• Barnett, T. P., , J. Ritchie, , J. Foat, , and G. Stokes, 1998: On the space– time scales of the surface solar radiation. J. Climate, 11 , 8896.

    • Search Google Scholar
    • Export Citation
  • Betts, A. K., , and R. Boers, 1990: A cloudiness transition in a marine boundary layer. J. Atmos. Sci, 47 , 14801497.

  • Black, T. L., 1994: The new NMC mesoscale Eta Model: Description and forecast examples. Wea. Forecasting, 9 , 265278.

  • Bohren, C. F., 1987: Multiple scattering of light and some of its observable consequences. Amer. J. Phys, 55 , 524533.

  • Clothiaux, E. E., , T. P. Ackerman, , G. G. Mace, , K. P. Moran, , R. T. Marchand, , M. A. Miller, , and B. E. Martner, 2000: Objective determination of cloud heights and radar reflectivities using a combination of active remote sensors at the ARM CART sites. J. Appl. Meteor, 39 , 645665.

    • Search Google Scholar
    • Export Citation
  • Conover, W. J., 1980: Some methods based on ranks. Practical Nonparametric Statistics, Wiley Series in Probability Mathematics and Statistics, John Wiley and Sons, 215–228.

    • Search Google Scholar
    • Export Citation
  • Deardorff, J. W., 1980: Cloud top entrainment instability. J. Atmos. Sci, 37 , 131147.

  • Del Genio, A. D., , and A. B. Wolf, 2000: The temperature dependence of the liquid water path of low clouds in the southern Great Plains. J. Climate, 13 , 34653486.

    • Search Google Scholar
    • Export Citation
  • Del Genio, A. D., , M. S. Yao, , W. Kovari, , and K. K. W. Lo, 1996: A prognostic cloud water parameterization for global climate models. J. Climate, 9 , 345357.

    • Search Google Scholar
    • Export Citation
  • Deng, A., , N. L. Seaman, , and J. S. Kain, 2003: A shallow-convection parameterization for mesoscale models. Part I: Submodel description and preliminary analysis. J. Atmos. Sci, 60 , 3456.

    • Search Google Scholar
    • Export Citation
  • Geleyn, J-F., , and A. Hollingsworth, 1979: An economical analytical method for the computation of the interaction between scattering and line absorption of radiation. Beitr. Phys. Atmos, 52 , 116.

    • Search Google Scholar
    • Export Citation
  • Ghan, S. J., , L. R. Leung, , and J. McCaa, 1999: A comparison of three different modeling strategies for evaluating cloud and radiation parameterizations. Mon. Wea. Rev, 127 , 19671984.

    • Search Google Scholar
    • Export Citation
  • Ghan, S. J., , R. C. Easter, , J. Hudson, , and F. M. Brèon, 2001: Evaluation of aerosol indirect radiative forcing in MIRAGE. J. Geophys. Res, 106 , 53175334.

    • Search Google Scholar
    • Export Citation
  • Huber, P. J., 1981: Regression. Robust Statistics, Wiley Series in Probability Mathematics and Statistics, John Wiley and Sons, 153–198.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., , R. Pincus, , C. Hannay, , and K-M. Xu, 2004: Use of cloud radar observations for model evaluation: A probabilistic approach. J. Geophys. Res.,109, 3202, doi:10.1029/ 2003JD003473.

    • Search Google Scholar
    • Export Citation
  • Kato, S., , T. P. Ackerman, , E. E. Clothiaux, , J. H. Mather, , G. G. Mace, , M. L. Wesely, , F. Murcray, , and J. Michalsky, 1997: Uncertainties in modeled and measured clear-sky surface shortwave irradiances. J. Geophys. Res, 102 , 2588125898.

    • Search Google Scholar
    • Export Citation
  • Liljegren, J. C., 1994: Two-channel microwave radiometer for observations of total column precipitable water vapor and cloud liquid water path. Preprints, Fifth Symp. on Global Change Studies, Nashville, TN, Amer. Meteor. Soc., 262–269.

    • Search Google Scholar
    • Export Citation
  • Liljegren, J. C., , E. E. Clothiaux, , G. G. Mace, , S. Kato, , and X. Dong, 2001: A new retrieval for cloud liquid water path using a ground-based microwave radiometer and measurements of cloud temperature. J. Geophys. Res, 106 , 1448514500.

    • Search Google Scholar
    • Export Citation
  • Long, C. N., , and T. P. Ackerman, 2000: Identification of clear skies from broadband pyranometer measurements and calculation of downwelling shortwave cloud effects. J. Geophys. Res, 105 , 1560915626.

    • Search Google Scholar
    • Export Citation
  • MacVean, M. K., , and P. J. Mason, 1990: Cloud-top entrainment instability through small-scale mixing and its parameterization in numerical models. J. Atmos. Sci, 47 , 10121030.

    • Search Google Scholar
    • Export Citation
  • Miller, M. A., , M. P. Jensen, , and E. E. Clothiaux, 1998: Diurnal cloud and thermodynamic variations in the stratocumulus transition regime: A case study using in situ and remote sensors. J. Atmos. Sci, 55 , 22942310.

    • Search Google Scholar
    • Export Citation
  • Min, Q., , and L. C. Harrison, 1996: Cloud properties derived from surface MFRSR measurements and comparison with GOES results at the ARM SGP site. Geophys. Res. Lett, 23 , 16411644.

    • Search Google Scholar
    • Export Citation
  • Morcrette, J-J., , and C. Jakob, 2000: The response of the ECMWF model to changes in the cloud overlap assumption. Mon. Wea. Rev, 128 , 17071732.

    • Search Google Scholar
    • Export Citation
  • Randall, D. A., 1980: Conditional instability of the first kind upside-down. J. Atmos. Sci, 37 , 125130.

  • Sengupta, M., 2002: Radiative impact of continental stratus at the Southern Great Plains: A climatological perspective. Ph.D. thesis, The Pennsylvania State University, 130 pp. [Available online at http://etda.libraries.psu.edu/theses/approved/WorldWideFiles/ ETD-179/phdthesis.pdf.].

