Mapping seasonal lood- ecession c opland ex en in he Senegal
Ri e Valley
Lau en B uckmann
a,b,*
, And ew Ogil ie
c
, Didie Ma in
c
, Finda Bayo Diakha ´
e
d
,
Amau y Tilman
a,b
a
Depa men o Ci il and Wa e Enginee ing, Uni e si ´
e La al, Quebec, Canada
b
Cen Eau, Wa e Resea ch Cen e , Quebec, Canada
c
G-EAU, Ag oPa isTech, BRGM, Ci ad, INRAE, Ins i u Ag o, IRD, Uni Mon pellie , Mon pellie , F ance
d
ISRA, BAME, Daka , Senegal
ABSTRACT
Flood- ecession ag icul u e (FRA) ep esen s a c ucial sou ce o li elihood o nume ous communi ies ac oss A ica who eside nea expansi e
loodplains and we lands. Howe e , i is cu en ly insu icien ly moni o ed. In his s udy, we p esen a me hodology o mapping FRA ha es ed
a eas in he Senegal Ri e Valley ha is bo h ep oducible and scalable. Ou me hodology en ails he in eg a ion o op ical and ada da a om
Sen inel pla o ms, conduc ed h ough a mul i empo al analysis wi h a seasonal ocus, and he applica ion o he Random Fo es algo i hm. The
esul s, suppo ed by a kappa coe icien o 91.9%, demons a e he i s comp ehensi e mapping o FRA in he Senegal Ri e alley, conduc ed
be ween 2019 and 2023. This mapping acili a es he iden i ica ion o he hyd ological ac o s ha in luence FRA ha es ing. The esul s o he
analyses ha e demons a ed he impo ance o in e annual a iabili y in he cul i a ed a eas o FRA, which ange om 14,000 o 75,000 ha
depending on he in ensi y o he annual lood. The du a ion and looded ex ension a e he p ima y ac o s ha egula e he c opping pa e n o FRA
o e he loodplain. The lood du a ion mus be a ound 35 days o pe mi he cul i a ion, wi h g ow h gene ally s a ing be ween 10 and 30
No embe . In conside a ion o hese indings, we ecommend ha u u e wa e managemen s a egies and u al de elopmen ini ia i es gi e due
conside a ion o FRA, o enhance he isibili y o a me s.
1. In oduc ion
A global analysis e eals ha 24% o c oplands a e si ua ed in loodplain a eas (D yden e al., 2021). Floodplains o la ge opical
i e s a e essen ial a eas o he p o ision o ecosys em se ices (De G oo e al., 2018) and o li elihoods (Asa e e al., 2022). In
A ica, lood-based a ming sys ems co e app oxima ely 25–30 million ha (Kool e al., 2018), and 3 million people depend di ec ly on
loodplains o ag icul u e (Rich e e al., 2010). One o he mos p e alen sys ems is loodplain ag icul u e (FRA), also known as
ecession ag icul u e. This sys em u ilizes he esidual mois u e om ex ended annual loods o cul i a e lood plains ollowing he
lood’s ecession (Molla d and Wal e , 2008). In A ica, his p ac ice is obse ed along majo i e s, including he Senegal Ri e . FRA
suppo s he p oduc ion o ice in Nige (Ba bie e al., 2011), maize and so ghum in Senegal (Saa nak, 2003; B uckmann, 2018), and
a ious o he c ops, including ui s and ege ables (Adamczewski e al., 2011). Fo example, lood- ecession so ghum co e s
app oxima ely 2 million ha in A ica (Kebe e al., 2001). FRA is pa o complex li elihood sys ems ha also include ishing and
g azing, enhancing ood p oduc ion and di e si y (Mo sumi e al., 2012). As FRA is p ima ily ha es ed du ing he d y season in
semi-a id en i onmen s, i signi ican ly con ibu es o ood secu i y, p o iding ei he household consump ion o cash income (Sidib´
e
e al., 2016; Balana e al., 2019). In he mid-Zambezi loodplain, FRA cons i u es 40% o he income o ag icul u al households
* Co esponding au ho . Depa men o Ci il and Wa e Enginee ing, Uni e si ´
e La al, Quebec, Canada
E-mail add ess: [email p o ec ed] (L. B uckmann).
Con en s lis s a ailable a ScienceDi ec
Remo e Sensing Applica ions: Socie y and
En i onmen
jou nal homepage: www.else ie .com/loca e/ sase
h ps://doi.o g/10.1016/j. sase.2025.101473
Recei ed 9 Oc obe 2024; Recei ed in e ised o m 10 Janua y 2025; Accep ed 23 Janua y 2025
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
A ailable online 30 Janua y 2025
2352-9385/© 2025 The Au ho (s). Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license
( h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/ ).
(Chimwe a e al., 2022). Howe e , lood-based a ming sys ems in A ica ha e been la gely o e looked by de elopmen policies, wi h
g ea e ocus placed on ull con ol i iga ion sys ems (Kool e al., 2018; S eenbe gen e al., 2011; Sidib´
e e al., 2016), e en hough
conside ing hem could o e oppo uni ies on he long- e m sus ainabili y o loodplains (Zenebe e al., 2022).
In he Senegal Ri e Valley (SRV), lood- ecession ag icul u e was he dominan ag icul u al sys em be o e i iga ion was in o-
duced in he 1960s, wi h cul i a ed a eas anging om 110,000 o 180,000 ha annually (C ousse e al., 1991). The es ablishmen o
i iga ion sys ems in he 1970s, suppo ed by he Manan ali and Diama dams in he 1980s, was aimed a mi iga ing he e ec s o
p olonged d ough s and egula ing i e lows. Howe e , hese dams ha e al e ed he hyd ological egime, educed lood olumes, and
inc eased i egula i y (Sall e al., 2020; B uckmann, 2021). Despi e his, FRA emains c ucial o local ood secu i y and complemen s
i iga ed c ops (B uckmann, 2018; Poussin e al., 2020). The ecen Mas e Plan (OMVS, 2022) has iden i ied a leas ou p oposed
dam de elopmen s ha gi e ise o conce ns abou he po en ial o looding o be los as a esul o ups eam discha ge con ol.
Moni o ing annually cul i a ed FRA su aces is essen ial o assess he impac s o u u e dams and op imize wa e alloca ion. While
i iga ion moni o ing is managed by Senegalese and Mau i anian na ional companies (SAED and SONADER), lood- ecession moni-
o ing is unde de eloped, pa ly due o challenges and lack o s akeholde in e es . Al hough he e ha e been emo e sensing ex-
pe imen s in he ea ly 2000s (Man´
e and F a al, 2001), and sho - e m esea ch p og ams like AGRICORA ha e explo ed
hyd o-ag ological unc ioning on speci ic basins (Poussin e al., 2020; Sall e al., 2020), a comp ehensi e, la ge-scale, in e annual
FRA moni o ing and mapping p og am has ye o be implemen ed. Mapping and moni o ing FRA can help balance ag icul u al and
wa e needs, and guide wa e managemen decisions on how o mee he con adic o y wa e demands o hyd oelec ic p oduc ion and
peak low suppo (F a al e al., 2002; Bade e al., 2003; Tilman e al., 2020). Accu a e da a can aid in modeling op imal ade-o s,
bene i ing local li elihoods and ecosys ems, and po en ially educing hyd oelec ic p oduc ion impac s om ups eam dams (Raso
e al., 2020). The esea ch on na u e-based solu ions demons a es he signi ican alue o loodplains o ecosys ems and ood secu i y,
emphasizing he impo ance o loodplain ag icul u e o li elihoods and sus ainable de elopmen goals (Singh e al., 2021).
Con e sely, la ge-scale i iga ion p og ams ha e seen limi ed success in he SRV (Poussin, 1998; Ga cia-Bola˜
nos e al., 2011),
p omp ing a ee alua ion o u al de elopmen s a egies in he egion.
