Keeping it simple: Monitoring flood extent in large data-poor wetlands using MODIS SWIR data

Wolski, Piotr; Murray-Hudson, Michael; Thito, Kgalalelo; Cassidy, Lin

doi:10.1016/j.jag.2017.01.005

Cited by 34 publications

(62 citation statements)

References 34 publications

(29 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This large wetland consists of a panhandle region, a channel system surrounded by permanent swamps, and a large, low gradient alluvial fan [6,8,9]. The delta is a flood pulsed system subject to an annual flood event which is asynchronous with the local rainy season; rainfall in the highlands of Angola flows into the Okavango River, entering the Botswana panhandle in April/May, and slowly moves down the fan, reaching maximum inundation extent in July-September [4,6,8,[10][11][12]. Intra and inter-annual variations in the frequency, duration, and extent of the inundation produces a complex mosaic of vegetation, supporting a vast number of ecological niches and a rich diversity of flora and fauna [3,13,14].…”

Section: Introductionmentioning

confidence: 99%

“…shrublands, grasslands) on satellite imagery is relatively simple, there is spectral overlap between inundated areas covered in aquatic vegetation (e.g. floodplain) and some dryland vegetation classes, making separation of these classes difficult and traditional water classification methods unviable [4,[10][11][12]15,16]. Therefore a range of classification methods (unsupervised, supervised, band thresholding, band ratios, indices, and combinations of these methods) have been trialled [4,6,8,[10][11][12]16].…”

Section: Introductionmentioning

confidence: 99%

“…floodplain) and some dryland vegetation classes, making separation of these classes difficult and traditional water classification methods unviable [4,[10][11][12]15,16]. Therefore a range of classification methods (unsupervised, supervised, band thresholding, band ratios, indices, and combinations of these methods) have been trialled [4,6,8,[10][11][12]16]. Recently, the relatively simple method of band thresholding has been successful [4,10,16], with thresholding of the short wave infrared (SWIR) band producing high accuracy results on MODIS imagery [11].…”

Section: Introductionmentioning

confidence: 99%

“…The majority of delta inundation studies have used imagery with broad spatial resolution (MODIS (250m, 500m, and 1km) [10,11,16] and NOAA AVHRR (1km) [8,12], but see [4,16]), taking advantage of the high temporal resolution of these sensors, which allows sub-monthly analysis of inundation [11]. However, dependent on the intended use of the inundation maps, such broad spatial resolution may result in unacceptable simplification of the complex mosaic of the delta [10].…”

Section: Introductionmentioning

confidence: 99%

“…In addition, broad spatial resolution increases the likelihood of mixed pixels (e.g. half inundated, half dry) which can confuse classification attempts [6,11].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Automated Inundation History Mapping over Large Areas using Landsat and Google Earth Engine

Inman¹,

Mb²

2020

Preprint

View full text Add to dashboard Cite

Accurate inundation maps for flooded wetlands and rivers are a critical resource for their management and conservation. In this paper we automate a method (thresholding of the short-wave infrared band) for classifying inundation, using Landsat imagery and Google Earth Engine. We demonstrate the method in the Okavango Delta, northern Botswana, a complex case study due to the spectral overlap between inundated areas covered with aquatic vegetation and dryland vegetation classes on satellite imagery. Inundation classifications in the Okavango Delta have predominately been implemented on broad spatial resolution images. We present the longest time series to date (1990-2019) of inundation maps at high spatial resolution (30m) for the Okavango Delta. We validated the maps using image-based and in situ data accuracy assessments, with accuracy ranging from 91.5 - 98.1%. Use of Landsat imagery resulted in consistently lower estimates of inundation extent than previous studies, likely due to the increased number of mixed pixels that occur when using broad spatial resolution imagery, which can lead to overestimations of the size of inundated areas. We provide the inundation maps and Google Earth Engine code for public use.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…In addition, broad spatial resolution increases the likelihood of mixed pixels (e.g. half inundated, half dry) which can confuse classification attempts [6,11].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Automated Inundation History Mapping over Large Areas using Landsat and Google Earth Engine

Inman¹,

Mb²

2020

Preprint

View full text Add to dashboard Cite

show abstract

Landscape‐wide pulse events predict trait‐based responses among wetland birds in perennial channels of a dryland wetland