    • Search Google Scholar
    • Export Citation
  • Sengupta, M., , E. E. Clothiaux, , T. P. Ackerman, , S. Kato, , and Q. Min, 2003: Importance of accurate liquid water path for estimation of solar radiation in warm boundary layer clouds: An observational study. J. Climate, 16 , 29973009.

    • Search Google Scholar
    • Export Citation
  • Tiedtke, M., 1989: A comprehensive mass flux scheme of cumulus parameterization in large scale models. Mon. Wea. Rev, 117 , 17791800.

  • Tiedtke, M., 1993: Representation of clouds in large-scale models. Mon. Wea. Rev, 121 , 30403061.

  • Zhao, Q., , and F. H. Carr, 1997: A prognostic scheme for operational NWP models. Mon. Wea. Rev, 125 , 19311953.

  • Zhao, Q., , T. L. Black, , and M. E. Baldwin, 1997: Implementation of the cloud prediction scheme in the Eta Model at NCEP. Wea. Forecasting, 12 , 697712.

    • Search Google Scholar
    • Export Citation
  • Zhong, S., , and J. C. Doran, 1997: A study of the effects of spatially varying fluxes on cloud formation and boundary layer properties using data from the Southern Great Plains Cloud and Radiation Testbed. J. Climate, 10 , 327341.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Histograms of daytime, nighttime, and all cloud-base heights from observations (left). The consolidated 4-yr dataset, while (right) data are divided into years

  • View in gallery

    Histograms of seasonal cloud-base height from observations

  • View in gallery

    Histograms of daytime, nighttime, and all cloud liquid water paths from observations. (left) The consolidated 4-yr dataset, while (right) the data are divided into years

  • View in gallery

    Histograms of seasonal cloud liquid water path from observations

  • View in gallery

    Histograms of 4-yr consolidated, annual, and seasonal normalized cloud forcing obtained from observations

  • View in gallery

    Histograms of cloud liquid water path occurrence from observations and the ECMWF and MOLTS models for daytime, nighttime, and all cases

  • View in gallery

    Histograms of cloud-thickness occurrence from observations and the ECMWF and MOLTS models for daytime, nighttime, and all cases

  • View in gallery

    Histograms of cloud liquid water content occurrence from observations and the ECMWF and MOLTS models for daytime, nighttime, and all cases

  • View in gallery

    Histograms of seasonal cloud liquid water path occurrence from observations and the ECMWF and MOLTS models for all cases

  • View in gallery

    Histograms of seasonal cloud-thickness occurrence from observations and the ECMWF and MOLTS models for all cases

  • View in gallery

    Histograms of seasonal cloud liquid water content occurrence from observations and the ECMWF and MOLTS models for all cases

  • View in gallery

    Histograms of cloud-base height occurrence from observations and the ECMWF and MOLTS models for daytime, nighttime, and all cases

  • View in gallery

    Histograms of seasonal cloud-base height occurrence from observations and the ECMWF and MOLTS models for all cases

  • View in gallery

    Histograms of normalized cloud forcing occurrence from observations and the ECMWF and MOLTS models

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 30 30 5
PDF Downloads 30 30 3

Climatology of Warm Boundary Layer Clouds at the ARM SGP Site and Their Comparison to Models

View More View Less
  • 1 Pacific Northwest National Laboratory, Richland, Washington
  • | 2 The Pennsylvania State University, University Park, Pennsylvania
  • | 3 Pacific Northwest National Laboratory, Richland, Washington
© Get Permissions
Full access

Abstract

A 4-yr climatology (1997–2000) of warm boundary layer cloud properties is developed for the U.S. Department of Energy Atmospheric Radiation Measurement (ARM) Program Southern Great Plains (SGP) site. Parameters in the climatology include cloud liquid water path, cloud-base height, and surface solar flux. These parameters are retrieved from measurements produced by a dual-channel microwave radiometer, a millimeter-wave cloud radar, a micropulse lidar, a Belfort ceilometer, shortwave radiometers, and atmospheric temperature profiles amalgamated from multiple sources, including radiosondes. While no significant interannual differences are observed in the datasets, there are diurnal variations with nighttime liquid water paths consistently higher than daytime values. The summer months of June, July, and August have the lowest liquid water paths and the highest cloud-base heights. Model outputs of cloud liquid water paths from the European Centre for Medium-Range Weather Forecasts (ECMWF) model and the Eta Model for 104 model output location time series (MOLTS) stations in the environs of the SGP central facility are compared to observations. The ECMWF and MOLTS median liquid water paths are greater than 3 times the observed values. The MOLTS data show lower liquid water paths in summer, which is consistent with observations, while the ECMWF data exhibit the opposite tendency. A parameterization of normalized cloud forcing that requires only cloud liquid water path and solar zenith angle is developed from the observations. The parameterization, which has a correlation coefficient of 0.81 with the observations, provides estimates of surface solar flux that are comparable to values obtained from explicit radiative transfer calculations based on plane-parallel theory. This parameterization is used to estimate the impact on the surface solar flux of differences in the liquid water paths between models and observations. Overall, there is a low bias of 50% in modeled normalized cloud forcing resulting from the excess liquid water paths in the two models. Splitting the liquid water path into two components, cloud thickness and liquid water content, shows that the higher liquid water paths in the model outputs are primarily a result of higher liquid water contents, although cloud thickness may a play a role, especially for the ECMWF model results.

