TY - JOUR

T1 - Variation of the Omo Delta between 1990 and 2018

T2 - What remote sensing data reveal and models explain

AU - Kangi, Gladys

AU - Dondeyne, Stefaan

AU - Kleinschroth, Fritz

AU - Van Orshoven, Jos

N1 - Publisher Copyright: © 2023 John Wiley & Sons Ltd.

PY - 2024/1/30

Y1 - 2024/1/30

N2 - Deltas, being areas where river sediments accumulate, are sensitive to changes in climate and anthropogenic processes that affect sediment generation. The Omo Delta, on the Ethiopia–Kenya border, is constantly changing due to fluctuating lake levels, a variable climate, and rapidly changing land use. Due to field data scarcity and the region's limited accessibility, we relied on remote sensing (RS) data to investigate delta extent between 1990 and 2018 and attempted to disentangle the effect of climatic from anthropogenic variables. We obtained a time series of delta extent from Landsat imagery using Random Forest (RF) classification. Using parametric and non-parametric regression techniques, we regressed the delta extent on Lake Turkana levels, Omo River discharges, Omo-Gibe Basin rainfall, tree cover loss, and irrigation extent. The RF algorithm distinguished water from land with high (>90%) accuracies and revealed Omo Delta extent fluctuations ranging from 949 km2 in 1993 to 651 km2 in 2000. Lake water level, which depends on the rainfall over the Omo-Turkana Basin, emerged as the best predictor of delta extent. However, the annual rainfall over the Omo-Gibe Basin showed no correlation with delta extent. The regression models further show a connection between delta extent, irrigation extent, and tree cover loss. We conclude that rainfall indirectly influences delta extent across the Omo-Turkana Basin. Regression models indicate additional cumulative effects of human activity in the Omo-Gibe Basin but fall short in explaining delta dynamics.

AB - Deltas, being areas where river sediments accumulate, are sensitive to changes in climate and anthropogenic processes that affect sediment generation. The Omo Delta, on the Ethiopia–Kenya border, is constantly changing due to fluctuating lake levels, a variable climate, and rapidly changing land use. Due to field data scarcity and the region's limited accessibility, we relied on remote sensing (RS) data to investigate delta extent between 1990 and 2018 and attempted to disentangle the effect of climatic from anthropogenic variables. We obtained a time series of delta extent from Landsat imagery using Random Forest (RF) classification. Using parametric and non-parametric regression techniques, we regressed the delta extent on Lake Turkana levels, Omo River discharges, Omo-Gibe Basin rainfall, tree cover loss, and irrigation extent. The RF algorithm distinguished water from land with high (>90%) accuracies and revealed Omo Delta extent fluctuations ranging from 949 km2 in 1993 to 651 km2 in 2000. Lake water level, which depends on the rainfall over the Omo-Turkana Basin, emerged as the best predictor of delta extent. However, the annual rainfall over the Omo-Gibe Basin showed no correlation with delta extent. The regression models further show a connection between delta extent, irrigation extent, and tree cover loss. We conclude that rainfall indirectly influences delta extent across the Omo-Turkana Basin. Regression models indicate additional cumulative effects of human activity in the Omo-Gibe Basin but fall short in explaining delta dynamics.

KW - climate change

KW - irrigation

KW - Landsat images

KW - multiple linear regression

KW - Omo-Turkana

KW - Random Forest

UR - http://www.scopus.com/inward/record.url?scp=85174566262&partnerID=8YFLogxK

U2 - 10.1002/ldr.4958

DO - 10.1002/ldr.4958

M3 - Article

AN - SCOPUS:85174566262

VL - 35

SP - 867

EP - 883

JO - Land Degradation and Development

JF - Land Degradation and Development

SN - 1085-3278

IS - 2

ER -