PNNL

Abstract

Retrievals of cloud top heights from the ARM 35-GHz MMCR located on Manus Island are compared to those from the GMS-5 satellite as a means to inter-evaluate the accuracy of both MMCR and GMS-5 retrievals as well as to ascertain their limitations. Comparisons are carried out for retrievals of both single-layer and multilayer clouds as seen by radar, but only for clouds with 100% amount within a 0.3o x 0.3 o domain centered at the ARM site with one cloud type (low, middle, or high) as observed from the satellite. It is shown that mean differences, with 95% confidence intervals included, between radar- and satellite-retrieved cloud top heights are 0.3±0.3 km for single-layer clouds and -0.7±0.3 km for multilayer clouds. The study reveals that for thick clouds (cloud base =1 km and cloud thickness =10 km), for which the satellite retrievals should be accurate, retrievals from the MMCR agree with those from satellite with mean differences of 0.0±0.4 km and -0.2±0.3 km for single-layer and multilayer clouds, respectively. For clouds of lesser thickness, the MMCR retrievals of cloud top heights are accurate since beam attenuation is not as prominent. In such cases the satellite derived cloud top heights appear to be underestimated by as much as 2.0 km as a consequence of the non-blackbody emissions from these relatively optically thin clouds.