Current affiliation: CIRA, Colorado State University, Fort Collins, Colorado

Corresponding author address: Manajit Sengupta, CIRA, Colorado State University, Foothills Campus, Fort Collins, CO 80523. Email: sengupta@cira.colostate.edu

Abstract

A 4-yr climatology (1997–2000) of warm boundary layer cloud properties is developed for the U.S. Department of Energy Atmospheric Radiation Measurement (ARM) Program Southern Great Plains (SGP) site. Parameters in the climatology include cloud liquid water path, cloud-base height, and surface solar flux. These parameters are retrieved from measurements produced by a dual-channel microwave radiometer, a millimeter-wave cloud radar, a micropulse lidar, a Belfort ceilometer, shortwave radiometers, and atmospheric temperature profiles amalgamated from multiple sources, including radiosondes. While no significant interannual differences are observed in the datasets, there are diurnal variations with nighttime liquid water paths consistently higher than daytime values. The summer months of June, July, and August have the lowest liquid water paths and the highest cloud-base heights. Model outputs of cloud liquid water paths from the European Centre for Medium-Range Weather Forecasts (ECMWF) model and the Eta Model for 104 model output location time series (MOLTS) stations in the environs of the SGP central facility are compared to observations. The ECMWF and MOLTS median liquid water paths are greater than 3 times the observed values. The MOLTS data show lower liquid water paths in summer, which is consistent with observations, while the ECMWF data exhibit the opposite tendency. A parameterization of normalized cloud forcing that requires only cloud liquid water path and solar zenith angle is developed from the observations. The parameterization, which has a correlation coefficient of 0.81 with the observations, provides estimates of surface solar flux that are comparable to values obtained from explicit radiative transfer calculations based on plane-parallel theory. This parameterization is used to estimate the impact on the surface solar flux of differences in the liquid water paths between models and observations. Overall, there is a low bias of 50% in modeled normalized cloud forcing resulting from the excess liquid water paths in the two models. Splitting the liquid water path into two components, cloud thickness and liquid water content, shows that the higher liquid water paths in the model outputs are primarily a result of higher liquid water contents, although cloud thickness may a play a role, especially for the ECMWF model results.

Current affiliation: CIRA, Colorado State University, Fort Collins, Colorado

Corresponding author address: Manajit Sengupta, CIRA, Colorado State University, Foothills Campus, Fort Collins, CO 80523. Email: sengupta@cira.colostate.edu

1. Introduction

Clouds influence radiative, dynamical, and hydrological processes. Hence, their accurate representation in numerical weather prediction (NWP) and general circulation models (GCMs) is crucial to the performance of these models. In recent years simulations of clouds, as well as of hydrological processes, have improved in both GCMs (e.g., Tiedtke 1993; Del Genio et al. 1996) and NWP models (e.g., Zhao et al. 1997) as a result of treating cloud liquid water as a prognostic variable. While there have been assessments of model liquid water paths using relatively short-term ground-based measurements (e.g., Ghan et al. 1999) and long-term satellite measurements (e.g., Ghan et al. 2001), there have been no statistical comparisons with ground-based measurements using multiyear data. The continuous ground-based remote sensing measurements of cloud liquid water path now being made at the U.S. Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) Program Southern Great Plains (SGP) site in Oklahoma provides long-term datasets to assess model-derived cloud liquid water paths.

Using ARM SGP datasets, we derive cloud liquid water path statistics for warm boundary layer clouds over the ARM SGP site and compare them to values obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) model and the National Center for Environmental Prediction (NCEP) early Eta model. To evaluate errors in model surface irradiance resulting from errors in cloud liquid water path we developed a simple parameterization of boundary layer cloud radiative properties as a function of solar zenith angle and cloud liquid water path using the observations at the ARM SGP site. We then applied the parameterization to both the observed and modeled cloud liquid water paths and compared the resulting sets of surface irradiances. Using the parameterization to estimate surface irradiance from cloud liquid water path allowed us to avoid the problem of extracting solar transmission values from model output fields, thereby placing the radiative effects of observed and modeled liquid water paths on an equal footing and saving considerable computational expense.

2. Measurements of cloud properties at the ARM SGP site

The central facility of the ARM SGP site is situated in Lamont, Okhlahoma (36.605°N, 97.485°W). We made use of measurements at this facility from 1997 through 2000 to build a 4-yr database of cloud liquid water paths, cloud boundaries, and the radiative properties of warm boundary layer clouds. We now describe the methods for retrieving cloud liquid water paths and cloud boundaries from the measurements at the ARM SGP central facility.

a. Cloud boundary heights and thermodynamic profiles

We identified the vertical distribution of cloud hydrometeors with a ceilometer, a micropulse lidar, and a millimeter-wavelength cloud radar. Compared to the radar, the ceilometer and lidar are more sensitive to large concentrations of small particles as opposed to small concentrations of large particles. As a result, the ceilometer and lidar provide the best estimates of cloud-base height. The radar beam is not attenuated significantly by most low-altitude clouds and is capable of providing dependable cloud-top height estimates for these clouds. Ground clutter, however, is a problem for the radar during the late spring, summer, and early autumn months, potentially biasing high the cloud-top height estimates for low-altitude clouds during these periods.

Clothiaux et al. (2000) provide a detailed description of a method for combining the active remote sensor data that yields estimates of the vertical distribution of cloud particles. A cloud binary mask is generated from these data such that each altitude bin containing a significant hydrometeor detection is labelled with a 1 and all other altitude bins are labeled with a 0. Using the cloud mask, identification of cloud layers and their boundary heights is straightforward.

To classify a cloud as warm we used atmospheric temperature profiles constructed from multiple sets of temperature observations, including those from radiosondes (G. G. Mace 2000, personal communication). Using the cloud-boundary height estimates from the active sensors and the temperature profiles, we classified those clouds having base and top temperatures above 273 and 253 K, respectively, as warm boundary layer clouds. We then removed from the study all clouds whose bases were above 3000 m.

b. Cloud liquid water path

When we consider the effects of boundary layer clouds on solar radiation, changes in cloud liquid water path (LWP), that is, the amount of condensed water in a vertical atmospheric column, are the dominant cause of variation in the flux reaching the surface. Therefore, accurate cloud liquid water path retrievals are essential for arriving at reasonable model calculations of surface solar irradiance based on the observed cloud liquid water paths. For this study, cloud liquid water paths are retrieved using brightness temperature measurements at 23.8- and 31.4-GHz frequencies from a dual-channel microwave radiometer (Liljegren 1994) that are input into a statistical retrieval (Liljegren et al. 2001). The Liljegren et al. (2001) retrieval also uses surface meteorology, temperature profiles, and millimeter-wave cloud radar hydrometeor detections as inputs for setting retrieval coefficients that depend on the atmospheric state at each time step.

