Spectroscopic determination of leaf water content using continuous wavelet analysis

Leaf water content as a percentage of dry mass (LWC_D), also termed fuel moisture content, is a critical variable for fire risk modeling. To date, the relationship between spectral reflectance and LWC_D is poorly understood due to the mixed effect of leaf water and dry mass on spectral response. Few studies have reported an adequate estimation of LWC_D from leaf reflectance spectra, in particular for a variety of species. In this study, we propose an effective approach, using continuous wavelet analysis, to determine LWC_D from reflectance spectra for 47 tropical forest species in Panama. The proposed method provides a new way to select a small number of wavelet features (wavelet coefficients) that are sensitive to changes in LWC_D. Wavelet features derived using this new method are compared with common spectral indices in terms of their relationship with LWC_D. Results demonstrate that LWC_D is poorly correlated to the water indices but strongly correlated to the derived wavelet features. Accurate estimates of LWC_D across a range of species are achieved using any of those wavelet features and can be slightly improved using a combination of them.