The impact of the diurnal variation of albedo on the remote sensing of thedaily mean albedo of grassland

The correction of a land surface albedo estimate made at one solar zenith a ngle (SZA) from a polar-orbiting satellite to a standard SZA or to a daily mean albedo requires knowledge of the dependence of the albedo on SZA. This paper uses ground-based measurements of the clear-sky albedo at a uniform grassland site at Uardry (34.39 degrees S, 145.30 degrees E) in southeaster n Australia to investigate the accuracy to which the daily mean albedo can be inferred from the albedo at 1030 LST given knowledge of the SZA dependen ce of albedo to various levels of detail. During nine months in which the d aily mean albedo varied from 0.20 to 0.27, the albedo always had the expect ed minimum near noon but the strength of the albedo's SZA dependence varied greatly. For a few months, albedos were up to 0.04 higher in the afternoon than in the morning, and variations on finer timescales of up to 0.02 also appeared in the diurnal albedo cycle for days or weeks. These features of the diurnal variation were all seen at two or three surface points separate d by up to 750 m and so are expected to appear at the similar to 1-km resol ution of many satellite sensors. For the Uardry grassland site, the error i n estimating the daily mean albedo from the 1030 LST, albedo can be up to 0 .03, which is 15% of an albedo of 0.20, if the albedo is assumed to be cons tant through the day. The maximum error is reduced to about 0.02 if a simpl e model of the SZA dependence is used with even an approximate value for th e parameter that controls the strength of the dependence, and to 0.01 or le ss if the strength of the dependence is appropriate to the state of the veg etation on the day. Afternoon-morning asymmetry in the albedo can contribut e almost 0.01 to the error in inferring a daily albedo from a morning measu rement.