As opposed to ground-based methods, remote sensing deciphers the reflected, instead of the transmitted radiation. The plant communities are full of chlorophylls, a set of pigments which absorb part of the solar radiation for photosynthesis. As a result, reflectance of radiation in different spectral bands, especially the two widely used infrared and red ones, is changed proportionally to the amount of green vegetation. Since large-area maps of LAI are needed for global land-surface modeling, plenty of empirical relationships (i.e., statistical correlations) have been proposed between satellite or airborne image reflectance and ground-based (in situ) estimations of LAI. There are many vegetation indices developed from radiances in a wide range of channels corresponding to spectral bands. LAI estimation from satellite data requires ground data for validation and testing for bias. Satellite data must be corrected for atmospheric effects, thus requiring additional information on the state of the atmosphere (especially water vapor, aerosols, and ozone). Nondestructive (optical) measurements are the preferred approach for obtaining ground measurements. Classical values of LAI derivation from remote sensed vegetation indices (like Normalized Difference Vegetation Index,
NDVI) range from 0 to 4-5, fitting an empirical exponential function with a plateau indicating a saturated signal for higher LAI. Such a relationship was first established for wheat and maize, at various states of growth. Unfortunately, LAI often reaches values above 5 and up to 15 in temperate mixed broadleaved forests or coniferous plantations. Then remote LAI derivation is not well adapted for such kinds of vegetation. New algorithms using the intrapixel variability of signals have been proposed but remain to be tested using ground-based data sets of forest LAI differing from the calibration data set.
Was this article helpful?