Comparison of airborne and surface spectral bidirectional reflectance factors, spectral hemispherical reflectance and spectral vegetation indices

Abstract

Six sets of airborne advanced solid‐state array spectroradiometer (ASAS) and ground modular multiband radiometer (MMR) remotely sensed bidirectional measurements acquired over one First International Satellite Land Surface Climatology Project (ISLSCP) Field Experiment site in 1987 and 1989 were compared for the following parameters: bidirectional spectral reflectance factor, spectral hemispherical reflectance, simple ratio, and normalized difference vegetation index. ASAS at‐sensor radiances were atmospherically corrected and converted to surface reflectance factors. Selected MMR bands were simulated with ASAS data, and hemispherical reflectance for both ground and airborne data sets was estimated using data collected only in or close to the solar principal plane. The shapes of the reflectance factor response curves (as a function of view zenith angle) were strongly affected by solar zenith angle and compared well between instruments. Off‐nadir anisotropy in reflectance factors was comparable between sensors, while actual values of reflectance factors differed 2–35% relative between instruments in the green channel and 0–38% relative in the red channel. For the data set giving the overall closest agreement, ASAS reflectance factors differed from MMR values by 5–17% relative (0.3–1.8% absolute) in the green and 5–10% relative (0.2–0.5% absolute) in the red. These differences showed no correlation with solar zenith angle. Some of the differences were arbitrarily introduced by the 5° offset in view zenith angles (except nadir) between the two instruments. Other differences were caused by the discrepancy in solar zenith angle for some of the sets, variable deviations from the solar principal plane, hotspot and associated shadowing effects not consistently recorded by both instruments, and sampled ground area variations. Estimates of hemispherical reflectance compared very well between sensors, with differences of only 3–14% relative (0.3–1.6% absolute) in the green and 11–27% relative (0.8–2.4% absolute) in the red. For the ASAS data, using atmospherically corrected reflectance factors (instead of at‐sensor radiances) significantly increased the values of the spectral vegetation indices (SVIs). Off‐nadir anisotropy for the SVIs derived from both sensors' data sets was less than that observed for the bidirectional reflectance factors, and NDVI off‐nadir deviations were much less than those of the simple ratio (SR). Large differences in the values of SR and NDVI between sensors indicate SVIs calculated from broadband (MMR) versus narrowband (ASAS) data are not comparable.