Recen e o s ha e aimed a mapping po en ial FRA a eas, such as iden i ying lood-p one zones in E hiopia using mul isou ce
emo e sensing (Gumma e al., 2022) o applying a Bayesian app oach o es ima e FRA p obabili y in Kenya and E hiopia (Liman
Ha ou e al., 2020). While hese s udies p o ide aluable insigh s, hey don’ cap u e ac ual FRA cul i a ion. O he esea ch has
ocused on mapping looded a eas in Wes A ican loodplains, including Senegal, as he ounda ional basis o suppo ing FRA using
da ase s like MODIS, Landsa , and Sen inel (Ogil ie e al., 2015; Ogil ie e al., 2020; B uckmann e al., 2022;Ogil ie e al., 2025 ).
Howe e , mapping a single ype o ag icul u e in a he e ogeneous ag icul u al landscape poses signi ican challenges. In he Senegal
Ri e basin, FRA occu s du ing he season ollowing bo h he ainy and lood seasons. The landscape ea u es a mix o i iga ion,
ain ed c ops, o es , and g ass, which can o e lap wi h FRA be ween la e Oc obe and ea ly Decembe . Fu he mo e, FRA c ops a e
sown a di e en imes along he loodplain as he ag icul u al calenda is de ined by he local loodwa e dynamics. E ec i e FRA
mapping mus p io i ize a me hodology ha combines lood and ag icul u al mapping, g ounded in a deep unde s anding o he
hyd o-ag oecological p ocesses ha d i e i . The in ica e wa e - ege a ion dynamics, also seen in a eas like i iga ed pe ime e s,
equi e sophis ica ed app oaches. Va ious me hods ha e been applied o map seasonal ege a ion o ag icul u al land use in such
complex en i onmen s. Fo ins ance, ice has been a ocal poin due o i s o e lap wi h o he c ops du ing he g owing s age. Mapping
e o s should conside using speci ic spec al bands like he ed edge (Jiang e al., 2021) o a mul i empo al app oach (Gumma e al.,
2014; Dong e al., 2015). In eg a ing da a om mul iple sou ces can enhance he accu acy o ice ield de ec ion (Zhao e al., 2021).
Tempo al analysis is also aluable o examining lood pulse dynamics and hei e ec s on landscape and ege a ion (Hu e al., 2015;
Kool e al., 2022). Machine lea ning is inc easingly u ilized o iden i y and map speci ic land uses ac oss he globe, pa icula ly in
di e se ag icul u al landscapes. The choice o classi ica ion me hodology is c i ical, and nume ous s udies ha e sough o de e mine
he mos e ec i e app oaches in a ious con ex s. Gxokwe e al. (2020) e iewed me hods o semi-a id we lands mapping using
Sen inel-2 and ound ha Random Fo es (RF) and a i icial neu al ne wo ks, combined wi h classi ica ion and eg ession ee (CART),
deli e ed he bes esul s. This aligns wi h he indings by Thanh Noi and Kappas (2017), who compa ed non-pa ame ic classi ie s like
RF, k-Nea es Neighbo (kNN), and Suppo Vec o Machine (SVM) o land co e mapping in he Red Ri e Del a. Ano he app oach o
map seasonal ege a ion in ol es using mul i empo al image se ies o aid classi ica ion. In he Cambodian Mekong loodplains,
O ieschnig e al. (2021) iden i ied RF and G adien Boos ing T ees (GTB) as he mos accu a e classi ie s. Inco po a ing in a-seasonal
da a, such as phenology, can u he imp o e we land classi ica ion p ecision (Tian e al., 2016). One limi a ion o using machine
lea ning and mul i empo al da a o e la ge loodplains is he compu a ional demand o analyzing nume ous images. Cloud compu ing
pla o ms, like Google Ea h Engine (GEE), can o e come hese challenges by o e ing as s o age and compu a ional powe . GEE is
widely used oday, wi h many case s udies highligh ing i s e ec i eness in applica ions, such as la ge-scale we land mapping (Wang
e al., 2020; Gxokwe e al., 2022).
This s udy aims o de elop an app oach capable o mapping seasonal lood- ecession ag icul u e in he Senegal Ri e Valley o
moni o he empo al ends o his i al ag icul u al li elihood. The main con ibu ions o his wo k a e i s he de elopmen o an
adap able wo k low o FRA mapping ha combines he bene i s o using well-known machine lea ning algo i hms and a mul i-
empo al app oach inside GEE cloud-compu ing pla o m. The wo k low is designed o be easily eplica ed and adap ed in o he lood-
ecession ag icul u al egions. Fu he mo e, his app oach u ilizes open-access da a om Sen inel-1 and Sen inel-2 sa elli es, he eby
ensu ing ha he me hodology is no only accessible bu also cos -e ec i e o esea che s and p ac i ione s wo ldwide. Secondly, we
u ilize spa ial analysis and ex ensi e locally collec ed g ound u h da a o p o ide a de ailed desc ip ion o he spa ial dis ibu ion and
ex en o FRA a eas in he Senegal loodplain du ing ecen yea s, om 2019 o 2023. Fu he mo e, we analyze he spa ial d i e s
condi ioning he implemen a ion o FRA in SRV wi h op ical da a, including he cha ac e is ics o looding in space (ex en ) and ime
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
2
Fig. 1. – Map o he Senegal Ri e Basin and Valley. Whi e boxes a e he ou ocus a eas p esen ed below; Black ci cles a e he main ci ies o he
alley; Da k blue is pe manen wa e s de ined using Sen inel-2 and MNDWI. Backg ound om Google Sa elli e.
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
3
(s a , end, du a ion), o assess he h eshold ha ini ia es he cul i a ion o FRA a eas. The s udy aims o analyze spa ial pa e ns and
empo al changes in FRA, p o iding scien i ic e idence o suppo he sus ainable de elopmen o loodplains and add ess he wa e -
ene gy- ood-ecosys em nexus con lic s in he egion.
2. S udy a ea: he Senegal i e alley
The Senegal Ri e basin is a la ge ansbounda y wa e shed (337,000 km
2
) sha ed by ou coun ies: Guinea, Mali, Mau i ania, and
Senegal. In he downs eam pa , he Senegal Ri e Valley is a la ge loodplain o 10–20 km wide and 790 km long. Topog aphy is la ,
wi h an a e age decline o 1–3 cm pe km (Bade e al., 2014). The alley wa e esou ce is allogenic due o i s loca ion downs eam o
he h ee p ima y ibu a ies: he Fal´
em´
e, Ba ing, and Bakoye. The Senegal loodplain is delimi ed be ween he wo owns o Bakel
ups eam and Dagana downs eam (Fig. 1). The annual lood is ed by ain all in he uppe basin. In he Valley, ain all is modes and
a ies om 250 mm in he no h o 600 mm in he sou h. The lood occu s be ween he mon hs o Augus and No embe , and he
iming o he peak depends on he loca ion along he i e as well as he ampli ude o he lood, wi h la ge loods p opaga ing less as
downs eam. FRA is p ac iced in he loodplain a e he lood, wi h a me s sowing seeds p og essi ely as he lood ecedes. C ops a e
sown be ween Sep embe and No embe depending on he da e he wa e e ea s om land in he loodplain. Fa me s g ow mos ly
so ghum, maize, cowpeas, and swee po a oes. The ha es happens be ween Feb ua y and Ma ch, depending on he yea and he
dis ance downs eam. In he SRV, h ee ypes o c op a ming a e dis inguished acco ding o he o igin and con ol o he wa e
esou ce: ain- ed c ops, i iga ed c ops unde o al con ol, and lood- ecession c ops. I iga ion is p ac iced h oughou he Senegal
Fig. 2. – Poin s o calib a ion and alida ion o e he 4 ocus a eas.
