Investigating the Influence of Land Use Change on River Channel Dynamics in Douala, Cameroon

INVESTIGATING THE INFLUENCE OF LANDUSE CHANGE ON RIVER CHANNEL DYNAMICS IN DOUALA, CAMEROON ROMARICK LEINYUY TEWIY ii INVESTIGATING THE INFLUENCE OF LAND USE CHANGE ON RIVER CHANNEL DYNAMICS IN DOUALA, CAMEROON. Disse a ion supe ised by P o . Ped o Cab al (NOVA IMS, Po ugal) Co-supe ised by D . Bakh ia Feizizadeh (Uni e si y o Müns e , Ge many) & P o . Filibe o Pla (UJI, Spain) Feb ua y 2024 iii DECLARATION OF ORIGINALITY I decla e ha he wo k desc ibed in his documen is my own and no om someone else. All he assis ance I ha e ecei ed om o he people is duly acknowledged and all he sou ces (published o no published) a e e e enced. This wo k has no been p e iously e alua ed o submi ed o NOVA In o ma ion Managemen School o elsewhe e. Roma ick Leinyuy Tewiy [ he signed o iginal has been a chi ed by he NOVA IMS se ices] i ACKNOWLEDGMENTS I would like o exp ess my deepes g a i ude o P o . Ped o Cab al, my main hesis supe iso , o his unwa e ing suppo , in aluable guidance, and schola ly insigh h oughou he en i e esea ch p ocess. I am also g a e ul o my co-supe iso s, P o . Filibe o Pla and D . Bakh ia Feizizadeh o hei cons uc i e eedback and hough ul sugges ions, which g ea ly con ibu ed o he success o his wo k. I also ex end my app ecia ion o God and o my amily o hei encou agemen , unde s anding, and pa ience du ing his demanding pe iod. Thei lo e and suppo ha e been a cons an sou ce o mo i a ion. Finally, I would like o acknowledge my iends and colleagues who ha e sha ed hei expe ise and p o ided encou agemen h oughou his academic jou ney. This hesis would no ha e been possible wi hou he collec i e suppo and inspi a ion om all hese indi iduals. INVESTIGATING THE INFLUENCE OF LAND USE CHANGE ON RIVER CHANNEL DYNAMICS IN DOUALA, CAMEROON. ABSTRACT Ri e s ha e pe ennially sculp ed he ea h's landscape, nu u ing ci iliza ions and sus aining biodi e si y. In Douala, Came oon, he in ica e dance be ween human p og ess and he na u al low o he Wou i Ri e p esen s a c i ical case s udy o his dynamic. In his comp ehensi e s udy, he impac o land use changes on i e channel dynamics in Douala, Came oon, is me iculously analyzed h ough he lens o ad anced emo e sensing da a and sophis ica ed machine lea ning algo i hms. This s udy emba ks on a empo al jou ney, un a eling he ans o ma ion o Douala's landscape o e ecen decades, om lush ege a ion and we lands o an inc easingly u banized milieu, and i s p o ound implica ions on he Wou i Ri e 's mo phology. Employing a blend o Landsa image y and Google Ea h Engine capabili ies, he esea ch delinea es shi s in land use pa e ns, emphasizing he enc oachmen o u ban de elopmen in o c i ical na u al habi a s. The indings unde sco e he u gency o in eg a ing sus ainable u ban planning wi h en i onmen al conse a ion s a egies o sa egua d i e ine ecosys ems agains he backd op o explosi e u ban g ow h. The hesis a icula es a ision o u u e u ban de elopmen ha is in ha mony wi h he ecological and hyd ological in eg i y o i e sys ems, ad oca ing o policy amewo ks ha p io i ize en i onmen al sus ainabili y alongside economic and demog aphic expansion. This seminal wo k con ibu es signi ican ly o he discou se on sus ainable de elopmen , o e ing a bluep in o policymake s, u ban planne s, and en i onmen al s ewa ds aiming o mi iga e he ecological oo p in o u baniza ion while os e ing esilien and sus ainable u ban ecosys ems. This s udy con ibu es o he achie emen o he ollowing Sus ainabili y De elopmen s Goals (SDGs), Goals 6 (Clean Wa e and Sani a ion), 11 (Sus ainable Ci ies and Communi ies), 13 (Clima e Ac ion), and 15 (Li e on Land) i KEYWORDS In luence o Land Use Change Ri e channel Dynamics Google Ea h Engine ii ACRONYMS LULC - Land Use and Land Co e GEE - Google Ea h Engine USGS - Uni ed S a es Geological Su ey CEMAC - Cen al A ican Economic and Mone a y Communi y DLA - Douala In e na ional Ai po GIS - Geog aphic In o ma ion Sys em RF - Random Fo es (Machine Lea ning Algo i hm) SVM - Suppo Vec o Machine (Machine Lea ning Algo i hm) GEE - Google Ea h Engine NDWI - No malized Di e ence Wa e Index UHI - U ban Hea Island SDGs - Sus ainable De elopmen Goals RGB - Red, G een, Blue (Colo Bands o Sa elli e Image y) NIR - Nea In a ed iii INDEX OF THE TEXT TITLE PAGE ..................................................................................................................................ii DECLARATION OF ORIGINALITY …………..........................................................................iii ACKNOWLEDGMENTS …………..............................................................................................i ABSTRACT ………........................................................................................................................ KEYWORDS ……………………………….…………………………………………………… i ACRONYMS................................................................................................................................. ii INDEX OF TEXT …...…………………………………………..………………….. iii INDEX OF TEXT …...…………………………………………..……………………ix INDEX OF TABLES………………………………………………………..…….…...x INDEX OF FIGURES …………………,………….……………………………..…...xi INDEX OF EQUATONS …………………….…………………..…………………..xi 1 INTRODUCTION .....................................................................................................................1 1.1 Rele ance o S udy ...................................................................................................................2 1.2 Resea ch Objec i es .................................................................................................................2 1.3 Resea ch Ques ions ..................................................................................................................3 2 LITERATURE REVIEW .........................................................................................................4 2.1 Land Use Change: T ends and D i e s……….