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

Titre : Geospatial Modeling for Drought Risk and Hazard Using Earth Observations and Multiscale Indicators

Auteur : Pandey, Varsha

Université de soutenance : Banaras Hindu University

Grade : Doctor of Philosophy (PhD) in Environmental Science 2021

Résumé partiel
Understanding various aspects of drought is essential due to its intricate relationship with socioeconomic and environmental conditions. Drought assessment and projection are challenging but a prerequisite task for minimizing risk and vulnerability and developing appropriate management frameworks and strategies. India is amongst the most vulnerable and drought-prone countries in the world. Recurrent droughts are also a significant problem in the Bundelkhand region of Uttar Pradesh, located in the Indo Gangetic plain and has fertile cropland with immense potential. This region is a data deficit one, with complex and diverse topography where drought monitoring primarily depends on limited gauge site data. The drought characterization is of utmost importance for preventive measures, adaptations against similar droughts in the future, and developing unified management policies for the vulnerable areas. The current study explores the suitability of remote sensing data and hydrological models for near real-time drought mapping and their long-term monitoring. Moreover, a novel approach has been proposed that combines multiple drought indicators and assess their effect on crop yield using Machine Learning techniques to fulfill the following research gaps :

1. The point-based observation data limits Spatio-temporal analysis of precipitation and therefore constrains effective drought monitoring. Alternatively, the high resolution satellite precipitation (integrated with the in-situ measurements) data products have been proven an effective alternative data source for Spatio-temporal drought analysis. Several attempts were made to compare satellite precipitation products with ground observed data at regional, continental, and global scales. However, a similar assessment at the local scale is inadequate, leading to uncertainty while satellite-based precipitation data products are used.

2. Agriculture drought assessments based on soil moisture are well studied in India. Such assessments primarily rely on point-based data and various Earth Observation (EO) estimates. Although the EO data overcomes Spatio-temporal contiguity and data inaccessibility constraints, long-term and root zone level Soil Moisture (SM) data are still lagging. The root zone SM data must put forth the antecedent conditions, which indicate the potential and available water storage for the plants. Moreover, the sub-surface SM is relatively more stable than the surface SM, which is dynamic and highly sensitive to various climatic variables. On the contrary, Land Surface Models (LSMs) or hydrological models employ long-term climate variables as inputs and provide SM data at frequent intervals for surface and surface soil layers, even at a daily scale. The use of such models for drought assessment is less explored over the Indian region.

Présentation et version intégrale (Shodhganga)