Abstract:In order to effectively integrate various land-surface processes and environmental variables retrieved by different spatial resolution remotely sensed data, scaling issues induced by transferring for one resolution data to another have drawn more and more attentions. Taking the Zhangye Oasis in the Heihe River Basin of northwestern China as a study site, using the LAI map derived from the Landsat ETM+ image and NASA published MODIS LAI product for the same day of the region, an experimental study for up scaling the 30m spatial resolution LAI to 900m was conducted to quantitatively analyzes the error transforms in the process of LAI up scaling with the final goal of establishing a new and much effective up scaling scheme in terms of NDVI pixel decomposing techniques. It was found that less than 10.5% errors were caused by the non-linearity of algorithms utilized in LAI up scaling, and its value is highly positively correlated with the terrain heterogeneities, terrain heterogeneity is the predominate causing for the errors introduced in LAI scale transferring. The maximum error caused by terrain heterogeneity can reach to about 45%, but no error can be caused for the pure pixel LAI scale transferring. A new algorithm for LAI up scaling, based on above mentioned analyses, in terms of pixel decomposing techniques of NDVI accounting for terrain heterogeneity in two key procedures, was developed for the up scaling of 30 m LAI to 900 m with fairly satisfied resolution. This algorithm was used to correct NASA published MODIS LAI product and excellent results were obtained, which was back up with the fact that the corrected MODIS LAI product well consistent with the high spatial resolution LAI map of the study region.