基于Google Earth Engine和NDVI时序差异指数的作物种植区提取

为提高农作物种植信息遥感监测的效率,扩展数据适用范围,本文提出了一种基于时间序列NDVI差异指数的作物种植区提取方法.随着海量遥感与云计算的发展,Google Earth Engine作为一个全球尺度地理空间分析云平台,弥补了单机计算耗时长的不足,为快速遥感分类带来了新机遇.基于Google Earth Engine平...

Crop Planting Area Extraction based on Google Earth Engine and NDVI Time Series Difference Index

In order to improve the efficiency of remote sensing monitoring of crop planting and expand applications of remote sensing data, a method of crop planting area extraction based on NDVI time series difference index is proposed. With the development of remote sensing and cloud computing technologies, Google Earth Engine, as a global-scale geospatial analysis cloud platform, overcomes the disadvantages of traditional single-machine computing and brings new opportunities for rapid remote sensing classification. In this study, taking Qi County in Henan province as the study area, the NDVI time series difference index of different crops is constructed according to the characteristics of time series NDVI curve of each crop to extract crop planting information and distinguish different crop types using multi-temporal Sentinel-2 images in 2019-2020 based on the Google Earth Engine platform. The extraction accuracy is verified and compared with other existing methods. The results show that the NDVI time series difference index is based on crop phenology information and developed using GEE's high-performance computing capability, which forms a framework for rapid crop planting information extraction and has obvious advantages over traditional local computing. The winter wheat and garlic planting areas in Qi County have obvious spatial variation. The winter wheat planting areas are mainly concentrated in the northwest and southern rural residential areas of the study area. While the garlic in Qi County is mainly concentrated in the central and northeastern part of the study area due to the needs of transportation. Compared with other methods using support vector machine and maximum likelihood, the overall accuracy of crop planting area extraction using the NDVI time series difference index reaches 83.72%, and the Kappa coefficient is 0.67. The overall accuracy and the Kappa coefficient are 10.02% and 0.21 respectively higher than the maximum likelihood method, and are 4.18% and 0.09 respectively higher than the support vector machine method, which indicates that our method can extract crop planting information with high efficiency and high accuracy. We develop an efficient and accurate monitoring method for regional crop planting information extraction and expand the application of remote sensing data in the agricultural field, which has significant value for future agricultural applications.