Improvement of MODIS 8-day LAI/FPAR product with temporal filters to generate high quality time-series product

Abstract

Numerous studies have reported that the time-series terrestrial parameters such as the Normalized Difference Vegetation Index (NDVI), Leaf Area of Index (LAI), Fraction of Absorbed Photosynthetic Active Radiation (FPAR), derived from NOAA/AVHRR, SPOT/VEGETATION, TERRA or AQUA/MODIS, have played significant roles in researching the global environment, terrestrial ecosystems and related ecological researches. However, the remotely sensed signals are interfered severely by atmospheric conditions especially clouds and such noises exists in the time-series products as well. Therefore, to obtain a high quality time-series of terrestrial parameter is a necessary step before further studies. At present several methods have been applied to reduce the noise to construct a fine time-series of NDVI, but few studies concerning the other key terrestrial parameters, such as LAI, FPAR etc. In this paper, after comparing general methods in literatures, we designed a new method based on the Savitzky-Golay filter, which was applied to improve the quality of MODIS 8-Day LAI/FPAR Product to generate time-series of LAI and FPAR with high quality. Our validation results indicate that more smooth and realistic time-series curves of LAI/FPAR can be obtained by using this new method, which exemplify the dynamic change of forests, crop or plants and key input parameters in modeling the complex land surface processing.