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
4
Ri e Valley, bu mainly in he downs eam pa . Rain ed a ming is ound mos ly in he sou he n pa o he alley due o he minimal
annual ain all in he no he n, Sahelian pa o he wa e shed. The loodplain has a su ace a ea o app oxima ely 400,000 ha (4000
km
2
) o which 200,000 ha a e in ended o i iga ion (OMVS, 2018). To enhance unde s anding o FRA, he pape ocuses on ou
dis inc zones wi hin he loodplain (Fig. 1). Zone 1 is si ua ed in he icini y o Podo , he no he n ci y in Senegal (Fig. 2a). I en-
compasses a FRA a ea si ua ed o he wes o he u ban a ea. This is he a ea examined by Ogil ie e al. (2020) and Poussin e al. (2020).
The second zone is si ua ed no h o Ndioum ci y (Fig. 2b), and zones 3 and 4 a e loca ed in he cen al alley espec i ely a ound he
illages o Mboumba (Fig. 2c) and Diaba Dekele (Fig. 2d).
3. Da a and me hods
The FRA mapping me hodology comp ises ou s eps. The ini ial s ep is o collec g ound u h poin s o di e en land uses.
Spec al band a io indices om op ical and ada image y a e employed in s ep wo o comp ehend he beha io o lood ecession
c ops on an in a-seasonal scale and o de ine sub-seasons speci ic o lood ecession ag icul u e. In s ep h ee, a syn he ic composi e
image o he indices is c ea ed in o de o ain a machine-lea ning algo i hm. Finally, he accu acy o he esul s is e alua ed p io o
ex apola ion o subsequen yea s.
3.1. Da a
3.1.1. Calib a ion and alida ion da a
One c ucial s ep is he es ablishmen o a da abase o egions o in e es , which is u ilized by he algo i hm o ain and alida e he
classi ica ion. Da a we e collec ed om high- esolu ion Google Ea h images in 2019, mainly in Decembe . We ga he ed 2951 poin s
ac oss 7 land uses (Table 1 & Fig. 2), spli 70/30 o calib a ion and alida ion. Addi ionally, 100 g ound- u h poin s om lood-
ecession plo s we e included om a Decembe 2022 ield ip along he SRV loodplain be ween Bakel and Podo . Poin s we e ac-
qui ed using a handheld GPS de ice wi hin plo s whe e ecession c ops we e g own in 2022.
3.1.2. Sa elli e images and cloud-compu ing
Da a we e analyzed using GEE due o he ex ensi e a ea unde s udy and he nume ous images a ailable. We u ilized Sen inel-2 MSI
images accessible ia GEE, employing he le el 2 su ace e lec ance p oduc , which became a ailable o ou a ea on Decembe 16,
2018. Sen inel-2 images a e p ocessed wi h cloud masking by using he CLOUDY_PIXEL_PERCENTAGE p ope y o e ain images wi h
less han 40% cloudy pixels and he cloud mask p obabili y p oduc . The COPERNICUS/S2_CLOUD_PROBABILITY Image Collec ion is
de i ed om Sen inel-2 da a and p o ides in o ma ion abou he likelihood o cloud p esence o each pixel in an image. Images wi h
mo e han 50% o cloud p obabili y we e emo ed om ou image collec ion. Addi ionally, we employed he Sen inel-1 G ound Range
De ec ed images. The in eg a ion o ada da a and op ical image y enables he di e en ia ion o ege a ion, pa icula ly o es and
ee co e , du ing lood pe iods, he eby enhancing he accu acy o classi ica ion (S einhausen e al., 2018; Slag e e al., 2020). All
Sen inel-1 GRD in GEE ha e unde gone p ep ocessing, which in ol es he s eps implemen ed by he Sen inel-1 Toolbox o de i e he
backsca e coe icien in each pixel. We c ea ed a homogeneous subse by il e ing using some me ada a p ope ies: all images a e IW
(In e e ome ic Wide Swa h) and om an ascending o bi .
While longe da ase s like Landsa ETM+and OLI o MODIS p o ide b oade co e age, Sen inel images a e p e e able due o hei
ine spa ial esolu ion. This esolu ion is c ucial o accu a ely mapping small lood- ecession cul i a ed a eas in luenced by hy-
d ological, pedological, and mic o- opog aphic condi ions. Addi ionally, Sen inel’s equen e isi s allow o in a-seasonal analysis,
which helps in dis inguishing and classi ying land co e based on i s spec al signa u e h oughou he season. Bo h Sen inel-2 and
Sen inel-1 ha e high e isi equencies (up o 5 days o S2 and 6 days o S1) and high esolu ion om 10 o 20 m. The image
collec ion used is composed o images om 7 iles o Sen inel-2 which a e hal -sha ed be ween S2A and S2B. Fo Sen inel-1 all he
images a e om S1A. Fo he calib a ion yea 2019, we used 568 Sen inel-2 and 120 Sen inel-1 images be ween July and Feb ua y.
De ails abou images used o each yea ly analysis can be ound in Appendix 1.
Table 1
– Da ase used o calib a ion and alida ion.
Land use Numbe o poin s
Flood- ecession ag icul u e 674
Flood- ecession basin uncul i a ed 400
I iga ed pe ime e s 537
Ba e soil 462
Fo es 531
O he na u al ege a ion 160
Wa e 187
To al 2951
Calib a ion 2066
Valida ion 885
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
5
3.1.3. Op ical and ada indexes
Se e al op ical indices we e es ed o classi ica ion based on hei empo al and spec al signa u es. These indices we e selec ed o
hei abili y o de ec wa e , mois u e, ege a ion, and ba e soil (Table 2). MNDWI, a e inemen o he No malized Wa e Di e ence
Index (NDWI), enhances wa e de ec ion while minimizing he in luence o buil -up a eas and soil, which is c ucial in Sahelian zones
whe e wa e is shallow (Campos e al., 2012). S udies ha e e ec i ely used MNDWI o delinea e looded a eas in he SRV (Ogil ie
e al., 2020; Ogil ie e al., 2025). The No malized Di e ence Mois u e Index (NDMI) moni o s soil and ege a ion mois u e, wi h
highe alues indica ing heal hy ege a ion wi h g ea e wa e con en . Th ee complemen a y ege a ion indices a e also used: he
widely used No malized Di e ence Vege a ion Index (NDVI) and he Modi ied Soil Adjus ed Vege a ion Index (MSAVI). MSAVI, which
inco po a es a soil adjus men ac o , is ad an ageous in egions such as he SRV, whe e i p o ides a mo e accu a e assessmen o
spa sely ege a ed pixels. The No malized Di e ence Red Edge Index (NDRE) is sensi i e o chlo ophyll and accoun s o di e en
ege a ion g ow h s ages, educing con usion in ege a ion mapping (Jiang e al., 2021). Finally, he Ba e Soil Index (BSI) is used o
de ec a ia ions in land co e .
Table 2
– Indexes used in he s udy.
Indice Fo mula wi h Sen inel bands Pixel
esolu ion
Fullname Sou ce
MNDWI (B3-B11)/(B3+B11) 20 Modi ied No malized Di e ence Wa e
Index
Xu, n.d.
NDMI (B8-B11)/(B8+B11) 20 No malized Di e ence Mois u e Index Gao (1996)
BSI ((B11+B4)-(B8+B2))/((B11+B4)+(B8+B2)) 20 Ba e Soil Index Rikima u e al.
