Abstract:

Tobacco is one of the important cash crops in China. Accurate extraction of tobacco planting area is of great significance to the adjustment of crop planting structure and scientific management and monitoring of tobacco fields. Based on GEE (Google Earth Engine) cloud platform, the authors used Sentinel-2A multi-temporal data from February to November of 2021 to exclude error factors according to tobacco planting characteristics, and combined with the distribution characteristics of field sampling data to conclude the vegetation characteristics. Using the threshold method, Linyi City of Shandong Province was taken as an example. Study on automatic extraction of tobacco planting plots was carried out. The results showed that the combined model of GEE and random forest algorithm could accurately obtain the spatial distribution information of tobacco planting field, and the overall classification accuracy was 96.04%. The classification of this research method is of high quality, which can meet the research requirement and provide accurate, quick and scientific reference for the tobacco planting field management department.

Key words: tobacco, planting area, Sentinel-2A multi-temporal data, random forest, Google Earth Engine