Mapping crop type in Northeast China during 2013–2021 using automatic sampling and tile-based image classification

Abstract

Northeast China is one of the most major grain banks in China and has an overwhelming influence on food security. To mitigate the challenges caused by increasing food demands and soil protection, crop rotation and fallowing policies have been introduced in Northeast China. These soil protection policies change annual crop planting area and crop distribution. To monitor crop type and its changes on a regional scale in time series, we explore the automatic sampling approach by hexagon strategy and tile-based classification by random forest (RF) algorithm using time-series Landsat-8 Operational Land Imager (OLI) images during 2013–2021. The crop maps have high credibility with the overall accuracies (OA) wall-to-wall ranging from 0.89 to 0.97, and also have close agreement with statistical data city by city. This study provides a highly reliable long-term crop maps dataset, which can be helpful for food security and regional agricultural production management.

Keywords

• Crop mapping
• Google Earth Engine
• Time-series
• Tile-based classification
• Random forest