(2002)
NDVI (B8-B4)/(B8-B4) 10 No malized Di e ence Vege a ion Index Rouse e al. (1973)
MSAVI (2 * NIR +1 - sq (pow((2 * NIR +1), 2) - 8 * (NIR -
RED)))/2
10 Modi ied Soil Adjus ed Vege a ion Index Qi e al. (1994)
NDRE (B8-B5)/(B8+B5) 10 No malized Di e ence Red Edge Ba nes e al. (2000)
VV/VH VV/VH 10 Pola iza ion bands a io Slag e e al., 2020
VV +VH VV +VH 10 Pola iza ion bands sum
VV-VH VV-VH 10 Pola iza ion bands di e ence
Fig. 3. – Tempo al beha io o i e di e en land uses on di e en op ical and ada indices used in he s udy. (Numbe o sampling poin s used pe
land use: FRA =675, PI =542, Fo es =619, Ba e soil =675, Uncul i a ed looded a eas =407).
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
6
Fo ada , we use indices de i ed om a i hme ic ope a ions on he e ical ansmi , ho izon al ecei e (VH), and e ical
ansmi , e ical ecei e (VV) channels. Co- and c oss-pola ized backsca e in ensi ies (VV and VH) a e use ul o iden i ying c op
ypes, especially ice (Nguyen e al., 2016; Mandal e al., 2018). SAR da a a e also commonly used o o es de ec ion due o hei
s abili y wi hin a single season. The indices used include he VV/VH a io, VV-VH di e ence, and VV +VH addi ion, which enhance
ege a ion de ec ion and o es a ea disc imina ion by accoun ing o soil s uc u e and ex u e h oughou he season. P e ious s udies
ha e shown ha combining SAR and op ical da a imp o es he accu acy o LULC mapping in we lands (Slag e e al., 2020), as hese
da a sou ces complemen each o he (Veloso e al., 2017; O ieschnig e al., 2021). Inclusion o he VV/VH di e ence o a io can also
imp o e accu acy (Abdikan e al., 2016).
3.2. Me hod
3.2.1. De ini ion o sub-seasons by using empo al se ies o indexes
To de ine and map seasonal c ops whose ex en is pa ly de e mined by he annual lood, we ocused on he in a-seasonal a i-
a ions in he spec al signa u es o a ious elemen s (wa e , humidi y, ege a ion, ba e soil). By analyzing he sub-seasonal empo al
(Fig. 3) and spa ial (Fig. 4) a ia ions o hese spec al signals, we aimed o enhance he de ec ion o FRA.
The mul i- empo al MNDWI cu e o lood- ecession ag icul u e (FRA) a eas in 2019 shows dis inc pa e ns: MNDWI alues a e
posi i e be ween Sep embe and la e Oc obe du ing looding bu emain a ound −0.4 o he wise. As expec ed, MNDWI emains
nega i e in ba e soils, while in looded o es s and i iga ed zones MNDWI alues can empo a ily exceed ze o bu emain lowe han in
looded a eas. NDMI ollows a simila end, inc easing du ing looding and dec easing a e wa ds. In i iga ed a eas, he NDMI e-
mains high o much longe han in FRA a eas. The Ba e Soil Index (BSI) is use ul as lood ecession a eas a e ba e ou side he c opping
Fig. 4. – A e age seasonal alue o he a ious indexes used in Zone 2 – Ndioum om July 2019 o Feb ua y 2020.
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
7
pe iods, wi h BSI a ound 0.2 when ba e and d opping nega i e when c opped. The NDVI p o ile is low be o e he lood, nega i e
du ing he lood, and inc eases du ing c op g ow h o 0.2–0.4 in Decembe and Janua y. In con as , o he ege a ed a eas main ain
highe NDVI alues h oughou he season. MSAVI and NDRE mi o NDVI pa e ns, wi h i iga ion esul ing in highe alues o he
ege a ion indices. Rada indices a e e ec i e in de ec ing looded a eas, as wa e causes signi ican signal changes, in con as o he
s able signal om o es s. Fo ins ance, he VV/VH a io inc eases in looded a eas due o s ong VV backsca e and weak VH
backsca e .
Based on isual analysis, he 2019 looded ag icul u al season can be di ided in o ou sub-pe iods (Fig. 5): (1) P e- lood (July o
mid-Augus ), wi h ba e soil and no wa e o ege a ion; (2) Floodi (mid-Augus o la e Oc obe ), ea u ing ex ensi e wa e co e age;
(3) Pos - lood (No embe ), wi h eceding wa e and ba e soil; (4) pos - lood ege a i e pe iod (Decembe o Feb ua y), when c ops
and na u al ege a ion each peak g ow h. This di ision helps in iden i ying land-use pa e ns and dis inguishing FRAs om o he
a eas. Indeed, he sha p and sho e peak obse ed in he wa e indices dis inguishes lood ecession a eas om i iga ed a eas, o es s,
and ba e soils, while he peak in he ege a ion indices in Janua y–Feb ua y helps o dis inguish FRA om uncul i a ed looded a eas.
In o he yea s, hese sub-pe iods a e adjus ed o align wi h he lood pe iod, using a consis en isual app oach o de ine he empo al
bounda ies. The map o a e age index alues o he di e en sub-seasons is p esen ed in Fig. 6, which ocuses on a speci ic a ea wi hin
he SRV. I can be obse ed ha some indexes, such as MNDWI, NDMI, o ege a ion indexes, exhibi signi ican changes be ween sub-
seasons. In con as , he p e- lood pe iod demons a es minimal spa ial a ia ions compa ed o he o he seasons.
3.2.2. Algo i hm p ocessing and pos p ocessing
Be o e classi ica ion, a single image is gene a ed o cap u e he spec al cha ac e is ics o he ou sub-seasons. Six educ-
e s—minimum, maximum, mean, s anda d de ia ion, skew, and ku osis—a e used o c ea e one image pe each sub-season, e lec ing
pixel dis ibu ions and cen al alues. This p ocess esul s in 36 images pe sub-season om nine indices ( h ee ada , six op ical),
o aling a 216-band image o each yea . Subsequen ly, ou machine lea ning algo i hms in eg a ed in o GEE we e es ed o iden i y
he mos obus one. The algo i hms we e Random Fo es (RF), G adien T ee Boos (GTB), Classi ica ion and Reg ession T ee (CART),
and k-Nea es Neighbo (KNN). The RF algo i hm p oduced he mos accu a e esul s, al hough he di e ence in accu acy be ween i
and GTB was minimal (Table 3). These indings a e consis en wi h hose o p e ious s udies conduc ed by Gxokwe e al. (2020),
(Thanh Noi and Kappas, 2017), and O ieschnig e al. (2021) on loodplains in A ica and Asia. Addi ionally, hese indings a e
consis en wi h hose o o he s udies conduc ed in semi-a id en i onmen s (Zhao e al., 2024) and in c op classi ica ion (Gup a e al.,
2024). The Random Fo es algo i hm was he e o e selec ed based on i s pe o mance in e ms o accu acy sco e and i s equency as an
op imal classi ica ion model in he exis ing li e a u e, hus ensu ing he desi ed le el o ep oducibili y. The Random Fo es algo i hm
is hen applied o he composi e image. A emp ing he same me hod o a single season om Augus o Ma ch esul ed in lowe
Fig. 5. – A e age empo al beha io o di e en op ical indexes o e lood- ecession ag icul u e a eas in 2019/2020 (n =675).
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
8
Fig. 6. A e age alue o each sub-season o he a ious indices used in Zone 2 - Ndioum om July 2019 o Feb ua y 2020.
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
9
5. Discussion
5.1. Wa e go e nance implica ions
The esul s on FRA cul i a ed a eas p esen ed he e a e he i s obse a ions since s udies comple ed in he 1980s and ea ly 2000s
unde he P og amme d’Op imisa ion de Ges ion des R´
ese oi s (POGR). This p og am es ima ed, based on pa ial na ional s a is ics,
ha he a e age annual a ea o cul i a ed FRA be ween 1946 and 1999 was 52,000 ha (OMVS & IRD, 2002). This a ea a ied acco ding
o he hyd ological pe iods, eaching 110,000 ha (1100 km
2
) o e 1946–1970 du ing high- lood yea s and educing o 46,000 ha (460
km
2
) o e 1970–1999 du ing he subsequen d ough pe iod when lood ampli udes educed subs an ially (B uckmann e al., 2022).
