Estimation of Cloud Fraction Profile in Shallow Convection Using a Scanning Cloud Radar

Kollias, P., Stony Brook University

Cloud Distributions/Characterizations

Convective Processes

Oue M, P Kollias, KW North, A Tatarevic, S Endo, AM Vogelmann, and WI Gustafson Jr.. 2016. "Estimation of cloud fraction profile in shallow convection using a scanning cloud radar." Geophysical Research Letters, 43(20), 10.1002/2016gl070776.


Fig. 1: (a) Horizontal distributions of hydrometeor mixing ratio and Ka-band reflectivity at 4.19 km and 2.42 km above ground level at 2100 UTC on 9 June 2015 over the SGP. Radar sensitivity was applied to the reflectivity plots assuming cross-wind horizon-to-horizon scans (CWRHI). Dots represent KAZR locations that produced CFPs in (b). (b) The CFP from hydrometeor mixing ratio over the LES domain, and CFPs from 10 KAZR dwells with their mean CFP and standard deviation.


Fig. 2: (a) Contoured frequency altitude diagram (CFAD) of simulated Ka-band reflectivity for 9 June 2015 at 2100 UTC. Black line represents the mean profile and white lines represent CDF isolines of 5, 10, 15, 20, and 50% going from left to right. (b) Cloud fraction profiles corresponding to the 10% CDF isoline with changing scan duration time (colored lines). Black dashed line in (b) represents the LES domain-averaged CFP for hydrometeor mixing ratio ≥ 0.01 g kg-1.


Fig. 1: (a) Horizontal distributions of hydrometeor mixing ratio and Ka-band reflectivity at 4.19 km and 2.42 km above ground level at 2100 UTC on 9 June 2015 over the SGP. Radar sensitivity was applied to the reflectivity plots assuming cross-wind horizon-to-horizon scans (CWRHI). Dots represent KAZR locations that produced CFPs in (b). (b) The CFP from hydrometeor mixing ratio over the LES domain, and CFPs from 10 KAZR dwells with their mean CFP and standard deviation.

Fig. 2: (a) Contoured frequency altitude diagram (CFAD) of simulated Ka-band reflectivity for 9 June 2015 at 2100 UTC. Black line represents the mean profile and white lines represent CDF isolines of 5, 10, 15, 20, and 50% going from left to right. (b) Cloud fraction profiles corresponding to the 10% CDF isoline with changing scan duration time (colored lines). Black dashed line in (b) represents the LES domain-averaged CFP for hydrometeor mixing ratio ≥ 0.01 g kg-1.

Science

Shallow convection plays a critical role in the heat and moisture transfer between the boundary layer and free atmosphere above about 2 km. However, with an average spatial scale of 0.5–1.5 km, shallow cumuli are not resolved in weather forecast and climate models, and their broken cloud coverage results in uncertainties in estimations of domain-averaged cloud fraction profiles (CFP, Fig. 1).

Impact

An objective method is proposed for estimating domain-averaged CFP using observed statistics of Ka-band scanning cloud radar (Ka-SACR) hydrometeor detection with height to estimate optimum sampling regions. This method shows good agreement with simulated CFP from large-eddy simulations (LES). The new techniques developed in the study increase confidence in the retrieved CFP when looking at the true atmosphere and improve our ability to compare model output with cloud radar observations for shallow cumulus cloud conditions.

Summary

This study is the first to use LES output from the new LES ARM Symbiotic Simulation and Observation (LASSO) capabilities currently under development. LES from LASSO and of the RACORO field campaign provide detailed cloud fields that serve as proxies of true cloud fields. The LES combined with a radar simulator provide an estimate of what can be observed in the atmosphere.

Uncertainties in domain-averaged CFP estimates are addressed using LES of shallow convection over the SGP coupled with a radar simulator. The radar simulation analysis indicates that observation from a single vertically-pointing Ka-band radar (KAZR) is inadequate to provide reliable CFPs (Fig. 1b). Use of Ka-SACR, performing a sequence of cross-wind horizon-to-horizon scans, is not straightforward because radar sensitivity decreases with distance. Sampling a small region in the vicinity of Ka-SACR ensures higher sensitivity, but this results in undersampling the overall cloud field. Alternatively, a larger sampling region ensures sampling more clouds, but the lower sensitivity farther from the radar underestimates the CFP. Using the cumulative distribution function (CDF) of reflectivity as a function of altitude, as depicted in Fig. 2a, an optimal minimum detectable reflectivity at each height is determined from the CDF isoline that provides an optimum selection of region size and radar sensitivity. The Ka-SACR observations need to be conducted for 35 min or more for CFP estimates to converge with the LES-simulated CFP (Fig. 2b) with an RMSE less than 1% on the order of 5% CFPs.