Rice-crawfish map in China

Integrated aquaculture systems, that is, systems involving the cocultivation of aquatic food with crops, provide high protein and energy outputs per unit of land and raise farm incomes. However, they can also cause substantial environmental problems, such as water and soil pollution. We used time series of satellite images to map the evolution of a new farming system integrating paddy rice production with crawfish cultivation from its implementation in 2013 to 2020 in five provinces (805,600 km2) in China. We used an object-based water difference method that integrates pixel-based identification with object-based segmentation. We first mapped the distribution of the water bodies by using the automated water extraction index and then extracted the integrated rice-crawfish area at the pixel level based on changes in the presence of water between the fallow and rice planting periods. We used the noniterative clustering algorithm in GEE to segment the Landsat images. Finally, we intersected the segmentation results with the rice-crawfish pixels and calculated the percentage of crawfish pixels within each image object using 50 % as the threshold for rice-crawfish detection. The object-based water difference method captured integrated rice-crawfish production only when there was a change in the presence of water between the fallow and planting periods. Rice-crawfish farming systems that were not associated with a change in the presence of water could not be identified.综合养殖系统（integrated aquaculture systems），即水产食品与农作物复合种养系统，可在单位土地上产出高蛋白与高能量的农产品，并提升农户收入。然而此类系统亦可能引发严重的环境问题，如水体与土壤污染。本研究借助卫星影像时间序列，对中国5个省份（总面积80.56万平方千米）内2013年至2020年间稻虾综合种养这一新型农业模式的演化过程进行了遥感制图。本研究采用了融合像素级识别与面向对象分割（object-based segmentation）的面向对象水体差异法（object-based water difference method）：首先通过自动水体提取指数（Automated Water Extraction Index, AWEI）绘制水体分布，随后依据休耕期与水稻种植期的水体存在变化，在像素级别提取稻虾综合种养区域；本研究借助谷歌地球引擎（Google Earth Engine, GEE）中的非迭代聚类算法对Landsat影像（Landsat images）进行分割。最后将分割结果与稻虾种养像素进行空间叠加，以50%作为稻虾种养识别阈值，计算每个影像对象内虾田像素的占比。该面向对象水体差异法仅当休耕期与种植期存在水体变化时，才能识别稻虾综合种养区域；未伴随水体变化的稻虾种养系统则无法被该方法识别。