Bias analysis and validation method of the MODIS LAI product

Abstract

An assessment of the MODIS Leaf Area Index (LAI) product shows a clear focus on the consistency in value and temporal trend between remote sensing products and the "ground truth." However, few studies have comprehensively analyzed the bias sources and quantified the contribution of different biases on the global deviation. The current study evaluates the MODIS LAI product by analyzing the MODIS LAI bias in terms of three aspects: algorithm, reflectance data, and clumping effect. We then quantify the individual influence. The bias analysis and validation of the MODIS LAI product are conducted by utilizing measured data of corn plants in a field in Huailai County, Hebei Province. Results indicate an evident underestimation of the MODIS LAI by as much as 34.14% in this area. The mean LAI values of the reference LAI and MODIS LAI are 3.53 m2/m2 and 2.33 m2/m2, respectively. The reflectance data in the bias analysis has the most significant effect on the total deviation. The bias caused by the difference between the MODIS reflectance and Landsat 8 OLI reflectance is 57.50% of the total; the clumping effect accounts for 28.33% of the total; and the bias caused by algorithm is the smallest at only 14.17% of the total. The proposed method positively influences the validation of the remote sensing product and uncertainties analysis.