Forest leaf area index estimation using combined ICESat/GLAS and optical remote sensing image

Based on Gaussian decomposition of the geoscience laser altimeter system( GLAS) waveform, accurate waveform characteristics were extracted, and then laser penetrate index (LPI) was computed for each GLAS waveform. The new method of leaf area index (LAI) estimation using LPI derived from GLAS data was proposed. Forest LAI estimation model based on GLAS data was established( R-2 =0.84, RMSE = 0.64) and the model's reliability was assessed using the Leave-One-Out Cross-Validation (LOOCV) method. The result indicates that the regression model is not overfitting the data and has a good generalization capability. Finally, regional scale forest LAI was estimated using combined GLAS and TM optical remotely sensed image by artificial neural network. And then, the accuracy of the predicted LAIs based on neural network was validated using the other 25 field-measured LAIs. The results show that forest LAI estimation are very close to the field-measured LAIs with a high accuracy (R-2 =0. 76, RMSE =0. 69). Therefore, the estimated LAIs provide accurate input parameters to the study on ecological environment. The study provides new methods and ideas to estimate LAI with large regional scale using GLAS waveform data.