Mapping the forest height with multispectral and synthetic aperture radar (SAR) data

The availability of updated forest height (FH) maps is becoming increasingly important with the growing impacts of climate change on communities, namely through natural catastrophes as wildfires and other extreme weather events. Spaceborne light detection and ranging sensors have been successfully used to provide reliable information on the vertical structure of forests globally; however, these missions only provide limited data with sparse coverage. Considering this, the authors have proposed several models to map FH resorting to synthetic aperture radar (SAR) and multispectral sensors, which provide gap-free observations, enabling the production of wall-to-wall FH maps locally and globally. This chapter presents a brief review of the most used approaches for mapping the FH resorting to SAR, multispectral, and the combination of both sensors. It was verified that when using SAR, most authors opted to utilize interferometric models; whereas when using multispectral observations (alone or together with SAR), most used nonparametric models. Other approaches using tomographic SAR or parametric models are also reported in the literature. According to the results discussed in this chapter, the random volume over ground model was the most accurate interferometric model. Moreover, the inclusion of parameters that compensate for the temporal decorrelation and slope was found to have a positive effect on FH mapping. Regarding the approaches that included multispectral, nonparametric models were the most reliable (the random forests, being the most used).