……………………………………................4 2.2 Ri e Channel Dynamics: Responses o Land Use Change .....................................................4 2.3 In e play Be ween Land Use Change and Ri e Dynamics .....................................................5 2.4 Socio-Economic Implica ions o En i onmen al Change ........................................................5 2.5 Rela ed Wo k ............................................................................................................................6 2.6 Gaps ..........................................................................................................................................7 3 DATA AND METHODOLOGY ................………………………………………………….8 3.1 S udy A ea ...............................................................................................................................8 3.2 Da a ..........................................................................................................................................9 3.2.1 Remo e Sensing Da a ....................................................................................................... 9 3.3 Me hods ..................................................................................................................................11 3.3.1 Me hodology Wo k low……………………………………………...............................11 3.3.2 Da a P epa a ion ..............................................................................................................11 3.3.3 Indices .............................................................................................................................12 3.4 Land Use Land Co e Classi ica ion .....................................................................................12 3.4.1 Accu acy Assessmen .....................................................................................................14 3.5 Change De ec ion Analysis ....................................................................................................14 4 RESULTS AND DISCUSSION .............................................................................................16 4.1 Accu acy Assessmen ............................................................................................................16 4.2 Land Use Land Co e Classi ica ion .....................................................................................28 4.3 Change De ec ion ..................................................................................................................21 ix 4.3.1 T ansi ions om Ri e Channel o O he LULC Classes ..............................................22 4.3.2 T ansi ions om O he LULC Classes o Ri e Channel ..............................................23 5 DISCUSSION, CONCLUSION, AND RECOMMENDATIONS ……………………….26 5.1 Discussion .............................................................................................................................26 5.2 Conclusion ............................................................................................................................26 5.3 Recommenda ions .................................................................................................................27 Bibliog aphic Re e ences ...........................................................................................................29 4 CHAPER TWO 2: LITERATURE REVIEW The in ica e ela ionship be ween land use changes and i e channel dynamics is a c i ical a ea o s udy wi hin en i onmen al science, especially conce ning a eas like Douala, Came oon. This li e a u e e iew me iculously analyzes he in luence o land use change on i e channel dynamics h ough an ex ensi e examina ion o 17 di e se s udies. These s udies collec i ely o e a comp ehensi e insigh in o he mul i ace ed in e play be ween an h opogenic land use ans o ma ions and he na u al p ocesses go e ning i e sys ems. The e iew is s uc u ed a ound ou pi o al hemes: Land Use Change: T ends and D i e s, Ri e Channel Dynamics: Responses o Land Use Change, In e play Be ween Land Use Change and Ri e Dynamics, and Socio-Economic Implica ions o En i onmen al Change, aiming o p o ide a nuanced unde s anding o he subjec ma e . 2.1: Land Use Change: T ends and D i e s The con e sion o land o ag icul u al, u ban, and indus ial uses has accele a ed globally, d i en by economic de elopmen , popula ion g ow h, and he demands o u baniza ion. The ans o ma ion om na u al landscapes o modi ied ones has p o ound implica ions o local and egional hyd ology, geomo phology, and biodi e si y. James L.A. and Lecce S.A. (2013) emphasize he his o ical dep h o human impac on i e sys ems, acing al e a ions o he Neoli hic e a and unde sco ing he ole o echnological ad ancemen s and popula ion g ow h in hese changes. This his o ical pe spec i e is c ucial o unde s anding he cumula i e impac o human ac i i ies on i e sys ems. Simila ly, s udies like Mesmin e al. (2020) de ail he apid u ban expansion in Came oon, highligh ing he p essing issue o de o es a ion and he con e sion o land o ag icul u e and u ban de elopmen . Such changes a e emblema ic o b oade en i onmen al ans o ma ions, whe e land use modi ica ions dis up na u al hyd ological cycles, leading o al e ed soil composi ions and, ul ima ely, impac ing i e mo phology. The esea ch collec i ely unde sco es a global end o en i onmen al change d i en by an h opogenic ac i i ies, e ealing he unde lying d i e s o land use ans o ma ion as complex and mul i ac o ial, in ol ing economic, social, and poli ical dimensions. 