Based on his da a, Bade e al. (2003, 2015) p oposed a model o es ima e FRA a eas based on discha ge da a du ing Sep embe a
Bakel gauging s a ion. This model has signi ican ly in luenced wa e alloca ion policies in he basin. Ogil ie e al. (2025) ecen ly
upda ed his ela ionship based on hyd ological eg essions and ea h obse a ions o he looded a eas in he SRV (Fig. 14). Ou
compa ison o EO-based FRA es ima es wi h his modeling app oach e eals ha ou es ima es a e gene ally close o hose de i ed
om discha ge da a. The mapping p oposed in his wo k has he po en ial o enhance comp ehension o he annual and seasonal
ag icul u al dynamics o he Senegal Ri e alley.
Mo eo e , i can op imize he wa e esou ce managemen a he basin le el, as demons a ed by illus a ing he impo ance o
lood cha ac e is ics, such as du a ion and ex en . Indeed, a he ime o he Manan ali dam cons uc ion, ope a ional ules had se a
a ge o implemen an a i icial lood o main ain 55,000 ha (550 km2) o FRA a eas each yea o balance lood suppo wi h ene gy
p oduc ion. This a i icial lood has a ely been implemen ed, and since 2003 he Manan ali dam has ceased o suppo looding, wi h
OMVS elying on un egula ed ibu a ies o mee his objec i e. Ou esul s sugges ha his app oach may be insu icien , as FRA a eas
we e less han 22,000 ha (220 km
2
) in 2021 and 2023. The 55,000 ha a ge was es ablished in he 1980s o add ess d ough -induced
educ ions in FRA po en ial while a oiding nega i e impac s on ene gy p oduc ion. The a ge is he e o e inapp op ia e o he
cu en hyd ological pe iod and is now pu sued wi h g ea u gency in all o icial documen s, including he In eg a ed Wa e Man-
agemen P ojec (2016) and he la es Mas e Plan (OMVS, 2022). Ne e heless, ou ecen esul s show ha his a ge has been
exceeded in h ee o he las i e yea s, wi h a no able 36% inc ease in 2022, and 12 imes since he ea ly 2000s (Fig. 14), highligh ing
he po en ial and local in e es in inc easing FRA.
This unde sco es he need o imp o ed conside a ion o FRA a eas in u u e wa e managemen , especially wi h he an icipa ed
inc ease in dam cons uc ion, which may educe lood olumes. Cu en wa e managemen and ag icul u al policies in he Senegal
Fig. 14. – Annual FRA a eas in SRV om 1950 o 2022 based on p e ious modelling by IRD compa ed o ou s udy and a e age discha ge a Bakel
om Augus o Oc obe . Da a: Ogil ie e al. (2025).
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
16
Ri e Basin ha e p ima ily add essed d ough , e en as p ecipi a ion le els ha e isen since he la e 1990s (Bodian e al., 2020).
Upda ing hese policies can be e accommoda e sec o s like lood-based ag icul u e and ishing is essen ial. Ou esul s emphasize he
signi ican impac o wa e a ailabili y on FRA pa e ns. The u u e hyd ological egime in he Senegal Ri e Valley will be c i ical o
FRA dynamics (Ogil ie e al., 2025; B uckmann, 2018). Howe e , he e is a con lic be ween he wa e needs o FRA and he
inc easing wa e demands due o ene gy p oduc ion and d y season i iga ion. When lood condi ions a e a o able, FRA can co e
o e 75,000 ha (750 km2), which is conside able compa ed o he 130,000 ha (1330 km2) used annually o i iga ion (PARACI da a,
OMVS, 2018). Finally, ou s udy ocuses on cul i a ed a eas, bu p e ious esea ch highligh ed he mul i unc ionali y o looded a eas,
including hei c ucial ole in ishing and g azing (Ho owi z e al., 1990; B uckmann, 2018). Add essing ade-o s wi hin he WEFE
nexus mus accoun o his complexi y.
5.2. Assessing socio-economic and ood secu i y impac s in ag icul u e: he need o ield-based da a
EO can signi ican ly acili a e/enhance he assessmen o la ge-scale cul i a ed a eas, suppo ing in o med decision-making.
Howe e , a de ailed unde s anding o c op p ac ices ( ype, dis ibu ion, and yield) is u gen ly needed. In he Senegal Ri e Valley,
p ima y c ops include so ghum, cowpeas, and maize, bu hei dis ibu ion a ies ac oss he alley. In he uppe egions, maize is he
p edominan c op, while downs eam a eas p ima ily cul i a e so ghum, o en alongside cowpeas. Yield da a o hese c ops a e
spa sely documen ed. They can a y widely based on se e al ac o s, including he du a ion and iming o looding, soil wa e ese es,
pes p esence, sowing pe iod, weed p e alence, and a ailable labo . P e ious s udies ha e epo ed no able disc epancies in yields. Fo
ins ance, so ghum yields ha e been epo ed as low as 178 kg/ha (Saa nak, 2003) and as high as 280–330 kg/ha (B uckmann, 2018;
Poussin e al., 2020). These yields a e ela i ely low, pa ly due o a simpli ied echnical i ine a y wi hou e ilize s o plan p o ec ion
p oduc s.
The lack o de ailed ield-based da a makes i challenging o asce ain he impac o annual FRA c opping on ood secu i y o
household incomes. FRA plays a signi ican ole in supplying households, wi h a la ge po ion o he p oduce consumed locally. In
Mau i ania, i is es ima ed ha in a good yea , 26% o na ional ce eal p oduc ion comes om FRA so ghum p oduc ion (Kebe, 2002).
Fu he mo e, ha es ing in Feb ua y o Ma ch helps o ensu e a s eady dis ibu ion o ce eals h oughou he yea . Howe e , epo s
om he CILSS indica e ha he egion equen ly aces ood insecu i y. Low lood yea s ha e a no able nega i e impac on ood
secu i y ac oss he alley. Fo ins ance, ollowing yea s wi h pa icula ly low loods, such as 2017/2018 o 2021/2022, egions in
Senegal and Mau i ania along he alley a e classi ied by he CILSS as expe iencing a nu i ional c isis. Con e sely, yea s wi h no mal
loods a e ypically conside ed unde p essu e. While o he ac o s also in luence ood secu i y, and a di ec co ela ion be ween
nu i ional secu i y and FRA is complex, he b oade pa e n is clea .
These insigh s unde sco e he necessi y o egula moni o ing o looded and lood- ecession cul i a ed a eas, as well as he
collec ion o da a on hei p oduc ion. In eg a ing and using localized insigh s o g ound da a wi h he ex ensi e spa ial co e age
p o ided by EO da a is c ucial o he assessmen o ood secu i y, he economic unc ionali y o hese a eas, and he enhancemen o
o e all wa e and ood secu i y policies. The p o ision o up- o-da e, high- esolu ion da a h ough emo e sensing mapping can
acili a e he sus ainable and e icien managemen o wa e esou ces, which is o i al impo ance o a ming communi ies ha
depend on na u al lood i iga ion.
