Cloud/radiation deficiency in storm-scale NWP and application of reduced effective cloud droplet radius

Representation of clouds in weather and climate models is essential to estimate modulation of shortwave and longwave radiative fluxes incident at the earth’s surface. Accurate estimates of the surface energy budget require accurate estimates of different cloud types, their depth, and their radiative properties. Improved treatments of cloud-radiation effect were implemented within a new version of the NOAA 3km High-Resolution Rapid Refresh (HRRR) weather/earth-system model in December 2020.

This change to the HRRR and RAP models was enabled by a new set of NOAA observations used for model evaluation starting in 2015, surface radiation measurements from the NOAA SURFRAD radiation network (Augustine et al 2000, 14 stations including the ancillary SOLRAD network). From this point on, HRRR and other NWP models were evaluated by RMS error and bias of downward SW radiation comparing model predictions and SURFRAD observations. It was found that before 2020, the HRRR model has a lower-48-US-average daytime warm-season bias of downward SW radiation over CONUS of over 100 W/m². However, this bias was decreased in 2020 in large part by reducing effective radius for cloud droplets resulting in more reflection/scattering and absorption of shortwave radiation by clouds.

Nevertheless, in this paper, we demonstrate that despite these improvements enabled by validation using SURFRAD observations before 2020, an excessive bias in downward SW radiation continued with the HRRR model after 2020 and above all, even that that excessive bias was present in each of the 14 SURFRAD/SOLRAD stations with different climates and cloud types. This means that there is a uniform problem about downward solar radiation in the HRRR model, not just in a few locations. We found that this error was not from clear-sky errors but from insufficient attenuation by clouds. Recent experiments with HRRR for our study then showed that further reduction of effective radius for cloud droplets (both explicit and subgrid-scale- SGS) further reduces downward SW radiation bias toward zero.

This outcome is important for prediction of boundary-layer evolution with energy, aviation, severe weather, and hydrology implications. We will describe our hypotheses for excessive downward SW radiation and our HRRR experiments modifying cloud droplet size for both explicit and SGS clouds carried out in different seasons.