c. Measured surface solar irradiance

A pyrheliometer provides observations of direct normal downwelling shortwave irradiance, while unshaded and shaded pyranometers measure the total and diffuse downwelling shortwave irradiances, respectively. In order to remove unknown and constant biases in the observations (e.g., Kato et al. 1997), as well as solar zenith angle effects, we use normalized cloud forcing, instead of irradiance, in the comparisons. Normalized cloud forcing (NCF) is defined as
i1520-0442-17-24-4760-e1
with Fclr and Fcld representing clear- and cloudy-sky fluxes, respectively. To obtain values of observed normalized cloud forcing one needs an estimate of the clear-sky irradiance Fclr that would be observed if the cloudy-sky period was actually clear. To estimate Fclr during cloudy-sky periods, we used the method developed by Long and Ackerman (2000), which fits an empirical function to the measured downwelling shortwave irradiances for a day using all clear-sky periods during that day. On days with no clear-sky periods, interpolated clear-sky empirical function fit coefficients from surrounding days are used to estimate the clear-sky downwelling solar irradiance for the cloudy day.

3. A climatology of warm boundary layer clouds

Using 4 yr of measurements from 1997 through 2000, we characterized the properties of warm boundary layer clouds observed during the period. The properties that we considered are cloud-base height, cloud liquid water path, and normalized shortwave cloud forcing. Since we are interested in the radiative properties of warm boundary layer clouds, we only included in our study the properties of single-layer clouds close to the surface with no cloud overhead. A cloud is treated single layered if significant radar returns for a particular profile were only from contiguous radar volumes. Cloud-base height and cloud liquid water path statistics were produced for both day and night, while normalized cloud forcing could only be computed and analyzed for daytime periods. The data were analyzed at 20-s resolution with each variable being averaged or interpolated to this resolution depending on its own temporal resolution. Note that we do not consider cloud-top heights in our climatology as radar returns from clutter above cloud top is possible, especially during summer daytimes. We do, however, use radar-derived cloud-top heights to estimate boundary layer cloud thicknesses for comparison with the models, indicating the possible impacts on our study of overestimation of cloud-top height as a result of clutter.

We searched for trends in the statistics both across years and between the different seasons. The 4-yr dataset was subdivided into four seasons for each year using groups of 3 months. The subdivisions are December–January–February (DJF), March–April–May (MAM), June–July–August (JJA), and September–October–November (SON). Summary statistics for the 4-yr and annual observation periods are presented in Table 1 with seasonal statistics presented in Table 2. Because the distributions of cloud-base height and cloud liquid water path do not follow a normal distribution, we use the two-sample Wilcoxon rank-sum nonparametric test (Conover 1980), also called the Mann–Whitney test, at a 5% level of significance to quantify whether certain interannual or interseasonal differences were significant. This test combines data from two samples into one sample while assigning ranks in an ascending order. The test statistic compares the sum of ranks from one sample to the other. We iterate that we do not use the two-sample t test, which is only valid for Gaussian distributions. We now consider these results in more detail.

a. Cloud-base height

Probability of occurrence histograms of cloud-base height for the 4-yr dataset, for warm single-layer clouds with bases below 3000 m, are presented in Fig. 1. Both the daytime and nighttime distributions have peaks in the lowest kilometer of the atmosphere with relatively constant, low-percentage cloud-base height occurrences above about 2000 m. The nighttime cloud-base heights are observed to be consistently smaller than daytime values (Table 1). Results from the Wilcoxon rank-sum test established that the difference between nighttime and daytime cloud-base height distributions is statistically significant. In a convective planetary boundary layer (PBL) cloud bases lie close to the lifting condensation level (LCL). The LCL rises with increasing surface temperature due to a decrease in the surface air relative humidity—a direct consequence of the temperature dependence of saturation vapor pressure. As the surface is cooler at night than during the daytime, the LCL, and therefore the cloud bases, are lower during night.

As all the cloud-base height distributions are positively skewed, the median height is a better measure of central tendency than the mean height. From Table 1 we find that 1997 and 1999 have the highest median cloud-base heights, while 2000 has a median height which is at least 100 m below the other years. Comparing 1997, 1998, and 1999 with 2000 using a Wilcoxon rank-sum test, our observation regarding the yearly height difference is found to be significant.

The cloud-base height distributions for the different seasons are shown in Fig. 2. These histograms illustrate that the cloud-base heights for the DJF season are generally lower than for the other seasons, while the convective JJA season has clouds more evenly distributed from the surface up to 3000 m. This phenomenon is again a result of the LCL lowering with decreasing temperatures, as detailed for diurnal cloud-base height differences. Del Genio and Wolf (2000) present similar seasonal observations and show that there is a larger dependence of cloud-base height on synoptic phase in warm months than in cold ones, with warm sector conditions producing much higher cloud bases than under any other synoptic condition. The evenly distributed cloud-base heights during the JJA (summer) season are most likely the result of shifts in the LCL with changing surface temperature and moisture over the course of this season.

The weak bimodality seen in the nighttime cloud-base height statistics is a result of the JJA distribution, which has an average higher than all other seasons. The cloud-base heights do not vary from year to year for DJF. In the MAM season of 2000 we find a preponderance of clouds below 500 m, reducing the mean and median to lower cloud-base heights than for all other years and influencing the interannual trends. This observation is confirmed by the Wilcoxon rank-sum test. Closer inspection of cloud occurrence during the MAM (spring) season shows that the earlier part of the 2000 season had a higher occurrence of boundary layer cloud than the latter part of the season when compared to other years. As the MAM season is the transition period between winter and summer, March temperatures are generally colder than for May, with lower LCLs leading to lower clouds. For JJA the cloud bases do not display any significant interannual difference. In SON, though, 1999 has significantly higher bases than for other years, which is confirmed by the Wilcoxon rank-sum test.

b. Cloud liquid water path

Distributions of retrieved cloud liquid water paths for daytime, nighttime, and both periods together are illustrated in Fig. 3. In this figure cloud liquid water paths are truncated at 1 mm; values greater than 1 mm represent 6% of the total dataset. Close inspection of radar reflectivity for the periods with cloud liquid water paths greater than 1 mm generally indicated precipitation at the surface, in which case the microwave radiometer window was most likely wet and the retrieved value of cloud water biased high. Consequently, we neglected these values of cloud liquid water path both in our figures and analysis.

