A Study Of Normalization Of Directional Effects In Vegetation Index Using BRDF Of Typical Crops Form MODIS Data

Vegetation Indies (VI) have been used to estimate crop yield and monitor crop growth for many years. In order to minimize the cloud effect and other influences, we often composite vegetation index by combining multiple days of VI, angular and quality assurance data in certain manner. In the past, the maximum VI value composite (MVC) approach is mainly used to composite VI for AVHRR Data. MODIS vegetation composition algorithm uses the empirical BRDF model (developed by Walthall et al) to normalize the Sun/View angles to certain angle, and then composite the VI by several day's data. These methods are effective in monitoring globe vegetation. However, the simple BRDF model brings some errors, which ought to be avoided. Our work aims to improve the level of estimation crop yield and monitoring crops growth. In this paper, we present a new method to normalize vegetation index on pure pixels of crops, which can be recognized from MODIS image by high resolution land cover map. Firstly, some different BRDFs of typical crops (refer to winter wheat in this paper) are simulated by radiative transfer models using the plant canopy parameters obtained from field measurements in typical growth stages. Between the two typical growth stages, the canopy parameters are produced by crop growth model. Then we normalize the VI based on the simulated BRDFs. We used our method to get MODIS normalized VI of North China Plain. It shows our results are good consistent with the ground based measurements and our method is more effective to monitor the crop growth in regional scale.