Table_2_Satellite Data Reveal Cropland Losses in South-Eastern Ukraine Under Military Conflict.XLSX

While people are aware that there is a continuing conflict in Ukraine, there is little understanding of its impact. The military conflict in South-Eastern Ukraine has been on-going since 2014, with a major socio-economic impact on the Donetsk and Luhansk regions. In this study, we quantify land cover land use changes in those regions related to cropland changes. Cropland areas account for almost 50% of the Donetsk and Luhansk regions, and with the declining industry between 2014 and 2017, the role of agriculture to the regional economy has increased. We use freely available satellite data and machine learning methods to map cropland extent in 2013 and 2018 and derive corresponding changes in cropland areas. We use a multi-layer perceptron (MLP) to classify multi-temporal Landsat-7, Landsat-8, and Sentinel-2 images into cropland and non-cropland areas, and a sampling-based approach to estimate the areas of cropland change. We found that net cropland losses were not uniform across the regions, and were more substantial in the areas not under control of the Ukrainian Government (22% of net cropland area loss compared to cropland areas in 2013) and within a buffer zone along the conflict border line (46%), where combat activities occur. These results highlight the impact of the conflict on agriculture and the utility of spatially explicit information acquired from Earth observation satellites, especially for areas, where collecting ground-based data is impractical.

尽管公众普遍认识到乌克兰持续存在冲突，但对于其影响却知之甚少。自2014年起，乌克兰东南部的军事冲突持续不断，对顿涅茨克和卢甘斯克地区产生了重大的社会经济影响。在本研究中，我们量化了这些地区与耕地变化相关的土地利用覆盖变化。耕地面积约占顿涅茨克和卢甘斯克地区的50%，随着2014年至2017年间工业的衰退，农业在区域经济中的作用有所增强。我们利用免费的可获取卫星数据和机器学习方法，绘制了2013年和2018年的耕地范围，并推导出相应的耕地面积变化。我们采用多层感知器（MLP）将多时相的Landsat-7、Landsat-8和Sentinel-2图像分类为耕地和非耕地区域，并采用基于抽样的方法估计耕地变化面积。我们发现，耕地净损失在区域间并不均匀，在乌克兰政府控制之外的区域（与2013年耕地面积相比，耕地净损失率为22%）以及冲突边境线沿线的缓冲区（损失率为46%），即发生战斗活动的区域，损失更为严重。这些结果突显了冲突对农业的影响以及从地球观测卫星获取的空间明确信息的实用性，特别是在难以收集地面数据的地区。