2.2: Ri e Channel Dynamics: Responses o Land Use Change Ri e channels a e inhe en ly dynamic, esponding o a mul i ude o na u al and an h opogenic in luences. The al e a ion o land use pa e ns, especially h ough u ban expansion and ag icul u al de elopmen , has signi ican implica ions o i e mo phology. Inc eased uno om impe ious su aces, al e ed sedimen anspo dynamics, and changes in ege a ion co e due o land use change can lead o signi ican modi ica ions in i e channel o m and unc ion. Fo ins ance, Tiwa i e al. (2023) and Fo si e al. (2019) p o ide de ailed analyses o how 5 u baniza ion and land use changes in luence sedimen yield and uno , he eby a ec ing i e channel mo phology and beha io . These s udies illus a e he di ec co ela ion be ween land use in ensi y and he shi ing o i e channels, highligh ing he i e 's ole as a dynamic indica o o en i onmen al change. The wo k o Bhunia, Shi , and Pal (2016) on he Ganges and Sa ka e al. (2012) on he B ahmapu a Ri e u he elucida es he geomo phological impac s o land use changes, emphasizing he c i ical need o ongoing moni o ing and unde s anding o i e channel dynamics in he ace o apid en i onmen al ans o ma ion. The esponsi eness o i e channels o land use changes unde sco es he delica e balance be ween na u al i e p ocesses and he p essu es exe ed by human ac i i ies. 2.3: In e play Be ween Land Use Change and Ri e Dynamics The ela ionship be ween land use changes and i e dynamics is cha ac e ized by a complex eedback loop, whe e each in luences and is in luenced by he o he . Changes in land use, such as de o es a ion, u baniza ion, and ag icul u al expansion, p ecipi a e al e a ions in he hyd ological egime, which in u n, p omp mo phological changes in i e channels. These mo phological changes can mani es as a ia ions in sinuosi y, mig a ion pa e ns, and bank s abili y, which a e c i ical o main aining ecological balance, biodi e si y, and human li elihoods. Deb, Das, and Uddin (2012) and Hasanuzzaman e al. (2021) illus a e how i e dynamics can p omp u he land use changes, necessi a ing lood con ol measu es and he cons uc ion o p o ec i e in as uc u e. This cyclical in e play necessi a es a comp ehensi e unde s anding o bo h land use and i e ine sys ems o mi iga e ad e se impac s and p omo e sus ainable de elopmen . The s udies e iewed highligh he impo ance o adop ing an in eg a ed app oach o en i onmen al managemen ha conside s he dynamic in e ac ions be ween land use changes and i e mo phology. 2.4: Socio-Economic Implica ions o En i onmen al Change The socio-economic implica ions o land use changes and he consequen al e a ions in i e channel dynamics a e p o ound. Shi s in i e channels can e ode a able land, impac ing ag icul u al p oduc i i y, and necessi a e signi ican in es men s in u ban planning and disas e isk managemen o mi iga e inc eased lood isks. S udies like ha o S.S.U. (2014) on he Jamuna Ri e and Pa anelli e al. (2019) on he Reno Ri e Ca chmen elucida e he socio- economic impac s o i e dynamics, including esou ce des uc ion, popula ion displacemen , and he challenges o ensu ing sus ainable u ban de elopmen in he ace o changing i e ine landscapes. The in e connec ion be ween en i onmen al changes, i e dynamics, and socio- economic ac o s poin s o he need o in eg a ed land use and i e managemen s a egies ha conside he ull spec um o impac s and This li e a u e e iew unde sco es he in ica e and mul i ace ed ela ionship be ween land use changes and i e channel dynamics. D awing on a b oad a ay o s udies, i highligh s he c i ical 6 need o in eg a ed, sus ainable app oaches o en i onmen al managemen . These app oaches mus conside he complex in e play be ween geomo phological p ocesses, land use ans o ma ions, and socio-economic ac o s. Such holis ic s a egies a e i al o add essing he challenges posed by en i onmen al changes, pa icula ly in egions like Douala, Came oon, whe e he balance be ween de elopmen and en i onmen al sus ainabili y emains a p essing conce n. 2.5: Rela ed wo k The s udy by A e in, Mesh am, and Şeke (2021) on he Padma Ri e in Bangladesh illus a es he signi ican impac s o i e channel mig a ion on LULC changes, emphasizing he c i ical ole o GIS and Remo e Sensing (RS) echnologies in moni o ing hese dynamics. This esea ch could in o m he Douala con ex by highligh ing he need o ad anced echnological app oaches o assess he e ec s o LULC changes on i e channels. Dwa akish and Ganas i (2015) e iew unde sco es he c i ical ole o hyd ological modeling in unde s anding he hyd ological impac s o land use change. Thei discussion on a ious modeling app oaches, including de e minis ic and s ochas ic models, p o ides a amewo k o analyzing he hyd ological esponses o LULC changes in he Douala egion, s essing he impo ance o in eg a ing land use models wi h hyd ological simula ions. Kang and Kanniah (2022) demons a e he applica ion o machine lea ning classi ie s in analyzing LULC changes and hei impac on i e mo phology in he Joho Ri e Basin, Malaysia. The use o Google Ea h Engine (GEE) o long- e m LULC da a analysis could be pa icula ly ele an o he Douala s udy, o e ing insigh s in o employing cu ing-edge ools o accu a e LULC classi ica ion and unde s anding hei implica ions on i e dynamics. The esea ch by Bekele e al. (2021) in he Awash Ri e basin, E hiopia, employs he Soil and Wa e Assessmen Tool (SWAT) model o explo e hyd ological changes due o LULC dynamics. This s udy's me hodology and indings on he hyd ological impac s o land use changes could guide simila in es iga ions in Douala, emphasizing he impo ance o model calib a ion and alida ion o accu a e hyd ological p edic ions. Walling's (2017) ex on sedimen dynamics o e s a comp ehensi e o e iew o he challenges in p edic ing sedimen beha io , which is c ucial o unde s anding i e channel esponses o LULC changes. The insigh s in o sedimen anspo p ocesses and he impac o human ac i i ies on sedimen dynamics could in o m he analysis o sedimen - ela ed i e changes in Douala. Vanacke