Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Pays Bas → 2020 → Mapping small-scale informal irrigation using time series of high-resolution data from Sentinel 2A/B satellites : A case study from Kenya

UNESCO-IHE Institute for Water Education, Delft (2020)

Mapping small-scale informal irrigation using time series of high-resolution data from Sentinel 2A/B satellites : A case study from Kenya

Sesay, Alhaji Mohamed

Titre : Mapping small-scale informal irrigation using time series of high-resolution data from Sentinel 2A/B satellites : A case study from Kenya

Auteur : Sesay, Alhaji Mohamed

Université de soutenance : UNESCO-IHE Institute for Water Education, Delft

Grade : Master of Science (MS) 2020

Résumé partiel
Over the years, small scale informal irrigation have been continuously neglected from a water management perspective. Undeniably, the water use per farmer may have minimal or has no abject impact on the source, but considered to have huge impact collectively. This study focuses on mapping small scale informal irrigation along the banks of Olkeriai river in Mashuru river basin which is located in Kajiado County in South west Kenya. The study area is semi-arid and largely depend on Sand Rivers/alluvial aquifers for its agricultural practices. With the rising effects of climate change and the increasing human population, it is high time we find ways and means to better manage our limited resources (land and water). Therefore to better understand the dynamics of the farming activities and its extent in the region, a remote sensing based approach was implemented to delineate small-scale irrigation extents from time series images of Sentinel 2A/B satellite data. The Sentinel data covering a period of three years from 2017 to 2019 together with ground truth data obtained from field work was used to extract the small scale irrigation extents of three years. Two approaches of supervised classifications (Object-based and Pixel-based) were applied to test their performance in delineating small-scale irrigated area extents. In this study, pixel based approach gave higher accuracy with less processing time. In both approaches, several machine learning approaches were applied (Random Forest, Support Vector Machine, K-Nearest Neighbour, and Decision Tree) and found that the Random Forest approach resulted in better results. Further, change analysis were carried out to understand the trend in small scale irrigation extent over three years. Metrics related to landscape ecology were applied to understand the spatial patterns of small scale irrigation and underlying processes behind those patterns. Three final Land use/Land cover maps were developed with extracted classes “Small-scale Irrigation, Riparian Forest, Water Source and Others”.

Sujets  : mapping remote sensing satellite data irrigation Kenya

Présentation

Page publiée le 22 avril 2021