Simultaneous Estimation of LAI and Dynamic Model Parameters Using Dual EnKF from Time Series MODIS Data

Accurate estimates of leaf area index (LAI) from remote sensing data is very important for many climate and carbon models. In this study, a new algorithm is developed to online update model states and parameters to estimate LAI from time series MODIS data. Kuusk model is chosen as observation operator to link LAI with reflectance data, and dual Ensemble Kalman filter is used to estimate parameters and state variables of dynamic model when observation data are available. The method is tested with MODIS reflectance data (MOD09A1) at some BigFoot sites. We refer to ETM+ LAI imageries which are consistent with the field LAI measurements. Then results are compared with MODIS LAI and LAI time series from EnKF, which shows that the new method has improved LAI estimation significantly.