LAI Retrieval from CYCLOPES and MODIS Products using Artificial Neural Networks

In this paper, an artificial neural network approach to estimate LAI from the combination of CYCLOPES and MODIS products over the 2001 to 2003 period is described in detail. Reflectances in RED, NIR and SWIR band and LAI with good quality were chosen according to the Quality Control information and the temporal consistency between the two LAI products. Four different reflectance and LAI combinations from both sensors were used as the input and output variables of the ANNs with different land cover types for training. The prediction abilities of the trained ANNs were validated using the datasets which were not used in the training process. It is observed that the ANNs can be well trained and have promising prediction abilities. The time series LAI derived from the trained ANNs is charactered by better temporal consistency compared with the original MODIS LAI product.