INVERTING OPTICAL REFLECTANCE TO ESTIMATE SURFACE-PROPERTIES OF VEGETATION CANOPIES

An inversion method using a simple bidirectional reflectance model and data on optical reflectances remotely-sensed from satellites has been improved to derive surface properties such as the leaf area index (LA I). Such properties are important in deriving the resistance of the ve getative canopy to uptake of gaseous trace chemicals from the atmosphe re and in the study of radiation transfer processes. We found that a m ulti-pass retrieval technique can greatly improve a model's ability to retrieve surface properties. Because the sensitivity of the model inv ersion to initial values is an important issue that depends on (1) the partial derivative of reflectance with respect to each parameters to be retrieved (partial derivative R/partial derivative x) and (2) the d egree of independence among model parameters, we investigated the issu e with synthetic data constructed by a bidirectional reflectance model . The results revealed that, although the surface optical properties a re mostly independent of each other, their initial values did have som e effect on the retrieved value of the LAI, with the worse case caused by leaf angle distribution index, n, at close-to-nadir solar and view zenith angles of the reflectance data. At near-nadir angles, n and LA I were strongly correlated, and their retrieval was not unique. When a pplied to satellite remote sensing data obtained with the advanced ver y-high-resolution radiometer (AVHRR), the model-retrieved seasonal var iation of surface properties agreed reasonably well with independent g round measurements made in the First International Satellite Land-Surf ace Climatology Project (ISLSCP) Field Experiment (FIFE) campaign. App lication to Landsat data to retrieve spatial variation was less succes sful, largely because of the close-to-nadir solar and view zenith angl es in the data.