Quantifying the Impacts of Subpixel Reflectance Variability on Cloud Optical Thickness and Effective Radius Retrievals Based On High‐Resolution ASTER Observations

Werner, F ; Zhang, Z ; Wind, G ; Miller, D J ; Platnick, S

关键词: CLOUDS (METEOROLOGY);    REFLECTANCE;    SATELLITE OBSERVATION;    THERMAL EMISSION;    RADIOMETERS;    REMOTE SENSING;   

RP-ID  :  GSFC-E-DAA-TN56863

学科分类：地球科学(综合)

美国|英语

来源: NASA Technical Reports Server

【 摘 要 】

TOOLS SHAREAbstractRecently, Zhang et al. (2016) presented a mathematical framework based on a second‐order Taylor series expansion in order to quantify the plane‐parallel homogeneous bias (PPHB) in cloud optical thickness (τ) and effective droplet radius (r(sub eff)) retrieved from the bispectral solar reflective method. This study provides observational validation of the aforementioned framework, using high‐resolution reflectance observations from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) over 48 marine boundary layer cloud scenes. ASTER reflectances at a horizontal resolution of 30 m are aggregated up to a scale of 1,920 m, providing retrievals of τ and r(sub eff) at different spatial resolutions. A comparison between the PPHB derived from these retrievals and the predicted PPHB from the mathematical framework reveals a good agreement with correlation coefficients of r > 0.97 (for Δτ) and r > 0.79 (for Δr(sub eff)). To test the feasibility of PPHB predictions for present and future satellite missions, a scale analysis with varying horizontal resolutions of the subpixel and pixel‐level observations is performed, followed by tests of corrections with only limited observational high‐resolution data. It is shown that for reasonably thick clouds with a mean subpixel τ larger than 5, correlations between observed and predicted PPHB remain high, even if the number of available subpixels decreases or just a single band provides the information about subpixel reflectance variability. Only for thin clouds the predicted Δr(sub eff) become less reliable, which can be attributed primarily to an increased retrieval uncertainty for r(sub eff).