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

Titre : Remote sensing drought variability across different selected biomes of South Africa

Auteur : Diza, Duduzile.

Université de soutenance : University of KwaZulu-Natal

Grade : Master Environnemental Science 2022

Résumé
Drought has been recognized as the leading extreme weather event in Southern Africa which causes major impacts on terrestrial environments, water resources and general functioning of the society across the region. Southern African biomes are increasingly getting vulnerable to drought conditions due to climate change, affecting vegetation health and biome distribution. Establishing the major effects of drought conditions on vegetation health at a regional scale is intricate, considering landscape variability with changes in climatic conditions, elevation, and soil dynamics. Previous research studies on drought assessment and monitoring have applied meteorological drought indices derived using point data from ground weather stations. However, the allocation of ground weather stations is commonly limited, hence, these meteorological drought indices are point limited and lack precise spatial coverage and accuracy in evaluating and monitoring the spatial distribution of drought periods at regional scale. Hence, vegetation indices derived from remotely sensed data have proven to be an effective method to track changes in vegetation over time and at a comparatively large scale. In this regard, this study sought to i) evaluate the relative performance of conditional and combinative drought indices in quantifying the magnitude of drought across different major biomes of Kwa-Zulu Natal. ii) assess the impact of various moisture and temperature contribution coefficients in estimating drought severity across different vegetation types. The findings of this study illustrated that the drought magnitude across the country could be optimally estimated using combinative drought indices to an R2 and RMSE of 0.98 and 0.074 for the Savanna biome, 0.93 and 0.013 for the Grassland Biome and 0.99 and 0.016 for the Forest biome. The optimal index in this model was the Vegetation Moisture Stress Index (VMSI). In assessing the impact of various moisture and temperature contribution coefficients in estimating drought severity, high temperature predictor variables yielded the highest accuracies with VMSI_2 model (90%TCI:10%VMCI) being the most accurate model. The results produced an R2 and RMSE of 0.78 and 0.0589 for the Savanna biome, 0.66 and 0.0511 for the Grassland biome and 0.76 and 0.0034 for the Forest biome. These findings demonstrates the potential of new combinative drought indices that combines multiple drought factors in effectively quantifying and monitoring drought conditions across different vegetation types of Southern Africa.

Mots clés  : Drought, Biomes, Conditional Indices, Combinative Indices, Kwa-Zulu Natal.

Présentation

Version intégrale 2,7 Mb)