Abstract
Empirical studies using remotely sensed data to estimate forest stand parameters have shown that the biological variable which is most closely functionally linked to the spectral response of the canopy is the leaf area index (LAI). The LAI is highly important for the photosynthetic functioning of the canopy since it has a dominant influence on the light interception capability or efficiency ε(i). This efficiency depends on the LAI and on the optical properties of the canopy elements and it is also functionally linked to the canopy reflectance. Furthermore, this efficiency is linked to the biomass production of forest canopies since this latter is the result of photosynthesis, through which a fraction of the intercepted incident solar energy is converted into biomass.
We propose a semi‐empirical model between εi and canopy reflectance (in terms of normalized difference vegetation index (NDVI)) which results from coupling semi‐empirical models of reflectance and light interception which are based on the Bouguer‐Lambert‐Beer law. This model is designed to have a more analytical basis than the simple statistical relationships, while remaining simple enough to be used on a large scale with satellite data. It does not depend explicitly on the LAI, but rather it is a function of NDVI which is calculated from ground equivalent reflectances in the red and near‐infrared channels. These reflectances can be estimated routinely from data of visible/near‐infrared satellite sensors, such as the NOAA/AVHRR, which are corrected for topography, atmospheric, illumination and view angles as well as mixel effects.
There are some constraints to using such a relationship on a global scale. The relationship between radiometric and biological stand parameters is degraded by the shadows and the reflectance of woody canopy elements, of background and of understory. Therefore, a basic requirement is that the reflectance of background and understory is known. Another problem is that the parameters of this relationship are often spatial and temporal averages. This difficulty can however be surmounted by considering different coefficients depending on the stand type and age.