The mean cloud liquid water path of the 4-yr dataset is 0.141 mm with a standard deviation of 0.18 mm, while the median value is 0.067 mm. Daytime median liquid water paths are seen to be lower than the nighttime values by approximately 50%. This observation is confirmed as significant by the Wilcoxon rank-sum test. Miller et al. (1998) provide the following explanation for this observation. During the daytime, solar heating at cloud top stabilizes the cloud layer, thereby suppressing convection and cloud formation. During the nighttime, the absence of solar heating leads to higher convective activity and thicker clouds compared to daytime conditions, thereby leading to higher nighttime cloud liquid water paths compared to the daytime.

If we look at the histograms in Fig. 3, the distributions are highly skewed with a modal value in the lowest bin from 0.00–0.05 mm of water. Because the accuracy of two-channel microwave radiometer retrievals of cloud liquid water path is on the order of 0.01–0.02 mm, the most common value of cloud liquid water path across the 4-yr dataset is on the order of the instrument accuracy. Note that cloud detection was performed using the radar; thus, the uncertainty in the cloud liquid water paths from the microwave radiometer does not affect the number of samples.

Annual distributions of liquid water path for single-layer warm clouds for daytime, nighttime, and all periods are shown in Fig. 3. The most salient feature in these figures is that the nighttime liquid water paths are consistently higher than daytime for all 4 yr. The seasonal probability of occurrence of cloud liquid water path for all boundary layer cloud cases is shown in Fig. 4. The JJA season daytime clouds contain the least liquid water on average (Table 2). The lower liquid water paths in summer may result from increased surface heating leading to lower relative humidities near the surface, as well as increased cloud-top entrainment drying. It is possible that negatively buoyant, and therefore unstable, entrained air at cloud top promotes further entrainment and therefore a reduction in cloud liquid water path within a boundary layer cloud. For this kind of instability to occur the ratio of the difference between above-cloud to in-cloud equivalent potential temperature and total water mixing ratio should lie above a critical value (Randall 1980; Deardorff 1980). Randall (1980) and Deardorff (1980) have suggested a typical critical value of 0.23 while MacVean and Mason (1990) have suggested a more restrictive value of 0.70. Del Genio and Wolf (2000) find that the Randall (1980) and Deardorff (1980) value is exceeded in multiple summertime cases. Also, if the entraining air above cloud top decreases in relative humidity, an increase in the depletion of the cloud liquid water path may occur. Del Genio and Wolf (2000) show that increasing summertime temperatures do lead to a decrease in relative humidity above cloud top. A comparison of the daytime and nighttime averages shows that the diurnal liquid water path variation is higher in spring and summer (MAM and JJA) than fall and winter (SON and DJF).

c. Normalized shortwave cloud forcing

Normalized cloud forcing for the 4-yr period is plotted in Fig. 5 for all the daytime samples for which both a positive retrieved value of cloud liquid water path and a surface flux measurement existed. Most normalized cloud forcing values lie between −0.4 and −0.9 with a secondary peak close to 0.0. The secondary peak is a result of the microwave radiometer and radar detecting a cloud overhead but the direct beam impinging on the shortwave pyranometer with almost no attenuation due to breaks in cloud cover. The mean and median of the cloud forcing are −0.62 and −0.65 (Table 1). These results indicate that approximately 65% of the flux which would have reached the surface on a clear-sky day is either absorbed or reflected back to space in the presence of a warm boundary layer cloud with a typical liquid water path of 0.067 mm. The small differences in interannual values suggest that on a statistical level there were no significant annual variations (Fig. 5). The seasonal normalized cloud forcing (Table 2) follows the trends in daytime cloud liquid water path with JJA having the least negative normalized cloud forcings as a result of the lowest cloud liquid water paths during this season.

d. Summary of the 4-yr climatology

Overall, the cloud liquid water path distribution is highly positively skewed. Nighttime cloud liquid water paths are consistently higher than daytime values irrespective of annual or seasonal comparisons. Comparing seasons, the day–night difference is higher in spring and summer than fall and winter. Clouds that occur in the summer months of June, July, and August contain, on average, the least amount of water. Cloud-base heights had modes below 1 km irrespective of the time of year, with the summer months having consistently higher averages than for the other seasons. The cloud bases were generally lowest in the winter (DJF) months. In 2000 the average cloud base was lower than for other years primarily because the MAM season had a much lower average than for other years. When we consider the normalized cloud forcing statistics, we find that the results follow the liquid water path trends. In particular, there were no significant interannual variations in cloud forcing. We conclude that any particular year of data will be representative of any other year for the period from 1997 to 2000.

4. Comparison of observations with model outputs

Having developed a climatology based on observations, we now compare these observations with output from two weather forecasting models. The first is the ECMWF model, which provides short- and medium-range weather forecasts. The selected domain, centered over the SGP central facility, covers the region from 36°–37°N and 97°–98.3°W with a model resolution of 0.56° by 0.56°. We use forecast outputs with an initialization time of 1200 UTC and hourly forecast verifications 12–36 h from initialization. The ECMWF model outputs represent horizontal averages over the domain as well as time averages over an hour. The ECMWF model has 60 layers of which approximately 18 layers are below 3 km and 15 layers below 2 km, leading to an average layer thicknesses of 166 m and 133 m, respectively. The ECMWF model outputs cover the period from April 1995 to November 2001. For boundary layer clouds the ECMWF model uses a multilayer mass flux scheme and a balance between surface evaporation and shallow cloud moisture flux for closure in the parameterization described by Tiedtke (1993).