e al. (2005) ocus on he Deleg Ri e basin in he Andes, examining he e ec s o human-induced LULC changes on i e mo phology. This s udy's app oach o assessing spa ial o ganiza ion and connec i i y o land uni s could o e aluable pe spec i es on managing e osion and op imizing wa e esou ce u iliza ion in he Douala con ex . Finally, he s udy by Sadhwani e al. (2022) on he Pe iya Ri e Wa e shed, India, emphasizes he SWAT model's ole in analyzing he e ec s o LULC changes on low dynamics and sedimen 7 yield. The de ailed calib a ion and alida ion p ocesses, along wi h he emphasis on sub- wa e shed le el analysis, could p o ide a me hodological bluep in o assessing he in luence o LULC changes on i e channel dynamics in Douala. By in eg a ing he me hodologies, indings, and insigh s om hese s udies, a li e a u e e iew can shed ligh on he complex in e play be ween land use change and i e channel dynamics in Douala, Came oon. This e iew would no only con ibu e o he academic discou se bu also o e p ac ical guidelines o sus ainable land and wa e esou ce managemen in he egion. 2.5: Gaps The li e a u e e iew o e s a ho ough examina ion o he in ica e ela ionship be ween land use change and i e channel dynamics in Douala, Came oon, highligh ing he mul i ace ed d i e s o land use change, such as economic g ow h and u baniza ion, and hei consequen ial impac s on i e mo phology and hyd ological p ocesses. Despi e he aluable insigh s gleaned om exis ing li e a u e, signi ican gaps pe sis , pa icula ly in unde s anding he speci ic linkages be ween u baniza ion, de o es a ion, ag icul u al p ac ices, and i e sys ems in Douala, as well as he in eg a ion o mul iple heo ies o ully comp ehend hese in e ac ions. Add essing hese gaps is impe a i e o ailo ing in e en ions and policy p oposals o he local con ex , ensu ing e ec i e decision-making and sus ainable de elopmen ini ia i es in Douala, Came oon. Mo ing o wa d, his mas e 's hesis aims o b idge hese gaps by examining he ex en and na u e o land use changes in Douala o e he pas decades using machine lea ning algo i hms on Google Ea h Engine (GEE), analyzing he dynamics o i e channels wi h land use change o assess hei po en ial impac on i e channel dynamics. Th ough hese esea ch objec i es, he hesis endea o s o con ibu e o a deepe unde s anding o he complex dynamics o land use change and i e channels in Douala, ul ima ely in o ming sus ainable de elopmen and en i onmen al conse a ion s a egies in he egion. 8 CHAPTER 3 3: DATA AND METHODOLOGY 3.1: S udy a ea. Douala, he la ges ci y in Came oon, s ands as he economic and comme cial nucleus no only o he coun y bu also o he en i e Cen al A ican Economic and Mone a y Communi y (CEMAC) egion, encompassing Gabon, Congo, Chad, Equa o ial Guinea, Cen al A ican Republic, and Came oon. His o ically, he a ea a ound Douala has been inhabi ed by indigenous popula ions along he Wou i Ri e es ua y. Du ing he la e 19 h cen u y, Douala became a Ge man ading pos unde he name Duala du ing he Ge man colonial pe iod. Pos -Wo ld Wa I, Came oon was di ided be ween he F ench and B i ish, wi h Douala becoming pa o F ench Came oon upon independence in 1960. Figu e 1: Loca ion o Douala O e he yea s, Douala has ans o med in o a cosmopoli an ci y wi h a di e se popula ion and a dynamic cul u al scene. The ci y boas s Cen al A ica's la ges po , hough K ibi po is g adually becoming an essen ial complemen . Douala's s a egic loca ion on he Wou i Ri e es ua y acili a es ade, handling a signi ican po ion o Came oon's majo expo s, including oil, cocoa, co ee, imbe , me als, and ui s. Douala In e na ional Ai po (DLA) se es as a majo in e na ional ga eway, con ibu ing o he ci y's economic signi icance. As o 2023, Douala and i s su ounding a ea had an es ima ed popula ion o 5,768,400, e lec ing i s s a us as one o he mos popula ed ci ies in Came oon. Geog aphically si ua ed on he es ua y o he Wou i Ri e , Douala expe iences a opical clima e wi h high empe a u es and signi ican ain all. Ongoing de elopmen s, including he cons uc ion o he K ibi po , signi y he ci y's e ol ing economic landscape, a ac ing in es men s, and solidi ying i s pi o al ole in Came oon's economic g ow h. 9 3.2: Da a. 3.2.1 Remo e Sensing Da a This hesis le e ages emo e sensing da a o explo e he impac s o land use changes on i e channel dynamics wi hin he Douala egion, Came oon. A se ies o high- esolu ion Landsa sa elli e image y, each wi h a esolu ion o 30 me e s, was sys ema ically selec ed o acili a e his in es iga ion. Spanning om 1986 o 2022, he empo al scope o he image y allows o a comp ehensi e analysis o he long- e m in e ac ions be ween land use ans o ma ions and lu ial p ocesses. Speci ically, he s udy inco po a es image y om Landsa 5 o he yea 1986, Landsa 8 o 2015, and Landsa 9 o 2022 om USGS downloaded ia Google Ea h Engine. This selec ion no only p o ides a his o ical pe spec i e bu also o e s con empo a y insigh s, ensu ing a obus empo al compa ison. I is no ewo hy ha all sa elli e images we e acqui ed du ing Came oon's d y season, which spans om Decembe o Ap il, o minimize he in luence o seasonal hyd ological a iabili y on he analysis. Due o he pe sis en cloud co e cha ac e is ic o he Douala egion, which equen ly obscu es sa elli e isibili y, he in e als be ween he selec ed images a e no uni o m. This i egula i y is a ibu ed o he necessi y o choosing cloud- ee images o main ain da a in eg i y and analy ical p ecision. Consequen ly, he 1986 image y comp ises a mosaic o wo images om he espec i e yea , while he image y o 2015 and 2022 a e composi ed o op imize cla i y. These images we e p ocu ed om he Uni ed S a es Geological Su ey (USGS) ia he Google Ea h Engine pla o m, ensu ing a high s anda d o image quali y and accessibili y o he esea ch. Below a e he Landsa sa elli e images used o his hesis. The able below shows he a