Comparison of the inversion of two canopy reflectance models for mapping forest crown closure using imaging spectroscopy

We compare the inversion of two canopy reflectance models to estimate forest crown closure (CC) using an EO-1 Hyperion image: the Kuusk-Nilson forest reflectance and transmittance (FRT) model, and the Li-Strahler geometric-optical model. For predicting CC on a per-pixel basis, the FRT model inversion is carried out by minimizing a merit function that provides a measure of the difference between the reflectance simulated by the FRT model and the reflectance originating from optimal band selection of Hyperion data. The inversion of the Li-Strahler model mainly depends on the relationship between the scene component "sunlit background" and forest structural parameters. We complement prediction deficiencies of the inverted Li-Strahler model CC using a spatial interpolation algorithm (regression kriging) in infeasible regions. Field-measured CCs of 40 sample sites are used to validate the inversion quality of both models. The results indicate that the Li-Strahler model inversion (R-2 = 0.67, RMSE = 0.043) performs better than the FRT model inversion (R-2 = 0.53, RMSE = 0.072) for CC retrieval. Estimated CC using the Li-Strahler model inversion combined with spatial interpolation yield a final, continuous CC map for the Longmenhe forest nature reserve in China, which is used as a study area for this work. The advantages and disadvantages of these two models inversion combined with imaging spectrometer data for mapping forest CC are discussed.