Vegetation Monitoring Based On Full Polarization SAR

Land resource monitoring plays an important role in rational allocation of resources,promotion of economic development and ecological environment protection.As one of the important components of land resources,it is more necessary to obtain vegetation change information accurately and timely.Synthetic Aperture Radar(SAR),as one of the active remote sensing technologies,has all-day and all-weather imaging capabilities,and has gradually become one of the research hotspots and effective observation methods for vegetation monitoring.With the continuous development of radar technology,fully polarized data can provide information on vegetation growth and morphological structure characteristics,providing critical data for vegetation monitoring.Based on multi-temporal full-polarization SAR data,this paper monitors the main vegetation types,rice and forests,in the area of Jinhu County,Jiangsu Province.The research mainly on the identification and classification of forests,the classification of rice and the inversion of key phenological periods.The main research contents of this paper are as follows:(1)Based on the time series full-polarization SAR data and the sample points obtained from the field ground synchronization measurement,the polarization characteristic parameters of rice and forest were extracted respectively,and the radar response characteristics and polarization characteristic parameters caused by rice and forest were analyzed.(2)Time-domain analysis of polarization parameters of rice and forest,and selecting of polarization characteristic parameters sensitive to their growth changes.Based on the differences between different vegetation polarization parameters,the decision tree algorithm was used to identify rice and forest,and the indica and japonica,poplar and fruit trees in the study area were identified.The mapping precisions of indica and japonica rice were 98.54% and 92.12%,respectively,and the mapping precisions of fruit trees and poplars were 97.35% and 96.77%,respectively,and high-precision classification results were obtained.(3)Using the polarization characteristic parameters to construct a radar phenological index(RPI)which can reflect the key phenological changes of rice growth,and used Savitzky-Golay(S-G)filter to reconstruct the time series RPI.The variation of RPI with rice growth and the RPI characteristics of key phenological periods of rice were analyzed.The curve derivative method is used to extract the characteristic points(the max value points,etc.)of the reconstructed RPI time series curve,and the key phenological period of rice is identified,the recognition results are analyzed and verified.The results showed that the identified key phenological date of rice and the date of field survey experiment were all within 16 days,indicating that RPI can accurately identify the key phenology of rice.Based on the above research results,it is indicated that the fully polarized SAR data can identify vegetation types and monitor the growth of vegetation with high precision.Fully polarized SAR data has great potential for vegetation monitoring applications.