ious band combina ions and esolu ion o he sa elli e images 10 Table 1: Da a Sou ces The images used o his analysis can be seen below 1986 (Landsa 5) 2015 (Landsa 8) 2022 (Landsa 9) Figu e 2: Landsa Sa elli e Image ies Yea Landsa Sa elli e Da a Sou ce and Band Desc ip ion Image esolu ion (Me e s) 1986 Landsa 5 Seconda y Da a Sou ce; TM senso bands: Band 1 (Blue), Band 2 (G een), Band 3 (Red), Band 4 (Nea -In a ed), Band 5 (Sho wa e In a ed), Band 6 (The mal), Band 7 (Sho wa e In a ed) 30 2015 Landsa 8 Seconda y Da a Sou ce; OLI senso bands: Band 1 (Coas al/Ae osol), Band 2 (Blue), Band 3 (G een), Band 4 (Red), Band 5 (Nea -In a ed), Band 6 (Sho wa e In a ed 1), Band 7 (Sho wa e In a ed 2), Band 8 (Panch oma ic), Band 9 (Ci us); TIRS senso bands: Band 10 (The mal In a ed 1), Band 11 (The mal In a ed 2) 30 2022 Landsa 9 Seconda y Da a Sou ce; OLI-2 senso bands simila o Landsa 8; TIRS-2 senso bands: Band 10 (The mal In a ed 1), Band 11 (The mal In a ed 2), imp o ed calib a ion o he mal bands 30 11 3.3: Me hods 3.3.1: Me hodology wo k low This s udy used di e en pla o ms and So wa es o accomplish his ask. The Land use Land co e classi ica ion was pe o med using he Google ea h engine cloud compu ing pla o m. The change de ec ion and depic ion o i e channel dynamics was done on A c GIS. The me hodology s eps can be seen on he me hodology low cha below. Figu e 3: Me hodology low cha . 3.3.2: Da a p epa a ion. Fo isualizing he Landsa image ies, he RGB bands we e employed ac oss all selec ed yea s o main ain consis ency in he isual assessmen . The 1986 image y used he gene al RGB bands 'B4' ( ed), 'B3' (g een), and 'B2' (blue). Simila ly, he 2015 and 2022 images, om he Landsa 8 and Landsa 9 sa elli es espec i ely, u ilized hei co esponding RGB bands o isualiza ion: 'B4', 'B3', and 'B2' o 2015, and 'SR_B4', 'SR_B3', and 'SR_B2' o 2022. This s anda diza ion in he use o RGB bands ensu ed ha he isual compa ison o land use changes o e he yea s was consis en and accu a e o he s udy o i e channel dynamics in Douala, Came oon 12 3.3.3: Indices The No malized Di e ence Wa e Index (NDWI) is an essen ial ool in emo e sensing ha is u ilized o de ec and moni o he p esence o wa e on he land su ace. I is pa icula ly use ul o di e en ia ing liquid wa e om e es ial ege a ion and soil, especially in delinea ing open wa e ea u es such as lakes, i e s, and ese oi s. The o mula o NDWI is exp essed as NDWI = G een – NIR G een + NIR In his o mula, G een is he e lec ance in he g een band NIR is he e lec ance in he nea -in a ed band The index akes ad an age o he ac ha wa e abso bs mo e in a ed and sho -wa e in a ed ligh han i does in o he pa s o he spec um, which means ha open wa e has a low e lec ance in he SWIR pa o he spec um while ege a ion e lec s mo e NIR. The NDWI o mula is designed o maximize he e lec ance o wa e by using he NIR band, which ends o e lec well o wa e su aces, and o minimize i by sub ac ing he e lec ance om he SWIR band, which wa e abso bs. The di ision by he sum o NIR and SWIR e lec ance no malizes he index, scaling he alues be ween -1 and 1. This no maliza ion is c ucial as i accoun s o opog aphic and a mosphe ic a ia ions, imp o ing he eliabili y o wa e de ec ion. In his s udy, NDWI is employed o e alua e he spa ial dis ibu ion and empo al changes o wa e con en in he landscape o Douala. By applying NDWI, he esea ch is able o eliably quan i y he ex en o wa e bodies, ack hei changes o e ime, and gain insigh s in o he hyd ological dynamics a play. This con ibu es o a be e unde s anding o land-wa e in e ac ions and aids in he e ec i e managemen and planning o wa e esou ces in he egion. 3.4: Land use Land Co e classi ica ion In his s udy, he sa elli e da a unde wen an analy ical p ocess using a pai o machine lea ning echniques: Random Fo es (RF) and Suppo Vec o Machine (SVM), all wi hin he con ines o he Google Ea h Engine (GEE) en i onmen . The aim was o assess he pe o mance capabili ies o each algo i hm h ough me ics o o e all accu acy and he kappa s a is ic. This would enable he selec ion o he mos e ec i e machine lea ning app oach o subsequen classi ica ion asks in he in es iga ion. In his in es iga ion, he u ban landscape o Douala was delinea ed in o six disc e e Land Use and Land Co e (LULC) ca ego ies o he yea s 2015 and 2022, and se en ca ego ies o he yea 1986. These ca ego ies comp ise Buil -Up a eas, Ri e Channels, Vege a ion (including bo h 13 o es ed a eas and g asslands), Ag icul u al land, Ba en Land, and We lands. Fo he 1986 da ase , an addi ional classi ica ion o 'Clouds' was necessi a ed due o he p esence o cloud co e in he sa elli e image y. The able below p o ides a de ailed desc ip ion o he LULC classi ica ion classes u ilized in his s udy. Land Use Class Desc ip ion Ag icul u e A eas used o a ming and c op p oduc ion Ba en Land Land wi h li le o no ege a ion Buil -Up Zones wi h esiden ial, indus ial, comme cial buildings o pa ed oads Ri e _channel Pa hs o he na u al wa e cou ses Vege a ion Regions co e ed wi h g asses, ees, and o he plan li e We land Land a eas sa u a ed wi h wa e , ei he pe manen ly o seasonally Table 2: LULC Classi ica ion class desc ip ion In mi iga ing he challenge o limi ed ield-based da a collec ion, he on-sc een selec ion o sample poin s om high- esolu ion Google Ea h images is execu ed. This me hodological app oach, alida ed in p io s udies (Kanniah, 2017; Lin e al., 2018), add esses he necessi y o g ound u hing o aining and alida ion. A sys ema ic app oach in ol es he andom iden i ica ion o sample poin s, wi h hei dis ibu ion based on he a ea o each land use class. No ably, he emphasis is placed on main aining sample poin s a consis en loca ions h oughou he 36-yea s udy pe iod o educe he impac o hei loca ion on he classi ica ion esul s. The selec ed sample poin s a e hen in eg a ed in o he GEE sc ip o he classi ica ion p ocess, wi h andom di ision in o aining samples (80%) and alida ion samples (20%). This me hod ensu es obus ness and eliabili y in he absence o ex ensi e ield da a collec ion. This able below shows he sample poin s o each image as well as numbe o poin s used o aining and es ing Yea sample poin s T aining Tes ing 1986 768 619 149 2015 915 713 202 2022 1083 842 241 Table 3: Sample poin s o LULC Classi ica ion . 