Monitoring of Sugarcane Crop based on Time Series of Sentinel-1 Data

Abstract

Monitoring the spatial pattern and growth of sugarcane timely and accurately is of great importance at regional and global scales. In this paper, the focus was on sugarcane identification in Southern China with FuSui country as the study area. Classification was based on sentinel-1 different polarizations and sugarcane phenology. In order to explore the optimum periods and polar metric characters, time series of C-band dual polarization sentinel-1 data in 2017 totally 130 images were collected over the whole sugarcane growth season. Then the growth curve was built based on the former exploration. After that, there was a following analysis by combining growth curve and polarimetric characters of sugarcane, which contributes to setting attribute to identity. At last, the advanced rules were built to identify sugarcane according to growth curve above and subordinating degree function. Sugarcane extraction accuracy was verified by numerous ground data. The conclusions are as follows: (1)The results of this study show the importance of using C-band muti-temporal dual polarization data on crop identification especially for sugarcane comparing with traditional optical data. In other words, it's crucial for crop identification to extract the backscattering coefficient. When combining with a part of samples, the curve of crop growth used for classification can be portrayed. To deepen the difference between sugarcane and other typical features, additional three kinds of reference object like eucalyptus, water and buildings, all of which distributes in the experimental area, with an extensive representation. (2)The analysis of polarimetric characters has shown that the inherent SAR backscatter feature VH is superior in classification accuracy to the VV, which achieved an accuracy of 88.07%. During the stage of seedling and tillering, the amplitude from sugarcane is higher than that in other objects, proving the advantage of VV in sugarcane identification. On the contrary, the giant grass and aiphyllium appearing stable in sequential variation, corresponding banana and eucalyptus respectively. (3)Moreover, the sugarcane has shown strong difference in March when it comes to the optimum periods, the data is more sensitive to the change of sugarcane. There was an evidently reduction as time goes by, so choosing the data from March makes higher accuracy. Therefore, the data from March with the polarimetric character VH was used as the optimum periods.