Becker,

Boyes,

Wittmer

et al. 2023

Ecological Applications

View full text Add to dashboard Cite

Wetlands in arid or semi‐arid zones are vital for maintaining biodiversity but face growing threats. Flooding regime variability is a key driver of ecological dynamism in these systems, dictating primary productivity on a large spatial scale. Functional composition or diversity of wetland‐dependent bird species has been found to be sensitive to fluctuations in hydrological regimes and can thus be indicative of cascading ecosystem responses associated with climate change. In this paper we investigate whether large‐scale changes in inundation and fire – a significant additional biodiversity determinant in (semi‐)arid landscapes – were reliable predictors of functional group responses of wetland‐dependent birds along a perennial channel of the Okavango Delta, Botswana. We fit generalized additive models (GAM) to six years of bird survey data collected along ~190‐km‐long annual transects and use remotely sensed landscape‐level inundation estimates, as well as spatio‐temporal distance to fire, to predict responsiveness of seven trait‐based functional group abundances. During the surveys a total of 89 different wetland‐dependent bird species were recorded, including 76 residents, across all years, with below‐surface feeding waders consistently the most abundant functional group. Including estimated spatio‐temporal variability in flooding and fire, as well as their interactions, improved model fit for all seven functional groups, explaining between 46.8% and 68.3% of variability in functional group abundances. Covariates representing longer‐term variability in inundation generally performed better than shorter‐term ones. For example, variability in inundation over the 5 months preceding a survey best predicted the responses of all functional groups, which also all exhibited responsiveness to the interaction between flooding and fire. We were able to interpret the responses of individual functional groups, based on the resource exploitation assumption. Overall, our results suggest that perennial waters in dryland wetlands offer functional refugia to wetland‐dependent birds and highlight the indicative power of large‐scale trait‐based bird monitoring. Our findings demonstrate the potential utility of such a monitoring regime for dryland wetland ecosystems vulnerable to industrial‐scale anthropogenic pressure and associated climate change.ABSTRACT (Setswana)Morethetho wa go tshela ga metsi a makgobokgobo a a fitlhelwang mo nageng tsa komelelo, o rotloetsa mekgwa e dinonyane tse di fitlhelwang mo meeleng e e sa kgaleng ya makgobokgobo, di tsibogelang diemo ka teng.Makgobokgobo a a fitlhelwang mo dikgaolong tsa komelelo a botlhokwa mo go tshegetseng mefutafuta ya ditshidi, mme a lebaganywe ke matshosetsi. Go farologana ga ka fa metsi a makgobokgobo a tshelang ka teng, ke ntlha ya konokono e e tlhotlheletsang diphetogo mo nageng tse, ebile e laola ka fa dimela di dirang ka teng mo nageng eo ka bophara. Mefutafuta ya dinonyane tse di ikaegileng ka makgobokgobo e fitlhetswe e amiwa thata ke diemo tsa metsi tse di fetogafetogang, ka jalo mefuta e, e ka nna kaedi ya mekgwa e botshelo bo fetogang ka teng, segolobogolo fa re lebile phetogo e e sa tlwaelesegang ya loapi (Climate change). Mo mokwalong o, re sekaseka gore a diphetogo tsa maphashaphasha a metsi le melelo ya naga‐ nngwe ya dintlha tsa botlhokwa tse di rotloetsang mefutafuta ya ditshidi mo dinageng tsa komelelo ‐ ele bokao jo bo netlameng go ka supa ka fa mefuta ya dinoyane tse di ikaegileng ka makgobokgobo di tsibogelang diemo ka teng mo moeleng wa makgobokgobo a Okavango, mo Botswana. Go re thusa ka ditshekatsheko tse, re dirisitse boranyane jwa ditlhaka (Generalized Additive Models) go rarabolola patlisitso e e dirilweng mo dingwageng tse thataro ka dinonyane tse di bonweng mo tseleng ya sekgele sa boleele ja 190 km. Tsela e ya tshekatsheko e ne e tsamaiwa gangwefela mo ngwageng. Re ne gape ra dirisa maranyane a lefaufau (Remote sensing) go akanyetsa bophara jwa maphashaphasha a Okavango le go meta sekgele le sebaka sa molelo wa naga. Metlhale e ya tshekatsheko, e ne ya dirisiwa go akanyetsa pele phetogo ya dipalo tsa mefuta e supa ya dinonyane tsa makgokgobobo. Dipatlisiso di supile fa mo dingwageng tse thataro go gatisitswe mefuta e e farologanyeng ya dinonyane tse di ikaegileng ka makgobokgobo di le masome a a ferang bobedi le borobabongwe (89), go akarediwa tsa sennela ruri di le masome a supa le borataro (76). Go supegile gape fa mofuta wa dinonyane tse di jang dijo ko tlase ga metsi e le tsone di palo ntsi. Tharabololo ya maranyane a dipalo (model fit) e kgonne go tlhalosa gore fa gare ga masome mane le borataro le ntlha tse robabobedi mo lekgolong (46.8%) le masome marataro le boferabobedi le ntlha tse tharo mo lekgolong (68.3%) ya diphetogo tse di bonweng tsa mefuta ya dinonyane e bakilwe ke diphetogo tse di farologanyeng tsa go tshela ga metsi a makgobokgobo le molelo wa naga le ka fa dintlha tse di amanang ka teng. Mabaka a a rotloetsang diphetogo tsa maphashaphasha a metsi ka lebaka lo leleele, ka kakaretso a supegile a dira botoka go supa tsibogo ya dinonyane gona le a lebaka le le khutshwane. Sekai, diphetogo tsa selekanyo sa maphashaphasha a metsi mo kgweding tse tlhano (5 months) pele ga dipatlisiso di akanyeditse‐pele botoka ka fa ditlhopha tsotlhe tsa dinoyana di tsibogelang diemo ka teng. E supile tsibogelo diemo ya dinonyane e e bakiwang ke kamano magareng ga go tshela ga metsi le melelo wa naga. Re ne ra kgona go tlhalosa ka fa ditlhopha tsa dinonyane ka bongwe ja tsone di tsibogelang diemo ka teng, re dirisa kakanyetsong ya mekgwa ya tiriso ya ditsa tlholego. Ka kakaretso, maduo a rona a kaya fa metsi a sennela ruri a a bonwang mo makgobokgobong a naga ya komelelo, a a fa dinonyane tse di ikaegileng ka makgobokgobo lefelo la botshabelo, gape a senotse nonofo ya ditshekatsheko tsa peoleitho tse di dirwang ka dinonyane ka lebaka le selekanyo se segolo ebile se tseneletse. Maduo a rona a supile botlhokwa jwa tiriso ya ditlhotlhomiso tsa mofuta o mo makgobokgobong a a bonwang mo nageng tsa komelelo ebile a le mo diphatseng tsa kgotlelesego go tswa mo ditlhabololong le tse di amanang le phetogo e e sa tlwaelesegang ya loapi.