20 4.3: Change De ec ion An in e sec ion was pe o med be ween he 1986 and he 2022 LULC classi ica ion images o see he changes ha ha e occu ed o e he leng h o he s udy pe iod in Douala. The esul s o his can be seen in he able below. YEARS 2022 1986 LULC Classes Ag icul u e Ba eland Buil Up Ri e _channel Vege a ion We land To al Ag icul u e 0.57 0.08 26.15 0.0034 1.78 0.02 28.61 Ba eland 0.04 0.02 1.29 0.0001 0.09 0.001 1.44 Buil Up 0.44 0.19 58.70 0.16 4.82 0.08 64.39 Ri e _channel 0.24 0.07 2.35 68.38 8.70 0.52 80.26 Vege a ion 39.10 2.17 122.81 9.77 267.83 6.72 448.44 We land 1.17 0.06 6.57 1.00 8.43 0.86 18.1 To al 41.57 2.60 217.87 79.32 291.68 8.20 641.24 Table 10: LULC T ansi ion be ween 1986 and 2022 2022 LULC Classes Ag icul u e (%) Ba eland (%) Buil Up (%) Ri e _channel (%) Vege a ion (%) We land (%) TOTAL (%) 1986 Ag icul u e 2.00 0.29 91.41 0.01 6.21 0.07 100 Ba eland 3.00 1.13 89.31 0.01 6.47 0.08 100 Buil Up 0.68 0.30 91.16 0.25 7.49 0.12 100 Ri e _channel 0.30 0.09 2.93 85.20 10.84 0.65 100 Vege a ion 8.72 0.48 27.39 2.18 59.72 1.50 100 We land 6.47 0.35 36.30 5.55 46.57 4.74 100 Table 11: LULC T ansi ion be ween 1986 and 2022 (%) The LULC ansi ion ma ix indica es no able land co e changes in Douala om 1986 o 2022. Ag icul u al land has dec eased by app oxima ely 26 km², ansi ioning la gely o buil -up a eas, which ha e expanded by abou 176 km², e lec ing a signi ican u ban sp awl. The e has been educ ion o i e channel a eas which ha e dec eased o e all, mainly yielding o buil -up a eas and o ege a ion, likely due o his u ban enc oachmen o expansion o ege a ion. Vege a ion co e has educed by app oxima ely 124 km², wi h hese a eas also becoming u banized. We lands ha e seen a dec ease o nea ly 10 km², sugges ing a loss o hese ecologically i al a eas. These shi s om na u al and semi-na u al landscapes o u ban en i onmen s indica e a p o ound 21 ans o ma ion in Douala's land use o e he las 36 yea s, highligh ing he need o sus ainable u ban planning o mi iga e he impac on i e channel dynamics. 4.3.1: T ansi ions om Ri e channel o o he LULC classes be ween 1986 and 2022 Based on change de ec ion ca ied ou , i could be obse ed ha be ween 1986 and 2022, he i e channel ans o med in o o he land use classes. The able below shows his T ansi ion CLASS CHANGE AREA CHANGE (KMSQ) Ri e _channel - Ag icul u e 0.24 Ri e _channel - Ba eland 0.07 Ri e _channel - Buil Up 2.35 Ri e _channel - Vege a ion 8.7 Ri e _channel - We land 0.52 Table 12: LULC T ansi ion om Ri e channel o o he LULC classes Figu e 6: G aphical ep esen a ion o LULC T ansi ions om Ri e channel o o he LULC classes. The esul s indica e a no able shi in land use om he i e channel o o he landuse classes o e ime. A signi ican expansion in o buil -up a eas, wi h an a ea change o 2.35 squa e kilome e s, e lec s he p essu es o u baniza ion. The i e channel's con e sion o ege a ion is he mos subs an ial, a 8.7 squa e kilome e s, sugges ing po en ial e o es a ion o na u al ege a ion enc oachmen . Ag icul u al land shows a modes inc ease. 0 1 2 3 4 5 6 7 8 9 10 Ri e _channel - Ag icul u e Ri e _channel - Ba eland Ri e _channel - Buil Up Ri e _channel - Vege a ion Ri e _channel - We land AREA CHANGE (KMSQ) CLASS CHANGE RIVER CHANNEL TRANSITIONS TO OTHER LULC CLASSES (SQKM) 22 Figu e 7: T ansi ions om he Ri e Channel o o he LULC classes The map dis ibu ion illus a es hese ans o ma ions, highligh ing he i e 's dec easing a ea, pa icula ly in he main cou se o he Wou i Ri e . These changes ha e c i ical implica ions o he i e 's dynamics, including al e ed low egimes and ecological impac s due o u ban and ege a ion con e ion. 4.3.2: T ansi ions F om o he LULC classes o i e -channel be ween 1986 and 2022: Du ing he s udy pe iod, he i e unde wen bo h educ ion and expansion in a ea. Some land p e iously classi ied unde di e en uses has ansi ioned in o i e a ea, as demons a ed in he p o ided able and g aph. This indica es a dynamic change in land use and i e mo phology. CLASS CHANGE AREA CHANGE (SQKM) Ag icul u e - Ri e _channel 0.0034 Ba eland - Ri e _channel 0.0001 Buil Up - Ri e _channel 0.16 Vege a ion - Ri e _channel 9.77 We land - Ri e _channel 0.86 Table 13: LULC T ansi ion om o he LULC classes o he i e channel 23 Figu e 8: G aphical ep esen a ion o LULC T ansi ions om o he LULC classes o Ri e Channel As can be obse ed, he able and g aph abo e indica e a ansi ion o land co e om o he classes o i e channel, wi h he mos signi ican change being a con e sion om ege a ion o he i e channel a 9.77 squa e kilome e s. This sugges s a subs an ial inc ease in i e ine a ea po en ially due o changes in land managemen o na u al eclama ion o p e iously ege a ed a eas. Buil - up a eas show a educ ion o 0.16 squa e kilome e s, indica ing a dec ease in u ban land use adjacen o he i e , possibly e lec ing e o s o educe enc oachmen on he i e channel o he e ec s o e osion. Below is he ansi ion map om o he land use classes o i e channel 0 2 4 6 8 10 12 Ag icul u e - Ri e _channel Ba eland - Ri e _channel Buil Up - Ri e _channel Vege a ion - Ri e _channel We land - Ri e _channel AREA CHANGE (KMSQ) CLASS CHANGE LULC TRANSITIONS TO RIVER CHANNEL (SQKM) 24 Figu e 9: T ansi ion om o he LULC classes o he Ri e _channel The map isualizes hese ansi ions and highligh s he pe iphe al gains in he i e 's a ea, con as ing wi h losses in he main cou se. The expansion a ound he pe iphe ies could be a ibu ed o sedimen deposi ion o changes in land use p ac ices, while he na owing o he main cou se could esul om inc eased sedimen a ion o al e ed low pa e ns due o ups eam land use changes. These shi s a e c ucial o unde s anding he dynamics o he i e sys em, as hey can a ec hyd ological connec i i y, sedimen anspo , and loodplain dynamics. 