| Popis: |
The war in Tigray, Ethiopia has triggered a massive humanitarian crisis, displacing millions. Yet, the impact on cultivated land and local food production remains poorly understood, impeding effective aid. Leveraging Sentinel-2 satellite imagery and a decision tree algorithm with Normalized Difference Vegetation Index (NDVI) time series, we developed a model to map well-cultivated cropland, defined as fields judged by field surveyors to have satisfactory to optimal crop condition in 2021–2022 field observations. Assessing satellite estimated well-cultivated land in highland croplands (> 1200 m a.s.l), we found a net loss of 543 sq. km (95% CI: 81 sq. km) well-cultivated land in highland croplands equivalent to ≈ 8% of the average total surveyed cropland estimate from Central Statistical Agency between 2015 and 2019 (ESS, 2023a) in potential highland cropland. The net loss was positively associated with the density of recorded conflict incidents and sub-regions with high numbers of internally displaced persons (IDPs), consistent with a causal effect of the conflict on cultivated land.Employing a two-way fixed effect model causal analysis with rainfall covariates, we quantified the impact of conflict incidents on cultivated land during the pre-war (2019/20) and in-war (2021) periods. Results indicated a ≈ 6.17 sq. km (SE: 2.06) additional loss per unit increase in conflict incidents during the growing season (June to October), eight times higher than total incidents occurring throughout the entire study period. We estimated the kilocalories lost due to loss of well-cultivated croplands in 2021 could have supported at least 90% of all recorded IDPs in Tigray as of June 2021, discounting for Western Tigray. Our study showcases the utility of satellite data, coupled with local agricultural knowledge, for timely and cost-effective information crucial for aid agencies and long-term rehabilitation initiatives in smallholder farming contexts. |
