Modeling Paddy Rice FPAR Using C-band Radar Remote Sensing Data

Fraction of absorbed photosynthetically active radiation(FPAR) shows the vegetation canopy energy absorption capacity. FPAR is an important biophysical factor in vegetation growth monitoring, a critical parameter in the terrestrial ecosystem modeling and a key indicator in global climate change study. Rice means life for millions of people and it is planted in many regions of the world. In China, 18.7 percent of grain acreage is planted with rice and the total area if rice-planted land has reached 29.63 million hectares. Rice is therefore of significant importance to food security and social stability in China. At the regional and global scale, to retrieve rice FPAR accurately and timely with remote sensing technology can provide important reference data and theoretical basis for monitoring the growth status of rice, estimating gross primary productivity and the absorption of carbon dioxide. Rice growth environment is warm and humid, so rice planting areas are mostly cloudy, foggy and rainy, which limits the application of satellite remote sensing in vegetation monitoring largely. Therefore, spaceborne radar with all-weather, around-the-clock earth observation and cloud penetration abilities, is not only an important tool for vegetation monitoring, but also an valuable way to establish reliable and stable vegetation remote sensing monitoring system in the future.In this study, the rice fields were mapped using mulita-temporal RADARSAT-2 image in Meishan, Sichuan Province. The FPAR and Leaf Area Index(LAI) were observed per 12 days during the whole growth. Combined with field observations of experimental data and other information, the rice FPAR experience and theoretical inversion model were developed. The FPAR of the study area were estimated in the paddy rice growth stage.(1) After the preprocessing of RADARSAT-2 remote sensing data, the paddy rice fields were mapped using the threshold method through analyzing the differences of multi-band and multi-temporal backscattering characteristics between rice and the other plants in growing season. The result shows the high accuracy in rice fields mapping by checking the study site, the neighboring ground objects and the coordinates of control points.(2) Based on the backscattering coefficients of three RADARSAT-2 images, FPAR data which obtained from 30 experimental sites in the study area during the growing season, the statistical relationship and FPAR experimental inversion model of the paddy rice were developed. Then, we chose three Radarsat-2 images in different time phases, extracted their backscattering coefficients in all polarizations of study sites and ratios and fitted it to the measured FPAR. Comparing the results, the σ0VV/σ0VH and measured FPAR also showed high correlation, and, the coefficient of determination(R2) reached to 0.713.(3) Based on the structural parameter data obtained from field experiments in every time periods during growing season of rice, the study simulated the backscattering coefficients in all polarizations of rice from every study sites with the improved MIMICS(Michigan Microwave Canopy Scattering) model and a semi-empirical backscattering model. After analyzing the backscattering coefficients of simulated experimental sites and the measured FPAR, we found that there were high correlations between the σ0VV/σ0VH simulated data of two models and FPAR. R2, the coefficient of determination, reached to 0.814 and 0.64.(4) The correlation analysis between C-band radar remote sensing data and the measured FPAR shows that the use of the whole growth period of radar remote sensing data to estimate rice FPAR is completely feasible and has good effect on the inversion. From the point of rice growing stage, in the early, because of the absorption of large water of radar signals and the influence of rice dwarf, lead to rice FPAR inversion effect is poorer; in the mid, geometric structure characteristics is distinctly and the influence of water is smaller, so the C-band radar remote sensing data works well on the inversion of FPAR in this stage; latterly, because of the influence of higher coverage and grain, so the C-band radar remote sensing data can’t be used alone for rice FPAR inversion in this stage.