Relationships between tropical forest structure and remotely sensed synthetic aperture radar data

Relationships among forest structural parameters and SAR data were investigated in stands of swamp forest in Georgia and in tropical forest in eastern and central Amazon to evaluate the feasibility of retrieval of information about forest structure and canopy characteristics with radar data. Tree abundance and stem diameter at breast height (dbh) were measured and parameters of dbh distribution, stem density and basal area were calculated for different stands in each site. Hemispherical images of the stand canopies were registered in the field simultaneously to the SAR data acquisition. Canopy gap fraction was calculated by the integration of the fraction of the hemisphere with unobstructed view of the sky. SAR acquisitions were made for C band HH and VV in eastern Amazon, polarimetric L and C bands in central Amazon and polarimetric C, L and P bands in Georgia. Canopy gap fraction had significant negative correlations with stem density and with the slope of dbh histogram at all three sites, indicating that light penetration to the forest floor is controled by the understory. In the Amazon, significant correlations (P {dollar}<{dollar} 0.05) between SAR data and forest parameters were found at C band. Canopy gap fraction had significant negative correlation with CVV (r{dollar}sp2{dollar} = 0.61), in central Amazon and with the ratio CVV/(CHH+CVV) (r{dollar}sp2{dollar} = 0.85) in eastern Amazon. In central Amazon, basal area had significant negative correlations with CHV (r{dollar}sp2{dollar} = 0.72) and with the ratio LHV/CHV (r{dollar}sp2{dollar} = 90). The relationships between SAR backscatter and canopy gap fraction, the better results found with C band than with L band, and the better relation of cross-polarized backscattering with basal area, indicate that in tropical forests, radar interacts predominantely with the canopy layer in L and C band.