Improving the Retrieval of Forest Canopy Chlorophyll Content From MERIS Dataset by Introducing the Vegetation Clumping Index

The accurate retrieval of canopy chlorophyll content (CCC) is essential to the effective monitoring of forest productivity, and environmental stress. However, the clumping index (CI), a vital canopy structural parameter, affects inaccurate remote sensing of forest CCC. In this article, we proposed a concept of effective CCC (CCC
_e
) and an integrated CI approach to retrieving forest CCC using empirical regression and random forest regression. First, the PROSPECT-D and four-scale models were used to simulate forest canopy spectra, and the forest CCC
_e
including CI was found more feasible to be remotely sensed than the CCC. Then, an empirical regression model and random forest model trained using different combinations of the medium resolution imaging spectrometer (MERIS) terrestrial chlorophyll index (MTCI), reflectance, and CI values were used to estimate the CCC. Finally, the proposed approach was tested using satellite-based CI and MERIS product. Using the empirical regression model, the results showed that the retrieval of forest CCC using the MTCI was greatly improved by the inclusion of the CI (RMSE from 63.64 to 36.51 μg cm
^-2
for broadleaf; RMSE from 96.02 to 58.49 μg cm
^-2
for coniferous). Using the random forest approach and the model trained using the reflectance in red and red-edge bands, MTCI, and CI performed best, with RMSE = 27.95 μg cm
^-2
for broadleaf and RMSE = 34.83 μg cm
^-2
for coniferous. Overall, it is concluded that include the CI, particularly the approach using forest random regression, have the potential for satellite-based forest CCC mapping at regional and global scales.