6. Conclusion
Flood Recession Ag icul u e (FRA) is a i al li elihood in A ican loodplains. Dependen on he annual lood, i emains highly
ulne able o hyd ological a ia ions o e ime and ups eam changes including dam cons uc ion. The objec i e o his s udy was o
p o ide an accu a e mapping o he cul i a ed a eas in he Senegal Ri e alley. To achie e his, exis ing and eadily a ailable
me hodologies based on machine lea ning algo i hms we e mobilized o his no el applica ion, which holds signi ican ele ance o
wa e and u al managemen in Wes A ica. The esul s o he s udy e eal ha Random Fo es (RF) algo i hms combined wi h op ical
and ada indices allows o obus mapping and moni o ing o annual FRA a eas a a p ecise scale, acili a ed by he pixel esolu ion o
Sen inel images. While he classi ica ion accu acy is s ong, addi ional g ound u h da a o o he yea s and en i onmen s will allow
e inemen o a ange o hyd ological and land-use condi ions. This conside a ion is impo an in he Sahelian con ex whe e spec al
signa u e a iabili y ac oss yea s can be p onounced. Inc easing he numbe o egions o in e es is key o enhancing accu acy and
acili a ing mo e p ecise and con ex -speci ic moni o ing.
O e he pas i e yea s, FRA cul i a ion a eas in he Senegal Ri e alley ha e shown signi ican a iabili y, anging om 13,000
o 75,000 ha. The conside able deg ee o a iabili y obse ed highligh s he necessi y o egula moni o ing o FRA o manage wa e
esou ces and mi iga e he po en ial o ood c ises. The esea ch also highligh ed he in e ela ionship be ween inunda ion cha ac-
e is ics ( iming and du a ion) and he dynamics o he FRA in space and ime, indica ing ha FRA cul i a ion equi es a lood du a ion
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
17
o be ween 20 and 40 days. Unde s anding hese pa e ns is c ucial, as i p o ides c i ical insigh s in o how changes in lood egimes
a ec ag icul u al p oduc i i y and ood secu i y in lood-p one egions.
The lack o comp ehensi e FRA moni o ing has le a me s in isible and o en neglec ed in wa e and u al policies in he Senegal
basin. In his con ex , Ea h Obse a ion (EO) can play a pi o al ole in b idging he da a gap and os e ing coope a ion o e wa e
esou ces in la ge ansbounda y basins. The p esen s udy uses open-sou ce and eplicable ools such as GEE and he Random Fo es
algo i hm o p o ide an objec i e, scalable, obus , and cos -e ec i e applica ion o EO o moni o ing annual ag icul u al dynamics in
la ge loodplains. EO da a a e c i ical o in o ming wa e and u al policies ha suppo hese communi ies. As e idenced by s udies in
he Senegal and Nile basins, a ca e ul delinea ion o ade-o s be ween he demands o hyd opowe and ag icul u al demands is
impe a i e. EO da a o FRA moni o ing p o ides a aluable ool o build us among s akeholde s and add ess he mul i ace ed
demands o he Wa e -Ene gy-Food-En i onmen nexus.
CRediT au ho ship con ibu ion s a emen
Lau en B uckmann: W i ing – e iew & edi ing, W i ing – o iginal d a , Visualiza ion, Valida ion, So wa e, Resou ces,
Me hodology, In es iga ion, Fo mal analysis, Da a cu a ion, Concep ualiza ion. And ew Ogil ie: W i ing – e iew & edi ing, Vali-
da ion, Resou ces, Me hodology, In es iga ion, Funding acquisi ion. Didie Ma in: Valida ion, Resou ces, In es iga ion. Finda Bayo
Diakha ´
e: Resou ces, In es iga ion. Amau y Tilman : Supe ision, P ojec adminis a ion, Funding acquisi ion.
Decla a ion o compe ing in e es
The au ho s decla e ha hey ha e no known compe ing inancial in e es s o pe sonal ela ionships ha could ha e appea ed o
in luence he wo k epo ed in his pape .
Acknowledgemen s
This wo k was suppo ed by he GoNEXUS p ojec , which is unded by he Eu opean Union Ho izon P og amme call H2020-LC-
CLA-2018-2019-2020 - G an Ag eemen Numbe 101003722, and he AFD Cycle de l’Eau e Changemen Clima ique (CECC) p ojec .
The pa icipa ion o Canadian esea che s in GoNEXUS was made possible by he New F on ie s Resea ch Fund p og am (Canada)
(g an NFRFG-2020-00430) and by he Fonds Nou elles F on ie es en Reche che (Quebec) (g an 2022-FNFR- 310131).
We g a e ully acknowledge Google Ea h Engine o he cloud compu ing and we hank he edi o and he wo anonymous e-
iewe s o hei commen s ha helped s eng hen he manusc ip .
APPENDIX 1– Numbe o images used pe yea o analysis
Appendix 1a
Numbe o Sen inel-2 images pe mon h used o analysis be ween 2019/2020 o 2023/2024
Mon h 2019/2020 2020/2021 2021/2022 2022/2023 2023/2024
7 81 50 83 55 49
8 67 70 51 35 40
9 59 52 65 46 69
10 79 89 77 84 78
11 81 81 72 60 81
12 83 71 55 97 80
1 60 80 58 66 69
2 58 64 67 58 71
To al 568 557 528 501 537
Appendix 1b
Numbe o Sen inel-1 images pe mon h used o analysis be ween 2019/2020 o 2023/2024
Mon h 2019/2020 2020/2021 2021/2022 2022/2023 2023/2024
7 13 14 13 15 14
8 17 15 17 13 17
9 15 15 13 17 13
10 16 15 17 13 16
11 14 16 14 17 13
12 16 17 14 15 17
1 15 17 13 17 16
2 14 12 14 12 12
To al 120 121 115 119 118
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
18
APPENDIX 2– Accu acy alues and con usion ma ices
Appendix 1 c
Numbe o images used o analysis be ween 2019/2020 o 2023/2024
Sen inel-2 Tile 2019/2020 2020/2021 2021/2022/ 2022/2023 2023/2024
28PCC 30 33 28 25 27
28PDC 32 30 28 24 28
28PEC 34 32 34 27 32
28PFB 55 56 55 54 54
28PFC 63 61 61 54 58
28PGB 59 62 55 54 57
28PGC 67 56 59 54 58
28QCD 33 31 30 29 27
28QDD 65 66 60 64 67
28QED 33 35 27 28 33
28QFD 33 33 30 31 32
28QGD 64 62 61 57 64
To al 568 557 528 501 537
Appendix 2a
Calib a ion Con usion Ma ix o 2019/2020 using Random Fo es algo i hm
Land co e FRA I iga ed pe ime e Ba e soil Flooded a eas uncul i a ed Fo es o aqua ic ege a ion Wa e
FRA 463 0 0 0 1 0
I iga ed pe ime e 0 365 0 0 1 0
Ba e soil 0 0 318 0 0 0
Flooded a eas uncul i a ed 0 1 1 285 0 0
Fo es o aqua ic ege a ion 0 0 0 0 485 0
Wa e 0 0 0 0 0 128
Appendix 2b
Valida ion Con usion Ma ix o 2019/2020 using Random Fo es algo i hm
Land co e FRA I iga ed pe ime e Ba e soil Flooded a eas uncul i a ed Fo es o aqua ic ege a ion Wa e
FRA 203 0 0 5 2 0
I iga ed pe ime e 1 155 0 0 15 0
Ba e soil 1 1 135 5 2 0
Flooded a eas uncul i a ed 16 0 7 83 2 5
Fo es o aqua ic ege a ion 2 2 1 3 189 1
Wa e 0 0 0 2 0 57
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
19
APPENDIX 3– Time se ies o wa e and ege a ion indexes
Appendix 3a. – Time se ie o a e age MNDWI o e FRA and looded uncul i a ed a eas be ween July 2019 o Ma ch 2024.
Appendix 3b. – Time se ie o a e age MSAVI o e FRA and looded uncul i a ed a eas be ween July 2019 o Ma ch 2024.
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
20
Appendix 3c. – Time se ies o a e age MNDWI and NDVI alues o e di e en land co e du ing each yea analyzed.
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
21
Appendix 3c. (con inued).
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
22
Appendix 3c. (con inued).
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
23
Da a a ailabili y
Da a will be made a ailable on eques .
