3.1. Computing LAI from GF-1 and HJ-1 Data

Obtaining the LAI from remote sensing images is an important step in the data assimilation process in this article because LAI is an important input parameter for the WOFOST model. The correct inversion of LAI is critical to monitor crop growth. Many studies have shown that the use of the normalized difference vegetation index (NDVI) inversion of LAI is undesirable because the sensitivity of NDVI to LAI decreases and saturation occurs when the value of LAI is higher than 2 or 3 [48,49]. To avoid this problem, many researchers have conducted blue NDVI (BNDVI), green NDVI (GNDVI), blue-green NDVI (GBNDVI), red-blue NDVI (RBNDVI), green-red NDVI (GRNDVI), and green-red-blue (PNDVI) tests, and these tests found that GBNDVI and GNDVI had high accuracy in retrieving LAI [48]. It also has been found that, for HJ-1 images, the use of the inversion of LAI from GNDVI is better than that from GBNDVI [50]. In addition, we did not use the blue band, because the blue band is strongly influenced by atmospheric scattering. To control the variables, whether it was a GF-1 or HJ-1 image, this paper used GNDVI to invert the LAI. The GNDVI and LAI formulas [51] are as follows:

where ρ_NIR and ρ_G are the reflectance values of the near-infrared and green bands, respectively. In this paper, we calculated the four different resolutions of GNDVI and their corresponding LAI values.