Angular and seasonal variation of spectral surface reflectance ratios: implications for the remote sensing of aerosol over land

We obtain valuable information on the angular and seasonal variability of surface reflectance using a hand-held spectrometer from a light aircraft. The data is used to test a procedure that allows us to estimate visible surface reflectance from the longer wavelength 2.1 _tm channel (mid-IR). Estimating or avoiding surface reflectance in the visible is a vital first step in most algorithms that retrieve aerosol optical thickness over land targets. The data indicate that specular reflection found when viewing targets from the forward direction can severely corrupt the relationships between the visible and 2.1 _tm reflectance that were derived from nadir data. There is a month by month variation in the ratios between the visible and the mid-IR, weakly correlated to the Normalized Difference Vegetation Index (NDVI). If specular reflection is not avoided, the errors resulting from estimating surface reflectance from the mid-IR exceed the acceptable limit of Ap 0.01 in roughly 40% of the cases, using the current algorithm. This is reduced to 25% of the cases if specular reflection is avoided. An alternative method that uses path radiance rather than explicitly estimating visible surface reflectance results in similar errors. The two methods have different strengths and weaknesses that require further study. 1.0 Introduction When viewed from space, the striking difference between land and ocean surfaces is the uniformity of the ocean and the large spatial variability of the land. Land surface reflectance is dependent on the type of vegetation and the type of soil, as well as the relative fraction of each (Huete and Tucker, 1990). Surface reflectance varies as the vegetation progresses through its growing cycle. It also varies due to seasonal changes in soil moisture and hydrological