25 CHAPTER 5 5: DISCUSSION, CONCLUSION AND RECOMMENDATIONS 5.1: Discussion The esul s om he Land Use Land Co e (LULC) classi ica ion e eal a nuanced pic u e o Douala's ans o ma ion o e h ee decades, unde sco ing he c i ical in e play be ween u ban expansion and i e channel dynamics. The supe io pe o mance o he SVM classi ie o he yea s 1986 and 2022, and he RF classi ie o 2015, illus a es he nuanced di e ences in algo i hm e icacy dependen on he speci ic cha ac e is ics o each da ase . This disc epancy highligh s he impo ance o choosing he igh machine lea ning ool based on he da a's empo al con ex and he na u e o land co e changes. The signi ican u ban sp awl e lec ed in he expansion o buil -up a eas om 64.39 km² in 1986 o 217.87 km² in 2022 unde sco es he agg essi e pace o u baniza ion in Douala. This g ow h has led o conside able losses in ege a ion and we lands, c ucial o biodi e si y and lood mi iga ion, illus a ing he p o ound impac o human ac i i ies on na u al landscapes. The LULC ansi ion analysis be ween 1986 and 2022 u he de ails he complexi y o land use changes, whe e ag icul u al lands and na u al ege a ion a e inc easingly con e ed o u ban spaces. This shi no only ep esen s a loss o g een co e bu also signi ies changes in he hyd ological cycle and sedimen anspo dynamics, po en ially exace ba ing lood isks and al e ing i e ine ecosys ems. The ansi ion om i e channels o o he land use classes, pa icula ly o buil -up a eas and ege a ion, e lec s he dynamic na u e o Douala's landscape, whe e na u al wa e cou ses a e being eshaped by an h opogenic p essu es. These ans o ma ions necessi a e a ecalib a ion o u ban planning and en i onmen al conse a ion s a egies o p ese e he ecological and hyd ological unc ions o i e sys ems amids apid u ban de elopmen . This discussion unde lines he impo ance o sus ainable u ban expansion ha ha monizes wi h i e ine ecosys em p ese a ion, ensu ing he esilience o u ban landscapes agains en i onmen al and clima ic challenges. 5.2: Conclusion This mas e 's hesis s ands as a pi o al explo a ion in o he nuanced in e play be ween land use changes and i e channel dynamics wi hin Douala, Came oon, agains a can as o apid u ban de elopmen and en i onmen al shi s. Le e aging ad anced me hodologies such as emo e sensing da a analysis, machine lea ning algo i hms, and comp ehensi e land-use and land-co e (LULC) classi ica ion, his esea ch illumina es he p o ound impac s o human ac i i ies on he na u al en i onmen . I unde sco es he u gen need o in eg a i e s a egies in sus ainable de elopmen and en i onmen al conse a ion, pa icula ly in he ealm o u ban planning. The s udy unco e s a s a k ans o ma ion in Douala's landscape o e ecen decades, cha ac e ized by apid u ban expansion, a signi ican dec ease in ege a ion and we lands and al e a ions in ag icul u al land use pa e ns. This u ban sp awl, ueled by economic and demog aphic g ow h, 26 p esen s conside able challenges o he ecological and hyd ological in eg i y o he a ea, especially a ec ing he i e sys ems ha a e c ucial o he egion's ecological heal h and socio-economic sus ainabili y. The indings no only de ail he ex en and na u e o hese land use changes bu also a icula e hei dynamic epe cussions on he mo phology o i e channels, including shi s in sedimen deposi ion, e osion pa e ns, and channel mig a ion a es. A pa icula ly no ewo hy aspec o his hesis is he insigh in o he ansi ional dynamics be ween di e en land use classes and he i e channel. The s udy me iculously documen s he con e sion o signi ican i e channel a eas o buil -up egions and ege a ion, e lec ing a complex in e ac ion whe e land use changes and i e dynamics mu ually in luence each o he . This ela ionship unde sco es he necessi y o an in eg a ed app oach o u ban and en i onmen al managemen ha add esses he complex d i e s o landscape change, om human p essu es o na u al p ocesses. Inco po a ing he obse a ion o he na owing o he main i e cou se alongside pe iphe al gains en iches ou unde s anding o he delica e equilib ium be ween sedimen a ion, e osion, and land use p ac ices. This phenomenon poin s o c i ical implica ions o hyd ological connec i i y, sedimen anspo , and loodplain dynamics, highligh ing he sophis ica ed in e dependencies wi hin he i e ine ecosys em. By add essing he esea ch ques ions posed, his hesis signi ican ly con ibu es o elucida ing he subs an ial land use ans o ma ions and hei impac s on i e channel dynamics in Douala. I ad oca es o holis ic u ban planning and en i onmen al managemen s a egies ha p io i ize he conse a ion o na u al esou ces amid u ban de elopmen p essu es, aligned wi h he Uni ed Na ions Sus ainable De elopmen Goals. In conclusion, his esea ch lays a ounda ion o u u e inqui y and policy o mula ion aimed a econciling u ban de elopmen wi h en i onmen al conse a ion, o e ing a bluep in o policy and en i onmen al s ewa dship ha emphasizes he uni e sal challenge o ha monizing human de elopmen wi h he p ese a ion o he na u al wo ld. Th ough his comp ehensi e explo a ion, he hesis highligh s he c i ical ole o in o med, sus ainable de elopmen s a egies in ensu ing he longe i y and i ali y o i e ine ecosys ems o gene a ions o come, unde pinning he g ea ness encapsula ed in he s udy's indings and me hodologies. 