2019, h ps://doi.o g/10.5194/ c-17-1997-2023, 2023
•Palomaki, R. T. and Sp oles, E. A., Assessmen o l-band insa snow es ima ion
echniques o e a shallow, he e ogeneous p ai ie snowpack, Remo e Sensing o
En i onmen , 296:113744, 2023.
•Randall Bonnell, Daniel McG a h, Jack Ta icone, Hans-Pe e Ma shall, Ella Bump,
Ca oline Duncan, S ephanie Kamp , Yunling Lou, Alex Olsen-Miki owicz, Megan Sea s,
Kei h Williams, Lucas Zelle , and Yang Zheng, E alua ing L-band InSAR Snow Wa e
Equi alen Re ie als wi h Repea G ound-Pene a ing Rada and Te es ial Lida
Su eys in No he n Colo ado, EGUsphe e, h ps://doi.o g/10.5194/egusphe e-2024-
236, 2024
NASA L-Band InSAR SWE Wo king G oup
•Re ie ed snow dep h change using UAVSAR shows a e y good ag eemen wi h
LIDAR o e la 5km x 5 km egion on G and Mesa wi h d y snow
•Mo i a ed he ime se ies SnowEx campaign 2021
•Ai bo ne Lida ligh s aligned wi h UAVSAR o e ligh s, allowing di ec compa ison
o dep h change
•Ma shall, H.P., Deeb, E., Fo s e , R., Vuyo ich, C., Elde , K., Hiems a, C., and Lund, J.
(2021). L-band InSAR dep h e ie al du ing he NASA SnowEx 2020 Campaign:
G and Mesa, Colo ado. IGARSS, DOI:10.1109/IGARSS47720.2021.9553852
InSAR Wo king G oup Findings: UAVSAR o e G and Mesa
HP Ma shall e al, IGARSS 2021
TLS ΔSWE e ie als ( i s column) and UAVSAR ΔSWE e ie als
(second column) o e laid on a shaded elie as e , and pixel-wise
compa ison ( hi d column). Rows o ganized by da e/ ield si e.
•Righ igu e; UAVSAR s TLS (Te es ial Lida Scan)
•UAVSAR and TLS ΔSWE e ie als exhibi simila spa ial pa e ns and
ΔSWE magni udes
•Co ela ion: 0.72, RMSE = 19 mm
•Poo UAVSAR cohe ence o 3–23 Feb ua y 2021 – unw apping issues
•Below igu e; UAVSAR s GPR (G ound pene a ing Rada )
•ΔSWE e ie al compa ison: La ge sp ead, bu absolu e di e ence in
medians = 2 mm
•ΔSWE e ie al accu acy deg ades a lowe cohe ence
Figu e: (a) UAVSAR ΔSWE e ie als s. GPR ΔSWE e ie als. Box plo
dis ibu ions o UAVSAR and GPR ΔSWE e ie als o he (b) 2020 and (c) 2021
seasons.
InSAR Wo king G oup Findings: UAVSAR o e No he n Colo ado
Randall Bonnell, EGUsphe e 2024
(a) InSAR SWE be ween 19 and 26 Feb ua y agg ega ed o he 30 m Landsa
esolu ion. (b) The Landsa SCA be ween 18 Feb ua y and 5 Ma ch. The
colo scale o panel (a) was changed o −5 o 5 cm o exempli y he pa e ns.
InSAR Wo king G oup Findings: UAVSAR o e Jemez Moun ains, NM
Jack Ta icone e al, The C yosphe e 2023
•SWE loss and gain can be es ima ed
wi hin he same ada swa h
•Mo ning (~9 AM) acquisi ions held
cohe ence and p o ided easonable
ΔSWE esul s
•InSAR SWE loss isually simila o
Landsa SCA loss pa e ns
•UAVSAR e ie ed SWE s.
SNOWMODEL (RMSD = 0.023m,
= 0.60, n = 3e6)
•Bes pe o mance in high
ele a ion, d y snow egions
•UAVSAR e ie ed snow dep hs
s in si u (RMSE = 0.1m, = 0.80,
n = 64)
InSAR Wo king G oup Findings: UAVSAR o e Idaho
Zach Hoppinen e al, The C yosphe e 2024
•SWE = SWE a LIDAR da es
•Ve y good esemblance and
co ela ion be ween LIDAR
snow dep h and Sen inel SWE.