The second model considered is the NCEP Eta Model. We use the Eta 4D data assimilation system (EDAS) output for 104 model output location time series (MOLTS) stations, with the relevant outputs located between 31.9°–41.0°N latitude and 91.6°–103.6°W longitude. This analysis dataset covers the period between March 2000 and July 2002. Our previous comparison shows minimal interannual variability in the observations. This result is the basis for our assumption that comparisons with MOLTS model outputs are informative, even though the MOLTS model outputs do not totally overlap in time with the observations. The temporal resolution of the available MOLTS model outputs is 1 h and is similar to the ECMWF model output. The MOLTS outputs are available at 59 layers with a layer distribution similar to that of the ECMWF model. The NCEP Eta Model (Black 1994) uses the cloud parameterization approach described in Zhao and Carr (1997) and Zhao et al. (1997) in which combined shallow and deep cloud parameterizations are used to characterize boundary layer clouds.

As the ARM SGP site has no significant terrain variations, the point measurements at the SGP central facility are representative of the region on a statistical level (Zhong and Doran 1997). For comparison to the model outputs we average the high-resolution liquid water path data to 1-h averages, matching the time resolution of the model outputs. For a particular hour to be included less than 5% of the profiles in the hour could have multilayered cloud profiles. The hourly average was calculated using only the layers with single-layer clouds. The mean liquid water path for 20-s (Table 1) and 1-h (Table 4) averages varies from 0.141 to 0.096 mm. We note that a time varying average based on synoptic conditions (e.g., Barnett et al. 1998) or a probabilistic approach (e.g., Jakob et al. 2004) may provide more readily interpreted results than our fixed-time averaging. Our averaging technique, however, is easily implemented and does provide us with bounds on the parameters of interest in our comparisons.

Our goal is to compare the impact of observed and modeled differences in cloud liquid water path on corresponding surface fluxes at a statistical level. To remove the impact of solar zenith angle, as well as reduce the impact of measurement errors on the comparison, we chose to compare normalized cloud forcing. Models generally report a downwelling surface flux that is dependent on cloud fraction, in addition to reporting cloud liquid water path and solar zenith angle. To remove the impact of differences in model radiation schemes, as well as differences in modeled and observed cloud fractions, we estimated model normalized cloud forcing with a simple parameterization that we developed from observations and applied to model cloud liquid water paths and solar zenith angles. The use of the parameterization to estimate both modeled and observed normalized cloud forcing provides us an assessment of the impact of differences in cloud liquid water path exclusively.

In this section we first present the methodology for inferring cloud liquid water path from variables available in the model outputs. We then describe the parameterization for normalized cloud forcing and present comparisons of observed and modeled cloud liquid water path and normalized cloud forcing. We next retrieve the cloud thicknesses and cloud-base heights using lidar and radar returns and compare them with the model outputs. We emphasize that the observed cloud thickness may be biased high, especially during the summer season, due to insect contamination of the radar returns. Nevertheless, we present them as they still provide an insight into some model features. Finally, we compute layer-averaged cloud liquid water contents from the observed cloud thicknesses and liquid water paths and compare them with the model outputs.

a. Extraction of liquid water path from models

The models report pressure, temperature, specific humidity, cloud fraction, and specific cloud liquid and ice water content at model layer midpoints. We first compute the virtual temperature for each model layer using temperature and specific humidity. We then compute the vertical height and air density for each model layer using the pressure and virtual temperature. By averaging contiguous model layer heights, we obtained the heights of the model levels that bound the model layers. We are not able to compute the lowest level height, that is, the surface, unambiguously and we therefore treat the lowest model layer midpoint as our surface (0 m). This choice of the surface location impacts our model cloud-base height calculations, but has no impact on other computations. As model layer thicknesses are approximately 133 m in the bottom 2 km, the model cloud-base heights in this study are expected to be accurate to better than 133 m.

We converted model specific cloud liquid and ice water contents, which are averages over the domain, to cloudy-air-only values by dividing them by the cloud fraction for that model layer. If a model layer contained a cloud, the boundaries were taken to be at the previously computed levels which bound that model layer. We then used the cloud liquid and ice water contents, that is, the products of specific liquid and ice water contents and air density, to calculate liquid and ice water paths for that particular cloud layer. Model layer cloud-fraction-weighted summation of model liquid and ice water contents over the profile provided us with an estimate of model cloud liquid and ice water paths.

Profiles with only cloud liquid water paths and one cloud layer are used in this study. A model cloud layer is potentially composed of multiple cloudy model layers as long as the layers occur contiguously with height with no cloud-free layers in between. For these cases we assumed that the cloudy model layers were maximally overlapped (e.g., Morcrette and Jakob 2000; Geleyn and Hollingsworth 1979). All model cases for which there were two or more cloudy model layers separated by cloud-free model layers were removed from the study so that the maximum overlap assumption is strictly valid.

b. Parameterization of cloud forcing

In an analysis of ARM SGP site observations Sengupta et al. (2003) found that downwelling irradiances at the surface are primarily dependent on cloud liquid water path, with a much weaker dependence on cloud-drop effective radius. These results motivated us to develop a parameterization of normalized cloud forcing that is based, in part, on cloud liquid water path with no explicit dependence on cloud-drop effective radius. To develop the parameterization we used the 20-s-resolution data that are presented in section 3.

To fit a model of normalized cloud forcing to the observations we first divided the data into 8 equally spaced bins across cosines of the solar zenith angle ranging from cos(13°) to cos(80°). We did not use observations with solar zenith angles greater than 80° in order to avoid the numerous problems that one encounters at these large angles, including the breakdown of assumptions used in plane-parallel radiative transfer calculations and inherent problems in measuring the downwelling irradiance at these large solar zenith angles. We divided the cloud liquid water paths corresponding to each solar zenith angle bin into three groups and used combinations of functions of the form exp(a LWP) and 2/(2 + a LWP) (Bohren 1987) to fit the data in each group. The three equations that emerged from this process are
i1520-0442-17-24-4760-e2
for LWP < 0.015,
i1520-0442-17-24-4760-e3
for 0.015 ≤ LWP < 0.15, and
μ,e[a30(μ)+a31(μ)LWP]
for 0.15 ≤ LWP.
For each of the eight solar zenith angle bins we solve Eqs. (2) and (4) for the regression coefficients using a linear least squares fit, while for Eq. (3) we use the nonlinear Newton–Gauss fitting procedure (Table 3). We made the linear least squares fitting procedure “robust” by incorporating a filter on outliers (Huber 1981), so that large negative values of normalized cloud forcing did not dominate the fit at small liquid water path values. We finally fit polynomial functions to the coefficients obtained by the linear least squares fits to obtain
i1520-0442-17-24-4760-e5
The coefficients produced by the Newton–Gauss fitting procedure were not smooth, so we used the actual regression coefficients for each solar zenith angle bin in Table 3 to estimate the normalized cloud forcing for liquid water paths in this intermediate range.