Re e ences
Abdikan, S., Sanli, F.B., Us une , M., Cal`
o, F., 2016. Land co e mapping using sen inel-1 SAR da a. In . A ch. Pho og am. Rem. Sens. Spa ial In . Sci. 41, 757–761.
Adamczewski, A., He zog, T., Dosso, M., Jou e, P., Jamin, J.Y., 2011. L’i iga ion peu -elle se subs i ue aux cul u es de d´
ec ue. Cah. Ag ic. 20, 97–104. h ps://doi.
o g/10.1684/ag .2011.0469.
Asa e, E., Man yka-P ingle, C., Ande son, E., Belche , K., Cla k, R., 2022. E alua ing ecosys em se ices o ag icul u al we lands: a sys ema ic e iew and me a-
analysis. We l. Ecol. Manag. 30, 1129–1149. h ps://doi.o g/10.1007/s11273-022-09857-5.
Bade , J.-C., Lamaga , J.-P., Guiguen, N., 2003. Ges ion du ba age de Manan ali su le leu e S´
en´
egal: analyse quan i a i e d’un con li d’objec i s. Hyd ol. Sci. J. 48,
525–538.
Bade , J.-C., Cauchy, S., Du a , L., Sau a, P., 2014. Monog aphie hyd ologique du leu e S´
en´
egal. De l’o igine des mesu es jusqu’en 2011. In: Edi io, I.R.D. (Ed.), IRD.
Re ie ed om. h ps://www.documen a ion.i d. /ho / di:010065190.
Balana, B.B., San o, S., Ba bie , B., Williams, T., Kola alli, S., 2019. Assessmen o lood ecession ag icul u e o ood secu i y in No he n Ghana: an op imiza ion
modelling app oach. Ag ic. Sys . 173, 536–543. h ps://doi.o g/10.1016/j.agsy.2019.03.021.
Ba bie , B., Oued aogo, H., Demb´
el´
e, Y., Yacouba, H., Ba y, B., Jamin, J.-Y., 2011. L’ag icul u e i igu´
ee dans le Sahel oues -a icain. Cah. Ag ic. 20, 24–33.
Appendix 3c. (con inued).
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
24
Ba nes, E., Cla ke, T., Richa ds, S., Colaizzi, P., Habe land, J., Kos zewski, M., Walle , P., Choi, C., Riley, E., Thompson, T., e al., 2000. Coinciden de ec ion o c op
wa e s ess, ni ogen s a us and canopy densi y using g ound based mul ispec al da a. P oceedings o he Fi h In e na ional Con e ence on P ecision Ag icul u e
1619, 6. Blooming on, MN, USA, 16–19 July 2000.
B uckmann, L., 2018. C ue e d´
e eloppemen u al dans la all´
ee du S´
en´
egal : en e ma ginalisa ion e ´
esilience. Belgeo. Re ue belge de g´
eog aphie. h ps://doi.o g/
10.4000/belgeo.23158.
B uckmann, L., 2021. La ges ion pa ag´
ee de l’eau dans le bassin du leu e S´
en´
egal : ajec oi e, enjeux e pe spec i es. L’Oues Saha ien 13–14, 261–280. h ps://doi.
o g/10.3917/ousa.211.0261.
B uckmann, L., Delba , N., Desc oix, L., Bodian, A., 2022. Recen hyd ological e olu ions o he Senegal i e lood (wes A ica). Hyd ol. Sci. J. 67, 385–400. h ps://
doi.o g/10.1080/02626667.2021.1998511.
Campos, J.C., Sille o, N., B i o, J.C., 2012. No malized di e ence wa e indexes ha e dissimila pe o mances in de ec ing seasonal and pe manen wa e in he
Saha a–Sahel ansi ion zone. J. Hyd ol. 464–465, 438–446. h ps://doi.o g/10.1016/j.jhyd ol.2012.07.042.
Chimwe a, M., Nyakudya, I.W., Jimu, L., Musemwa, L., Mashingaidze, A.B., Musa a, J.P., Kashe, K., 2022. Flood- ecession c opping in he mid-Zambezi Valley: a
neglec ed a ming sys em wi h po en ial o imp o e household ood secu i y and income. Geog . J. 188, 57–75. h ps://doi.o g/10.1111/geoj.12418.
C ousse, B., Ma hieu, P., Seck, S.M., 1991. La Vall´
ee du leu e S´
en´
egal: ´
e alua ions e pe spec i es d’une d´
ecennie d’am´
enagemen s, 1980-1990. KARTHALA Edi ions.
De G oo , D., B ande , L., Finlayson, C.M., 2018. We land ecosys em se ices. In: Finlayson, C.M., E e a d, M., I ine, K., McInnes, R.J., Middle on, B.A., Van Dam, A.
A., Da idson, N.C. (Eds.), The We land Book. Sp inge , Ne he lands, Do d ech , pp. 323–333. h ps://doi.o g/10.1007/978-90-481-9659-3_66.
Dong, J., Xiao, X., Kou, W., Qin, Y., Zhang, G., Li, L., Jin, C., Zhou, Y., Wang, J., Bi ada , C., Liu, J., Moo e, B., 2015. T acking he dynamics o paddy ice plan ing a ea
in 1986–2010 h ough ime se ies Landsa images and phenology-based algo i hms. Remo e Sens. En i on. 160, 99–113. h ps://doi.o g/10.1016/j.
se.2015.01.004.
D yden, R., Anand, M., Lehne , B., Flue -Chouina d, E., 2021. Do we p io i ize loodplains o de elopmen and a ming? Mapping global dependence and exposu e o
inunda ion. Global En i on. Change 71, 102370. h ps://doi.o g/10.1016/j.gloen cha.2021.102370.
F a al, P., Bade , J.-C., Man´
e, L.K., Da id-Benz, H., Lamaga , J.-P., Diop Diagne, O., 2002. The ques o in eg a ed and sus ainable wa e managemen in he Senegal
Ri e Valley. ENVIROWATER, 5`
eme Con ´
e ence In e -R´
egionale su l’En i onnemen e l’Eau, Ouagadougou. h p://ag i op.ci ad. /511230/.
Gao, B., 1996. NDWI—a no malized di e ence wa e index o emo e sensing o ege a ion liquid wa e om space. Remo e Sens. En i on. 58, 257–266. h ps://doi.
o g/10.1016/S0034-4257(96)00067-3.
Ga cía-Bola˜
nos, M., Bo gia, C., Poblado , N., Dia, M., Seyid, O.M.V., Ma eos, L., 2011. Pe o mance assessmen o small i iga ion schemes along he Mau i anian
banks o he Senegal Ri e . Ag ic. Wa e Manag. 98, 1141–1152. h ps://doi.o g/10.1016/j.agwa .2011.02.008.
Gibb, S.A., 1987. Pa ne s, Elec ici ´
e de F ance & Eu oconsul . E ude de la ges ion des ou ages communs de l’OMVS. OMVS, Daka , S´
en´
egal.
Gumma, M.K., Thenkabail, P.S., Maunahan, A., Islam, S., Nelson, A., 2014. Mapping seasonal ice c opland ex en and a ea in he high c opping in ensi y en i onmen
o Bangladesh using MODIS 500m da a o he yea 2010. ISPRS J. Pho og amme y Remo e Sens. 91, 98–113. h ps://doi.o g/10.1016/j.isp sjp s.2014.02.007.
Gumma, M.K., Amede, T., Ge ne , M., Pinja la, B., Panjala, P., Legesse, G., Tilahun, G., Van Den Akke , E., Be del, W., Kelle , C., Siambi, M., Whi b ead, A.M., 2022.
Assessing po en ial loca ions o lood-based a ming using sa elli e image y: a case s udy o A a egion, E hiopia. Renew. Ag ic. Food Sys . 37, S28–S42. h ps://
doi.o g/10.1017/S1742170519000516.