Si e LIDAR Da e Co ela ion
Basin Summi (BS)
02-19-2020 0.51
Basin Summi (BS)
03-15-2021 0.47
Mo es C eek (MC)
02-09-2020 0.66
Mo es C eek (MC)
03-15-2021 0.59
D y C eek (DC)
02-19-2020 0.56
•SWE = SWE o in si u s a ions
•9 s a ions, wi h o al SWE e o < 2 cm
o he en i e ime se ies
•15 s a ions, wi h o al SWE e o > 2 cm
bu simila pa e n
•SnowEx Coo dina ion
•Sen inel-1 6-days e isi o e Idaho
•Closes o NISAR ime se ies
•240km co e s 32 in si u s a ions
•Con inuous ime se ies o he en i e win e (12/1 o 3/31)
•The co ela ion be ween Sen inel-1 and In Si u SWE is 0.8
and he RMSE e o is 0.93cm.
InSAR Wo king G oup Findings: Sen inel-1 C-band Time Se ies o e Idaho
Shadi O eisgha an e al, The C yosphe e 2024
InSAR p oduc s om NISAR
P oduc G id Pos ing Spa ial scope
GSLC Geocoded 5m x (40/10/5/2.5)m
Ta ge dependen
GUNW Geocoded 80 m Global
•All InSAR p oduc s will be p oduced o nea es neighbo pai s
•12 days global in e e og ams
•Global spa ial scope o Geocoded Unw apped In e e og am
(GUNW)
•Co ec ion laye s in GUNW p oduc s
•Ionosphe ic phase sc een
•T oposphe ic delay
•Gene a ing p oduc wi h be e esolu ion (spa ial co ela ion o .
Snow: 50m) o longe ime sampling (24, 36, … ) needs ools o
snow communi y
Th ee sample GUNW p oduc s p oduced wi h NISAR p ocesso (ALOS-1 aw da a > RSLC > GUNW)
NISAR o Snow Communi y
●NISAR launch is close – global co e age e e y 12 days, 80m esolu ion
●I will be an inc edible ime se ies da a o e ie ing SWE
○UAVSAR: shown he capabili y o L-band InSAR o e ie e changes in SWE and dep h
○Sen inel-1 da a analysis ha e shown he capabili y o InSAR ime se ies o e ie e o al SWE
●Ad an ages
○Global da a a 80m esolu ion, e e y 12 days
○The la ge ame co e s la ge numbe o in si u s a ions and SWE spa ial a iabili y
○Tempo al cohe ence is highe a L-band compa ed o Sen inel-1 C-band
○Ambiguous SWE is highe a L-band compa ed o Sen inel-1 C-band
○I p opaga es h ough he ege a ion mo e a L-band compa ed o Sen inel-1 C-band
●Limi a ion
○C-band is mo e sensi i e o DSWE han L-band
○NISAR collec s InSAR da a e e y 12 days, compa ed o 6-days o Sen inel-1. I educes he empo al cohe ence.
●We need o de elop so wa e wo k lows o gene a e InSAR da a wi h desi ed spa ial esolu ion and ime sampling
○UAVSAR_py ools, an open sou ce lib a y led by s uden s (now pos docs) s a ed a SnowEx Hackweek
●LIDAR da a is he bes esou ce o quan i y he sou ces o e o
●Ta ge ed ield obse a ions o iden i y mel , empe a u e, in e al boa ds, densi y a e e y help ul o iden i y he
limi a ion o his me hod
●Huge oppo uni y o snow communi y – bu lo s o wo k o do o a global snow e ie al
NISAR is almos he e…
Fo mo e in o ma ion:
h ps://nisa .jpl.nasa.go
Back Up
L-band InSAR ΔSWE e ie als in shallow, a iable
p ai ie snow (Palomaki and Sp oles, 2023, RSE )
99% con .
95% con .
90% con .
90% con .
95% con .
99% con .
None
High a iabili y
Low a iabili y
good absolu e ag eemen
poo absolu e
ag eemen
Subg id spa ial a iabili y
wi hin InSAR pixels con ibu es
o poo absolu e ag eemen .