We tested the parameterization against normalized cloud forcing observations and radiative transfer calculations for 19 overcast stratus cases between January 1997 and January 1998. The radiative transfer calculations used cloud liquid water paths derived from the microwave radiometer and cloud-droplet effective radii derived from 415-nm transmission measurements obtained by the Multifilter Rotating Shadowband Radiometer (Min and Harrison 1996). We found that the parameterization produced a 79% correlation with observations as compared to an 81% correlation for the explicit radiative transfer calculations (Sengupta et al. 2003). A forecast-type skill score assessment (Sengupta 2002) showed that parameterization performance was similar to that of the explicit radiative transfer calculations. Taking this comparison one step further, Sengupta (2002) found that the parameterization performed better than explicit radiative transfer calculations that used a climatological effective radius which was chosen to produce unbiased results when compared to observations.

c. Observed and modeled results

The ECMWF model produces skewed cloud liquid water path distributions with a mean of 0.202 mm (Fig. 6; Table 4). The MOLTS output (Fig. 6; Table 4) also has a skewed distribution but with a higher concentration of smaller liquid water paths than for the ECMWF model, resulting in a comparatively lower mean cloud liquid water path of 0.194 mm. The median values of the ECMWF and MOLTS model outputs, which better represent skewed distributions, are 0.148 and 0.108 mm, respectively. The cloud liquid water path observations from the ARM SGP site microwave radiometer have a mean of 0.096 mm, which is approximately half that of the models, and a median of 0.033 mm, which is approximately one-quarter of the models.

As differences in the cloud liquid water path can arise from differences in either cloud thickness or cloud liquid water content, we compare modeled and observed values for both of these variables in an attempt to identify the source of the cloud liquid water path differences. The 1-h averages of observed cloud thickness were calculated from the difference of the averaged cloud base and top computed from the 20-s data. Individual averages of the 20-s liquid water path and cloud thicknesses were used to compute the 1-h liquid water content. A comparison of cloud thicknesses for the warm boundary layer clouds (Fig. 7; Table 4) shows that the ECMWF model produces thicker clouds, while MOLTS produces thinner clouds, when compared to observations. The modeled liquid water contents for both the ECMWF and MOLTS models are 2–3 times the observed median values (Fig. 8; Table 4). The mean cloud liquid water content from the ECMWF model is about 30% higher than the observations, while it is almost double for MOLTS.

We infer from these results that the difference between modeled and observed liquid water path is primarily a result of much higher liquid water contents in modeled boundary layer clouds, although cloud thickness also has a contribution in the ECMWF model. This implies that the modeled clouds are more adiabatic than observations. One reason for this is the possibility that there is insufficient cloud-top entrainment in the models. We conclude, like Deng et al. (2003), that the ECMWF parameterization (Tiedtke 1989), which uses a balance between surface evaporation and shallow cloud moisture flux, may not be appropriate over land. In the Eta Model, which uses cloud parameterization concepts of Betts and Boers (1990), cloud-top mixing may be underestimated for continental stratus and stratocumulus (Black 1994; Deng et al. 2003).

As we mentioned earlier, the radar-derived cloud thicknesses may be biased high, especially during summer. Even so, both the mean and median cloud thicknesses produced by the ECMWF model are higher than observations (Table 4). For MOLTS, the cloud-thickness averages are slightly lower than for the observations (Table 4) and this is no doubt an artifact of the high bias in the observations. This argument is partly supported by the fact that wintertime observed and MOLTS model cloud-thickness averages are similar (Table 5) while for the other seasons, and especially the summertime, the observed cloud-thickness averages are higher (Tables 5–8).

We also investigated the possibility that models produce cloud fractions that are too low, leading to cloud liquid water contents that are too high, when compared to observations. For the observations we compute the cloud fraction as the proportion of times that the ground-based instruments detect an overhead cloud within each hourly period. For the models we use the maximum cloud fraction from the model layers containing a cloud. The mean and median cloud fractions from the observations are 0.35 and 0.23, respectively. The ECMWF model output mean and median cloud fractions are 0.48 and 0.42, respectively, while the corresponding values for MOLTS are 0.70 and 0.83. As the cloud fractions from the models are higher than for the observations, model cloud fractions are not the source of the observed and modeled cloud liquid water path differences.

An analysis of diurnal variations in cloud liquid water path (Fig. 6; Table 4) shows that the modeled daytime values are lower than for nighttime, similar to the observations. The cloud thicknesses in the models do not show significant diurnal variations, also in agreement with the observations (Fig. 7; Table 4). The liquid water content values are, therefore, lower during the daytime in both the model outputs and the observations (Fig. 8; Table 4).

We found earlier from the observations that there are smaller liquid water paths in the summer (JJA) months than for the other seasons (Fig. 9; Tables 5–8). While the MOLTS model output shows a similar signature, the actual values are much higher with the median being over 5 times that of the observations (Fig. 9; Tables 5– 8). Moreover, the seasonal signature is much stronger in the observations than the MOLTS model output. The ECMWF model results show an opposite signature in the cloud liquid water paths with the JJA median and mean values being higher than for the other seasons (Fig. 9; Tables 5–8). Moreover, boundary layer clouds produced in the ECMWF model are thicker in JJA (Fig. 10; Tables 5–8) compared to observations. Observed liquid water contents (Fig. 11; Tables 5–8) are lowest in summer and highest in winter. The MOLTS model output matches the observed trend, but with values around 2–3 times the observations, while the ECMWF model produces exactly the opposite trend (Fig. 11; Tables 5–8). Again, these differences may be a result of the closure used in the ECMWF parameterization, which may be more appropriate for water than for land (Tiedtke 1989; Deng et al. 2003), and the Eta Model cloud parameterization (Betts and Boers 1990) in which cloud-top mixing may be underestimated for continental stratus and stratocumulus (Black 1994; Deng et al. 2003).

