Biomass Mapping The set of field biomass training data and the MODIS observations were used to develop a regression tree model (Random Forest). Biomass predictions for the entire area were then generated by incorporating reflectance measurements from the first seven MODIS spectral bands (the land bands) into the regression tree model, effectively extending the model based on field training data to the entire region. Integration of GLAS Height Product and MODIS Observations for Biomass Mapping and Validation in Central Africa Alessandro Baccini, Scott J. Goetz, Mindy Sun, and Nadine Laporte The Woods Hole Research Center, Falmouth, MA Abstract Observations from the Moderate Resolution Imaging Spectroradiometer (MODIS) were used in combination with a large data set of field measurements in a tree-based model to map woody above-ground biomass (AGB) across tropical Africa. Analysis of lidar metrics from the Geoscience Laser Altimetry System (GLAS), which are sensitive to vegetation structure, were used to validate the biomass estimates and to investigate the relationship between lidar metrics and above-ground biomass. The results indicate that the model successfully stratified the landscape across the full range of biomass classes. The results also showed a strong positive correlation between the GLAS height of median energy and predicted AGB. Lidar Waveform and Biomass We geolocated the GLAS foot prints from the year 2003 and derived a set of lidar metrics for foot prints within MODIS pixels where above-ground biomass field measurements were available. We explored the waveform characteristics in relation to the pixel average biomass. To assess the accuracy of the predictions, a subset of the field data not used in model development were reserved for a cross-validation analysis. Following common protocol for data-intensive regression tree models, we used 10% (154 samples) of the field data, which were extracted using a random sampling design. The Random Forest model estimated using MODIS spectral reflectance proved effective for predicting above ground biomass (Fig. 2). The tree model explained 82% of the variance in above-ground biomass density, with a root mean square error (RMSE) of 50.5 Mg/ha. Figure 2. Above ground biomass derived from MODIS data and field measurements [Baccini et al., 2008]. Comparison with Lidar Metrics The analysis of the GLAS data showed a strong positive relationship between Random Forest predicted biomass aggregated in classes of 10 Mg/ha and the average vegetation height (r=95) and the ratio of HOME and height (r=95) (Figure 2). Conclusions We mapped above-ground biomass over tropical Africa using multi-year MODIS satellite observations and a wide range of field measurements. The results indicate that the MODIS data sets, used in a cross-calibrated regression tree model, captured the amount and spatial distribution of above-ground biomass across tropical Africa. Comparison with GLAS LIDAR height metrics, showed strong positive correlations with the mapped MODIS biomass density values. We are now conducting analyses using MODIS and GLAS data fusion. Acknowledgements This work was funded under NASA contract number G05GD14G and NNS06AA06A, the Roger and Victoria Sant, Joseph Gleberman, and The Linden Trust for Conservation Background Deforestation contributes about one fifth of total anthropogenic CO2 emissions to the atmosphere [Houghton, 2007]. Refining these estimates requires improved knowledge of the density and spatial distribution of forest biomass. Remote sensing has been extensively used as a basis for mapping forest structure and above-ground biomass and most recentely LIDAR (light detection and ranging) remote sensing has been used to successfully characterize vegetation vertical structure and height, and to infer AGB [Lefsky et al., 2005; Drake et al., 2002]. Figure 1. Mosaic of MODIS NBAR data. The black dots show about 30 % of the GLAS L2A (year 2003) shots after screening procedure and used in the comparison analysis with the predicted biomass. The Woods Hole Research Center References Baccini A., N. Laporte, S. J. Goetz, M. Sun, and H. Dong. A first map of Tropical Africa’s above- ground biomass derived from satellite imagery, (forthcoming, 2008). Lefsky M A, Harding D J, Keller M, Cohen W B, Carabajal C C, Espirito-Santo F D B, Hunter M O & de Oliveira R 2005 Geophysical Research Letters 32, 1–4. Drake B J, Knox R G, Dubayah R O, Clark D B, Condit R, Blair J B & Hofton M 2003 Global Ecology and Biogeography 12, 147–159. Data The GLAS instrument on board the Ice, Cloud, and Elevation Satellite (ICESAT) is a waveform sampling lidar sensor designed for global observation of the Earth. Lidar metrics have been extensively used to characterize vegetation structure. In this work we used about 1.3 million observations recorded from GLAS Laser 2 (L2A) between Oct-Nov 2003 (Figure 1), including the average vegetation height and the height of median energy (HOME) variables. Forest Inventory data Field biomass data sets were derived from forest inventories carried out in Congo, Cameroon, and Uganda. Allometric equations were used to convert timber volume measurements into above ground biomass. Remotely Sensed Data The MODIS Nadir Bidirectional Reflectance Distribution Function (BRDF) adjusted reflectances (NBAR) product (MOD43B4.V4) provides surface reflectances at 1 km spatial resolution and composited 16 day temporal resolution. The study uses seven bands designed for land applications with wavelength from 459 to 2155 nm. We analyzed ten 16-day products of NBAR data for each year between 2000 and 2003, and developed a mosaic of best quality observations. The plots on the right show the vertical lidar profile. The black dotted lines indicate the leading edge (canopy height) and signal end, the purple line is the height of median energy (HOME), the blue line is the trailing edge (ground) of the waveform, and the red line is the fitted waveform. Contacts: Figure 2. Relationship between GLAS derived height, ratio of HOME and height and predicted biomass (Mg/ha) aggregated in classes of 10 Mg/ha. The horizontal bars show the standard error for the GLAS metric within biomass bin.