Colorado State University General Circulation Model

Robert Carroll Levy

Levy, R., 1996: An analysis of southern great plains ARM cloudiness and surface radiation data. Masters thesis. Colorado State University, Ft. Collins., 145 pp.

Cloud fraction estimated by the micropulse lidar and by Pat Minnis' satellite cloud retrieval. Whole IOP averaged.

"Composite" time-series plots show the interaction of clouds, water, and radiation. The clouds (top panel) are positioned by three methods, cloud-base from the MPL(dots), cloud-top from Minnis (bars), and high relative humidity from the soundings (shading). For the water (middle panel), the precipitable water paths are represented by Xs and the liquid water path by dots (multiplied by 100 to fit on the same plot). Finally, the radiation (bottom panel) includes color lines representing time-series of BSRN observations. This example, Nov 9, shows that at 18UTC, low clouds were present along with high LWP, reduction of SW, and increased downward LW.

CLEAR SKY: Modeled radiative fluxes compared with observations : A "clear sky" was defined when the Micropulse Lidar (MPL) detected no clouds during an hour-long interval. 141 clear-sky cases were observed during the IOP, of which these plots show longwave and shortwave fluxes, calculated by the CSU GCM radiation parameterization and compared to BSRN observations. The shortwave plot shows good fit to the one-to-one line, however the model produces direct and total fluxes about 10% too high. Many diffuse fluxes are near zero, however there are a number of times when diffuse flux was recorded when the model didn't predict it, due to radiation reflecting off clouds in the sky that were not directly overhead and seen by the MPL. The longwave parameterization accurately calculates downwelling and net surface longwave fluxes, but does poorly on top-of-atmosphere outgoing longwave. This bias can be attributed to the parameterization not including the upper atmosphere (above 50 mb) in its calculations.

CLOUDY SKY: Modeled radiative fluxes compared with observations : These figures show calculated (with the CSU GCM radiation parameterization) downwelling shortwave fluxes, downwelling IR flux and upwelling (Top-of-atmosphere) IR flux compared with observations from the BSRN and the satellite (Minnis). Daytime values are red and nighttime are blue. Cloud bases and tops can be calculated by two ways, from sounding derived relative humidities, or by a combination of the Micropulse Lidar (MPL) and Minnis. The clouds themselves are broken into categories, depending on the temperature of the cloud: "liquid" (above 253 K), "ice" (below 253 K) or "mixed" (straddling 253 K). Note that these plots only include clouds that can be categorized as one-layer. The Liquid Water Path was distributed linearly (with zero water at the base) in liquid cloud regions (Slingo et al., 1980), and a characteristic ice-content was assumed based on mid-cloud temperature (Platt and Harshvardhan, 1989), and distributed linearly with zero ice at the top (Stackhouse and Stephens, 1989).

At first glance the plots using Sonde are very similar to those using the MPL/Minnis derived cloud positions, however upon close look, the quality of fit is different. The Lidar-derived LW plots better straddle the "best fit" lines, while the Sonde-derived LW plots have less scatter. This is true for the SW plots as well. In both cases, for both day and night, the longwave plots are much better than the shortwave plots, probably because we have better models for cloud absorption then of cloud-particle scattering.

Cloud bases and tops can be calculated by two ways, from sounding derived relative humidities, or by a combination of the Micropulse Lidar (MPL) and Minnis. The clouds themselves are broken into categories, depending on the temperature of the cloud: 'liquid' (above 253K), 'ice' (below 253K) or 'mixed' (straddling 253K). Note that these plots only include clouds that can be catagorized as one-layer. The Liquid Water Path was distributed linearly (with zero water at the base) in liquid cloud regions (Slingo et al., 1980), and a characteristic ice-content was assumed based on mid-cloud temperature (Platt and Harshvardhan, 1989), and distributed linearly with zero ice at the top (Stackhouse and Stephens, 1989).

There are not a lot of differences between Sonde-derived and MPL/Minnis derived plots. Most differences occur because the satellite recorded generally higher cloud tops than did the sonde, so for example, many clouds characterized as 'liquid' by the sonde, were characterized as 'mixed' by the lidars. For both cloud algorithms, liquid clouds produced good albedo fits, and ice clouds produced poor fits. While the clear sky calculations produced albedos close to observed, the weak slope of the regression line shows that the Briegleb et al., (1986) surface albedo model depended too strongly on solar zenith angle.

Red lines represent shortwave heating, blue lines are longwave heating, and the black lines represent the total heating. For each color, dotted lines represent the cloudy-sky calculations. Without clouds, the atmosphere generally cools by about one degree in longwave, and warms by up to one degree by shortwave in midafternoon. The combination creates a near zero heating. Clockwise from top left: a thin but vertically extensive cloud, a high cloud, a low cloud, and a thick cloud.

What happens to the heating profile when the clouds are changed from single-level to multi-level? Note that for the single layer cloud, water is assumed for levels up to about 490 mb, while ice is assumed for levels above that. For the two-layer cloud, water is assumed in the lower cloud, while ice is assumed for the upper cloud.