Gup a, P iyanka, Gup a, P a eek, Singh, S.K., Thaku , B., Kuma , M., 2024. Ensemble me hods-based compa a i e s udy o Landsa 8 ope a ional land image (OLI)
and sen inel 2 mul i-spec al images (MSI) o sma a ming c op classi ica ion. Mul imed. Tool. Appl. h ps://doi.o g/10.1007/s11042-024-19737-0.
Gxokwe, S., 2022. Le e aging Google Ea h Engine pla o m o cha ac e ize and map small seasonal we lands in he semi-a id en i onmen s o Sou h A ica. Sci. To al
En i on.
Gxokwe, S., Dube, T., Maz ima i, D., 2020. Mul ispec al emo e sensing o we lands in semi-a id and a id a eas: a e iew on applica ions, challenges and possible
u u e esea ch di ec ions. Rem. Sens. 12, 4190. h ps://doi.o g/10.3390/ s12244190.
Ho owi z, M.M., Salem-Mu dock, M., Magis o, J., Niasse, M., Nu all, C., 1990. Senegal Ri e Basin Moni o ing Ac i i y: Syn hesis Repo . Ins i u e o De elopmen
An h opology, Bingham on, New Yo k.
Hu, Y., Huang, J., Du, Y., Han, P., Huang, W., 2015. Moni o ing spa ial and empo al dynamics o lood egimes and hei ela ion o we land landscape pa e ns in
dong ing lake om MODIS ime-se ies image y. Rem. Sens. 7, 7494–7520. h ps://doi.o g/10.3390/ s70607494.
Jiang, X., Fang, S., Huang, X., Liu, Y., Guo, L., 2021. Rice mapping and g ow h moni o ing based on ime se ies GF-6 images and ed-edge bands. Rem. Sens. 13, 579.
h ps://doi.o g/10.3390/ s13040579.
Kebe, B., 2002. Le ˆ
ole du so gho de d´
ec ue dans la secu i ´
e alimen ai e au Sahel. La cul u e du so gho de d´
ec ue en A ique de l’Oues e du Cen e 23. Agence
Espagnole de Coop´
e a ion In e na ionale, ISBN 84-7232-898-8.
Kool, M., an S eenbe gen, F., Meha i Haile, A., Abbas, Y.M., Hagos, E., 2018. The p omise o lood-based a ming sys ems in a id and semi-a id a eas. In: Leal
Filho, W., de T inche ia Gomez, J. (Eds.), Rainwa e -Sma Ag icul u e in A id and Semi-a id A eas: Fos e ing he Use o Rainwa e o Food Secu i y, Po e y
Alle ia ion, Landscape Res o a ion and Clima e Resilience. Sp inge In e na ional Publishing, pp. 77–94. h ps://doi.o g/10.1007/978-3-319-66239-8_5.
Kool, J., Lhe mi e, S., H achowi z, M., B egoli, F., McClain, M.E., 2022. Seasonal inunda ion dynamics and wa e balance o he Ma a We land, Tanzania based on
mul i- empo al Sen inel-2 image classi ica ion. In . J. Appl. Ea h Obs. Geoin . 109, 102766. h ps://doi.o g/10.1016/j.jag.2022.102766.
Liman Ha ou, I., Whi ney, C., Kung’u, J., Luedeling, E., 2020. Mapping lood-based a ming sys ems wi h bayesian ne wo ks. Land 9, 369. h ps://doi.o g/10.3390/
land9100369.
Mandal, D., Kuma , V., Bha acha ya, A., Rao, Y.S., Siquei a, P., Be a, S., 2018. Sen4Rice: a p ocessing chain o di e en ia ing ea ly and la e ansplan ed ice using
ime-se ies sen inel-1 SAR da a wi h Google ea h engine. Geosci. Rem. Sens. Le . IEEE 15, 1947–1951. h ps://doi.o g/10.1109/LGRS.2018.2865816.
Man´
e, L.K., F a al, P., 2001. Sui i pa ´
el´
ed´
e ec ion des cul u es de d´
ec ue dans la Vall´
ee du Fleu e S´
en´
egal en saison oide 2000/2001. SAED-IWMI-WARDA epo .
Mo sumi, S., Magole, L., Kga hi, D., 2012. Indigenous knowledge and land use policy: implica ions o li elihoods o lood ecession a ming communi ies in he
Oka ango Del a, Bo swana. Physics and Chemis y o he Ea h, Pa s A/B/C 50–52 185–195. h ps://doi.o g/10.1016/j.pce.2012.09.013.
Nguyen, D.B., G ube , A., Wagne , W., 2016. Mapping ice ex en and c opping scheme in he Mekong Del a using Sen inel-1A da a. Remo e Sensing Le e s 7 (12),
1209–1218.
Ogil ie, A., Belaud, G., Delenne, C., Bailly, J.-S., Bade , J.-C., Oleksiak, A., Fe y, L., Ma in, D., 2015. Decadal moni o ing o he Nige Inne Del a lood dynamics
using MODIS op ical da a. J. Hyd ol. 523, 368–383. h ps://doi.o g/10.1016/j.jhyd ol.2015.01.036.
Ogil ie, A., Poussin, J.-C., Bade , J.-C., Bayo, F., Bodian, A., Dacos a, H., Dia, D., Diop, L., Ma in, D., Sambou, S., 2020. Combining mul i-senso sa elli e image y o
imp o e long- e m moni o ing o empo a y su ace wa e bodies in he Senegal i e loodplain. Rem. Sens. 12, 3157. h ps://doi.o g/10.3390/ s12193157.
Ogil ie, A., Fall, C.S., Bodian, A., Ma in, D., B uckmann, L., Dia, D., Leye, I., Ndiaye, P.M., So o, D.D., Danumah, J.H., Bade , J.-C., Poussin, J.-C., 2025. Su ace wa e
and lood-based ag icul u al sys ems: mapping and modelling long- e m a iabili y in he Senegal i e loodplain. Ag ic. Wa e Manag. 308, 109254. h ps://doi.
o g/10.1016/j.agwa .2024.109254.
OMVS & IRD, 2002. P og amme d’Op imisa ion e de Ges ion des R´
ese oi s (POGR). Phase inale, appo de syn hese, Annexe 1 : Cul u es de dec ue. Daka ,
Senegal, 2002.
OMVS, 2018. Plan d’ac ion ´
egional pou l’am´
elio a ion des cul u es i igu´
ees du bassin du leu e S´
en´
egal. Daka , S´
en´
egal., 2018
O ieschnig, C.A., Belaud, G., Veno , J.-P., Massuel, S., Ogil ie, A., 2021. Inpu image y, classi ie s, and cloud compu ing: insigh s om mul i- empo al LULC mapping
in he Cambodian Mekong Del a. Eu opean Jou nal o Remo e Sensing 54, 398–416. h ps://doi.o g/10.1080/22797254.2021.1948356.
Poussin, J.-C., 1998. Diagnos ic su les sys `
emes de izicul u e i igu´
ee dans la moyenne all´
ee a al du leu e S´
en´
egal. In: Bia nes, A. (Ed.), La condui e du champ
cul i ´
e: poin s de ue d’ag onomes. O s om, Pa is, pp. 133–161.
Poussin, J.-C., Ma in, D., Bade , J.-C., Dia, D., Seck, S.M., Ogil ie, A., 2020. Va iabili ´
e ag o-hyd ologique des cul u es de d´
ec ue. Une ´
e ude de cas dans la moyenne
all´
ee du leu e S´
en´
egal. Cah. Ag ic. Fo . 29, 23. h ps://doi.o g/10.1051/cag i/2020022.
Qi, J., e al., 1994. A modi ied soil adjus ed ege a ion index. Remo e Sens. En i on. 48, 119–126.
L. B uckmann e al.
Remo e Sensing Applica ions: Socie y and En i onmen 37 (2025) 101473
25