5.3: Recommenda ions In ligh o he agg essi e u baniza ion obse ed in Douala, Came oon, and i s conside able e ec s on i e channel dynamics and en i onmen al sus ainabili y, his mas e 's hesis s ongly ecommends ha u u e s udies pi o owa ds in es iga ing he spa io- empo al empe a u e a ia ions wi hin he ci y, pa icula ly in ela ion o clima e change. U ban expansion, a known ca alys o al e ing land use pa e ns and i e mo phologies, also signi ican ly in luences local clima e sys ems h ough mechanisms such as he u ban hea island (UHI) e ec , leading o a a ie y o ecological and socio-economic consequences. The need o unde s and how u baniza ion impac s local empe a u e a ia ions is c ucial, as i con ibu es o b oade clima e change implica ions, including al e ed wea he pa e ns and inc eased hea - ela ed heal h isks. Employing emo e sensing and geog aphic in o ma ion sys em (GIS) echnologies, which ha e p o en 27 e ec i e in analyzing land use change and i e dynamics, can p o ide p ecise da a on how u ban g ow h in luences empe a u e dispa i ies ac oss Douala. Such an in es iga ion will no only shed ligh on he mic o-clima ic e ec s o u baniza ion in opical ci ies bu also o e insigh s in o global clima e change adap a ion and u ban esilience s a egies. This esea ch di ec ion is i al o in o ming sus ainable u ban planning and clima e mi iga ion e o s, ensu ing ha u u e u ban de elopmen in Douala, and simila se ings, aligns wi h b oade en i onmen al sus ainabili y and clima e esilience objec i es. To add ess he impac s o u ban de elopmen on i e dynamics e ec i ely, in as uc u e p ojec s wi hin Douala mus be designed wi h a keen awa eness o he na u al i e low pa e ns. This en ails adop ing enginee ing p ac ices ha p io i ize he conse a ion o wa e cou ses, ensu ing ha cons uc ion does no impede he i e s’ ecological unc ions o con ibu e o lood isk. In eg a i e app oaches, such as he cons uc ion o g een in as uc u e and he implemen a ion o na u al wa e e en ion measu es, should be p io i ized. These me hods no only suppo lood managemen bu also enhance u ban biodi e si y and ec ea ional spaces, os e ing a ha monious coexis ence be ween u ban de elopmen and i e ine ecosys ems. This s a egic alignmen unde sco es he necessi y o c oss-sec o al collabo a ion among u ban planne s, en i onmen al scien is s, and ci il enginee s o de ise solu ions ha a e bo h inno a i e and sus ainable, ensu ing he long- e m heal h and esilience o Douala's i e sys ems amids ongoing u ban expansion. 28 BIBLIOGRAPHIC REFERENCES A e in, R., Mesh am, S. G., & Şeke , D. Z. (2021). Ri e channel mig a ion and land-use/land- co e change o Padma Ri e a Bangladesh: a RS- and GIS-based app oach. In e na ional Jou nal o En i onmen al Science and Technology, 18(10), 3109–3126. h ps://doi.o g/10.1007/s13762- 020-03063-7 Bekele, D., Alami ew, T., Kebede, A., Zeleke, G., & Melesse, A. M. (2021). Modeling he impac s o land use and land co e dynamics on hyd ological p ocesses o he Kele a wa e shed, E hiopia. Sus ainable En i onmen , 7(1). h ps://doi.o g/10.1080/27658511.2021.1947632 Bhunia, G. S., Shi , P. K., & Pal, D. K. (2016). Channel dynamics associa ed wi h land use/co e change in Ganges i e , India, 1989–2010. Spa ial In o ma ion Resea ch, 24(4), 437–449. h ps://doi.o g/10.1007/s41324-016-0045-7 Deb, M., Das, D., & Uddin, M. (2012). E alua ion o Meande ing Cha ac e is ics Using RS & GIS o Manu Ri e . Jou nal o Wa e Resou ce and P o ec ion, 04(03), 163–171. h ps://doi.o g/10.4236/jwa p.2012.43019 Des E Walling Uni e si y o Exe e 556 PUBLICATIONS 41, 402 CITATIONS. (2017). Sedimen e osion, anspo , and deposi ion. In Chap e . Dwa akish, G. S., & Ganas i, B. P. (2015). Impac o land use change on hyd ological sys ems: A e iew o cu en modeling app oaches. Cogen Geoscience, 1(1), 1115691. h ps://doi.o g/10.1080/23312041.2015.1115691 Fo si, Pou eau, B enon, Onguene, & E ame. (2019). Tempo al (1948–2012) and Dynamic E olu ion o he Wou i Es ua y Coas line wi hin he Gul o Guinea. Jou nal o Ma ine Science and Enginee ing, 7(10), 343. h ps://doi.o g/10.3390/jmse7100343 Hasanuzzaman, Md., Gayen, A., Haque, Sk. M., & Shi , P. K. (2021). Spa ial Modeling o Ri e Bank Shi ing and Associa ed LULC Changes o he Kaljani Ri e in Himalayan Foo hills. Resea ch Squa e Pla o m LLC. h ps://doi.o g/10.21203/ s.3. s-336521/ 1 James, L. A., & Lecce, S. A. (2013). 9.37 Impac s o Land-Use and Land-Co e Change on Ri e Sys ems. T ea ise on Geomo phology, 768–793. h ps://doi.o g/10.1016/b978-0-12-374739- 6.00264-5 Kang, C. S., & Kanniah, K. D. (2022). Land use and land co e change and i s impac on i e mo phology in Joho Ri e Basin, Malaysia. Jou nal o Hyd ology: Regional S udies, 41, 101072. h ps://doi.o g/10.1016/j.ej h.2022.101072 Mesmin, T., F édé ic, S., E ic, V., Philippes, M. F., Rose, N. M., Ismaël, I., & Séd ic, T. F. (2020). Land Use and Land Co e changes in he Cen e Region o Came oon. In P ep in s. 29 Pa anelli, D., Ca azza, C., La nić, S., & Toscano, A. (2019). The Long-Te m E ec s o Land Use and Clima e Changes on he Hyd o-Mo phology o he Reno Ri e Ca chmen (No he n I aly). Wa e , 11(9), 1831. h ps://doi.o g/10.3390/w11091831 Sadhwani, K., Eldho, T. I., Jha, M. K., & Ka maka , S. (2022). E ec s o Dynamic Land Use/Land Co e Change on Flow and Sedimen Yield in a Monsoon-Domina ed T opical Wa e shed. Wa e , 14(22), 3666. h ps://doi.o g/10.21203/ s.3. s-336521/ 1 Sa ka , A., Ga g, R. D., & Sha ma, N. (2012). RS-GIS Based Assessmen o Ri e Dynamics o B ahmapu a Ri e in India. Jou nal o Wa e Resou ce and P o ec ion, 04(02), 63–72. h ps://doi.o g/10.4236/jwa p.2012.42008 Tiwa i, A. K., Mish a, M., Singh, R., & Singh, G. S. (2023). Assessmen o i e channel dynamics and i s impac on land use/land co e in he middle Ganga plain, India. A abian Jou nal o Geosciences, 16(4). h ps://doi.o g/10.1007/s12517-023-11307-2 Vanacke , V., Molina, A., Go e s, G., Poesen, J., De con, G., & Decke s, S. (2005). Ri e channel esponse o sho - e m human-induced change in landscape connec i i y in Andean ecosys ems. Geomo phology, 72(1–4), 340–353. h ps://doi.o g/10.1016/j.geomo ph.2005.05.013