Synergy_Cropland_Mapping_Code

SOSA is based on the fusion of multiple existing cropland maps and multilevel statistics of the cropland area, which is independent of training samples. Firstly, cropland area statistics are used to rank the input cropland maps, and then a scoring table is built to indicate the agreement among the input datasets. Secondly, statistics are allocated adaptively to the pixels with higher agreement scores, until the cumulative cropland area is close to the statistics.

SOSA数据集基于多份现有耕地地图与耕地面积多级统计数据的融合构建，且不依赖训练样本。首先，通过耕地面积统计数据对输入的耕地地图进行排序，随后构建评分表以表征各输入数据集间的一致性程度。其次，将统计数据自适应分配至一致性评分更高的像素，直至累计耕地面积趋近于该统计值。