Estimating time-series leaf area index based on recurrent nonlinear autoregressive neural networks with exogenous inputs

The leaf area index (LAI) is a key parameter in many meteorological, environmental and agricultural models. At present, global LAI products from several sensors have been released. These single sensor-based LAI products are generally discontinuous in time and cannot characterize the status of natural vegetation growth very well. In this study, by fusing Moderate Resolution Imaging Spectroradiometer (MODIS) and Satellite Pour l'Observation de la Terre (SPOT) VEGETATION products, time-series LAIs were used to train recurrent nonlinear autoregressive neural networks with exogenous inputs (NARXNNs) for six typical vegetation types. The exogenous inputs included time-series reflectances in the red, near-infrared and shortwave infrared bands as well as the corresponding sun-viewing angles. These NARXNNs subsequently served to predict the time-series LAI. The validation results show that the predicted LAI of the NARXNN is not only more continuous and stable than the MODIS LAI as a function of time but is also much closer to the ground truth. Thus, the proposed method may be helpful for improving the quality of the MODIS LAI.