show abstract

Remote sensing to characterize inundation and vegetation dynamics of upland lagoons

et al. 2022

View full text Add to dashboard Cite

Understanding broad trends in the distribution and composition of wetlands is essential for making evidence‐based management decisions. Determining temporal change in the extent of inundation in wetlands using remote sensing remains challenging and requires on‐ground verification to determine accuracy and precision. Therefore, optimization and validation of remote sensing methods in threatened wetlands is a high priority for their conservation. Despite their ecological importance in the landscape, we have little knowledge of the variation in the spatial extent of inundation in upland lagoons, a threatened ecological community in New South Wales, Australia. Our project developed locally trained algorithms to predict the extent of water and emergent vegetation using imagery from the Landsat‐5, ‐7, and ‐8 satellites. The best model for upland lagoons used shortwave infrared reflectance (performing better than normalized difference spectral indices), with model accuracy against validation transects greater than 95%. We applied the model to images from 1988 to 2020 across 58 lagoons to generate a dataset that demonstrates the variable water regime and vegetation change in response to local rainfall over 32 years such as in the lagoons. Our results reduce threats to a dynamic threatened ecological community by filling an important knowledge gap and demonstrate a valuable method to understand historical and current changes in the hydrology of dynamic wetland systems more broadly.

show abstract

Keeping it simple: Monitoring flood extent in large data-poor wetlands using MODIS SWIR data

Cited by 34 publications

References 34 publications

Automated Inundation History Mapping over Large Areas using Landsat and Google Earth Engine

Automated Inundation History Mapping over Large Areas using Landsat and Google Earth Engine

Landscape‐wide pulse events predict trait‐based responses among wetland birds in perennial channels of a dryland wetland

Remote sensing to characterize inundation and vegetation dynamics of upland lagoons

Contact Info

Product

Resources

About