1
0
No malized LiDAR SWE
1
0
No malized InSAR SWE
0.57
-0.93
Di e ence (LiDAR –InSAR)
0.58
0
Local s anda d de ia ion
0
(a) (b)
(c) (d)
1
0
No malized LiDAR SWE
1
0
No malized InSAR SWE
0.57
-0.93
Di e ence (LiDAR –InSAR)
0.58
0
Local s anda d de ia ion
0
(a) (b)
(c) (d)
Lida and L-band InSAR show simila
spa ial pa e ns in shallow snow.
a e age snow dep h = 12 cm
Shown is abou 100 Tb o aw SAR
da a ha will be downlinked o e
each 12 day o bi cycle.
Many o NISAR’s science Cal/Val
si es a e join ly imaged by L-band
and S-band
Ta ge s include:
•India/ISRO (land, ice, ocean)
•Global dis ibu ion o Ecosys em
Cal/Val si es
•A c ic/An a c ic ice shee s
•De o ma ion–Wes e n U.S.
•Re lec o a ays –India, Alaska,
Cali o nia
Planned join L/S band obse a ions o one 12-day o bi cycle
NISAR Obse a ion plan
Dual-pol
40+5 MHz
Dual-pol
20+5 MHz
Dual-pol
40+5 MHz
Dual-pol
20+5 MHz
Ascending
Descending
●Cu en ly mos science
obse a ions o e he
globe a e domina ed by
20+5 MHz Dual-Pol H-
ansmi
●No h Ame ica is mos ly
co e ed by 40+5 MHz
Dual-Pol H- ansmi
acquisi ions
●The obse a ion plan is
di e se in space bu
consis en in ime ( ew
excep ions)
Geocoded Single Look Complex
om NISAR
Range
Azimu h
Eas ing
No hing
RSLC GSLC
●GSLC p oduc s a e Single Look Complex (SLC)
SAR image y geocoded on a map coo dina e
sys em
●The GSLC p oduc s a e p o ided on NISAR
ames and will be p oduced globally
●The g id o he p oduc s a e in UTM p ojec ion a
mid la i udes and in Pola s e eog aphic in pola
egions
●The g id o GSLC p oduc s om di e en ames
ha e an in ege spacing di e ence allowing o
mosaic he p oduc s wi hou in e pola ion.
Range bandwid h
No h spacing Eas spacing
5 MHz 5 m 40 m
20 MHz 5 m 10 m
40 MHz 5 m 5 m
80 MHz 5 m 2.5 m
●RSLC ep esen s simula ed NISAR
p oduc om UAVSAR
●GSLC is de i ed by geocoding he RSLC
•The RSLC p oduc s will be ~1.2x o e sampled in bo h ange and
azimu h di ec ions
•Azimu h spacing: 1520 Hz (equi alen o ~5 m pixel size)
•Slan ange spacing: mode dependen
•A Kaise (1.6) window o weigh he Range spec um
•The azimu h weigh ing is he an enna pa e n
Range
bandwid h
Azimu h
esolu ion
Slan ange
esolu ion
Azimu h
pos ing
Slan ange
pos ing
5 MHz ~6 m 30 m ~5 m 25 m
20 MHz ~6 m 7.5 m ~5 m 6.25 m
40 MHz ~6 m 3.75 m ~5 m 3.12 m
80 MHz ~6 m 1.95 ~5 m 1.56 m
RSLC esolu ion and pos ing
RSLC Speci ica ion
40 MHz
80 MHz
5
Range equency [MHz]
Range spec um o di e en ada modes
20 MHz 5
40 MHz
20 MHz
5
1257.5
The a ia ion o he wa eleng h (cen e
equency) ac oss all di e en L-band
modes is less han 3%.
NASA’s DAAC
(ASF DAAC o NISAR L1-
L2 p oduc s)
InSAR Scien i ic Compu ing En i onmen , Enhanced Edi ion
(ISCE3)
ISCE3 is he co e so wa e o p oduce he NASA’s SAR p oduc s
om NISAR (and Sen inel-1)
h ps://gi hub.com/isce- amewo k/isce3
ISCE3 HySDS Use s
