Global Cropland Extent全球250米耕地范围数据集（2000-2008）

该数据集中的数据获取自马里兰大学地理科学系的全球土地分析和发现 (GLAD) 实验室生产的Global Cropland Extent数据产品。该数据产品利用 250m MODIS（中分辨率成像光谱仪）数据绘制全球生产农田范围图，用一组包含四个 MODIS 陆地带、NDVI（归一化差异植被指数）和热数据的多年 MODIS 指标来描述2000至2008 年期间的农田物候，训练生成一组全局分类树模型，从而产生全局每像素农田概率层。然后使用来自 USDA-FAS生产、供应和分配 (PSD) 数据库的数据对概率产品进行阈值处理，以创建离散的农田/非农田指标图。随后使用五种全球土地覆盖分类对全球 MODIS 农田范围图进行区域评估。参考四种主要的全球粮食作物：玉米、大豆、小麦和水稻，进一步研究了全球概率层。总体结果表明，MODIS 图层表征最好的是密集的阔叶作物生产区域（玉米和大豆），既与现有地图相对应，也与相关的高概率匹配阈值一致。小麦产区的概率阈值较低，而水稻产区的相关置信度最低。没有农业集约化的地区，如非洲，无论作物类型如何，都很难表征。

The data in this dataset is sourced from the Global Cropland Extent data product produced by the Global Land Analysis and Discovery (GLAD) Laboratory, Department of Geographical Sciences, University of Maryland. This data product uses 250m Moderate Resolution Imaging Spectroradiometer (MODIS) data to map global cropland extent. It employs a set of multi-year MODIS metrics including four MODIS land bands, Normalized Difference Vegetation Index (NDVI), and thermal data to characterize cropland phenology during the period 2000–2008, and trains a set of global classification tree models to generate a global per-pixel cropland probability layer. Subsequently, thresholding was applied to the probability product using data from the Production, Supply and Distribution (PSD) Database produced by the USDA-Foreign Agricultural Service (USDA-FAS) to create a discrete cropland/non-cropland indicator map. Regional validation of the global MODIS cropland extent map was then conducted using five global land cover classifications. The global probability layer was further investigated with reference to four major global staple food crops: maize, soybean, wheat, and rice. Overall results indicate that the MODIS layer performs best in characterizing dense broadleaf cropland production areas (maize and soybean), which aligns with both existing maps and the associated high-probability matching thresholds. Cropland probability thresholds are lower for wheat-producing regions, while the corresponding confidence levels are the lowest for rice-producing areas. Regions without agricultural intensification, such as Africa, are difficult to characterize regardless of crop type.

搜集汇总

背景与挑战

背景概述

全球250米耕地范围数据集（2000-2008）基于MODIS数据生成，覆盖全球范围，重点关注四种主要粮食作物的耕地分布。数据集通过多年度MODIS指标和分类树模型生成耕地概率层，特别适用于密集阔叶作物生产区的表征。