Smart agriculture data analysis approach for drought, food and water security using satellite data

This thesis proposes a smart agriculture data analysis approach that integrates satellite imagery and meteorological data to address pressing global challenges such as drought, food security, and water security. In the context of escalating climate change and its growing impact on agricultural productivity, the development of early warning and decision support systems has become crucial. The proposed system employs Sentinel-2 satellite data to calculate spectral indices related to soil moisture and vegetation health, and applies machine learning algorithms such as RandomForestRegressor to generate spatial predictions. The study introduces a dual-stream architecture. The first stream focuses on analyzing satellite data to assess agricultural suitability and environmental stress, while the second stream processes meteorological data retrieved from the OpenEO platform to provide user-specific insights and recommendations. One of the system's core features is its ability to perform real-time, location-based analyses defined by user-input coordinates, enabling adaptive and flexible decision-making. Additionally, three deep learning models were developed to assess drought risk (DroughtNet-X), water security (WS-RiskNet), and food security (FoodNet-C). These models were trained and evaluated using Python in a Google Colab environment, and their performance was benchmarked against conventional methods in the literature. The findings demonstrate that the combined use of satellite and meteorological data significantly enhances the effectiveness of agricultural decision-making processes and contributes to the resilience of food systems under environmental stress. This study offers a comprehensive and practical framework that is both scientifically rigorous and applicable to real-world agricultural management scenarios.