A comparison of cloud-base heights (Fig. 12; Table 4) shows that ECMWF model cloud bases are slightly higher, while the MOLTS cloud bases are much lower, than the observations. Daytime cloud bases are higher than nighttime cloud bases for both the models and observations. A seasonal comparison (Fig. 13; Tables 5– 8) shows cloud-base heights to be much higher in summer (JJA) for both observations and models when compared to the other seasons.

We calculated the impact of the excess cloud liquid water paths in the models on normalized cloud forcing using the parameterization developed in the previous section (Fig. 14; Table 4). To provide an estimate of how the parameterization compares with actual observations we also plot the actual measurements of normalized cloud forcing in Fig. 14a. The mean and median values of the observed normalized cloud forcing are −0.45 and −0.44, respectively, while applying the parameterization to observed cloud liquid water paths yields mean and median values of −0.48. Note that the differences in mean and median normalized cloud forcing between the 20-s (Fig. 5; Table 1) and 1-h (Fig. 4; Table 4) datasets are a result of broken clouds predominantly occurring when cloud liquid water paths are smaller. This predominance of lower liquid water paths corresponding to lower cloud fractions leads to a higher weighting of these cases in 1-h-averaged statistics when compared to the 20-s statistics. The correlation between the parameterized and observed normalized cloud forcing is 0.81. The cloud liquid water paths produced by the ECMWF model yield values of −0.76 and −0.75 for the mean and median normalized cloud forcings, respectively, while the MOLTS model cloud liquid water paths yield values of −0.69 and −0.71 for the mean and median normalized cloud forcings, respectively, which are close to the ECMWF results. The large discrepancies in model cloud liquid water paths lead to large differences in solar transmission.

5. Conclusions

We developed a climatology of cloud liquid water path and cloud-base height for warm boundary layer clouds. We found that nighttime cloud liquid water paths were consistently higher than daytime values. Day to night comparisons showed that spring and summer had larger differences than the other two seasons. On average, the summer (JJA) months had lower cloud liquid water paths than the other seasons. Cloud-base heights were generally below 1 km, with summer months having, on average, higher bases than the other seasons. The winter (DJF) months were found to have the lowest cloud bases. Therefore, higher ambient temperatures do not necessarily lead to larger cloud liquid water paths as the clouds may consistently form at higher altitudes leading to potentially thinner clouds with less liquid water path. There were little to no interannual variations in the data set and statistics from any year were representative of all four years.

We developed a parameterization of normalized cloud forcing from the observations that is dependent of solar zenith angle and cloud liquid water path. This parameterization was found to perform nearly as well as explicit radiative transfer model calculations that used detailed information of cloud and atmospheric properties. While we used the parameterization to assess differences in observed and modeled normalized cloud forcings that resulted from differences in observed and modeled cloud liquid water paths, it can be used as a baseline for testing sophisticated schemes for estimating downwelling shortwave irradiances, as well as a means for providing quick estimates for researchers who need results without running complicated radiative transfer models.

Both the ECMWF and Eta models were found, on average, to have around 4–5 times the median cloud liquid water path of observations. The diurnal trend of lower cloud liquid water paths during the daytime when compared to nighttime is seen in both the observations and models. The observations showed lower than annual average cloud liquid water paths during summer (JJA), which was also the case for the MOLTS model output. The ECMWF model, however, exhibited an opposing trend with summers having higher liquid water paths than other seasons. While the observed cloud thicknesses may be biased high, the ECMWF model output clouds are thicker than the observations. This higher cloud thickness in the ECMWF model output coupled with a larger liquid water content produces the significantly higher liquid water paths when compared to observations. The cloud thickness from the MOLTS data is smaller than observations but the liquid water content appears to overcompensate, resulting in significantly higher liquid water paths. Observed and modelled cloud fraction differences do not compensate for the higher liquid water paths in models, as observed cloud fractions were significantly lower than the cloud fractions from the ECMWF and MOLTS output. The larger cloud liquid water paths in the models led to a significant low bias of approximately 50% in modeled normalized cloud forcing as compared to observations. Such major differences in the redistribution of solar energy between observations and models may potentially have a significant impact on the quality of model predictions.

Acknowledgments

This research was supported by the Office of Biological and Environmental Research of the U.S. Department of Energy under contract numbers DE-AC06-76RL01830 and DE-FG02-90ER61071 as part of the Atmospheric Radiation Measurement Program.

REFERENCES

  • Barnett, T. P., , J. Ritchie, , J. Foat, , and G. Stokes, 1998: On the space– time scales of the surface solar radiation. J. Climate, 11 , 8896.

    • Search Google Scholar
    • Export Citation
  • Betts, A. K., , and R. Boers, 1990: A cloudiness transition in a marine boundary layer. J. Atmos. Sci, 47 , 14801497.

  • Black, T. L., 1994: The new NMC mesoscale Eta Model: Description and forecast examples. Wea. Forecasting, 9 , 265278.

  • Bohren, C. F., 1987: Multiple scattering of light and some of its observable consequences. Amer. J. Phys, 55 , 524533.

  • Clothiaux, E. E., , T. P. Ackerman, , G. G. Mace, , K. P. Moran, , R. T. Marchand, , M. A. Miller, , and B. E. Martner, 2000: Objective determination of cloud heights and radar reflectivities using a combination of active remote sensors at the ARM CART sites. J. Appl. Meteor, 39 , 645665.

    • Search Google Scholar
    • Export Citation
  • Conover, W. J., 1980: Some methods based on ranks. Practical Nonparametric Statistics, Wiley Series in Probability Mathematics and Statistics, John Wiley and Sons, 215–228.

    • Search Google Scholar
    • Export Citation
  • Deardorff, J. W., 1980: Cloud top entrainment instability. J. Atmos. Sci, 37 , 131147.

  • Del Genio, A. D., , and A. B. Wolf, 2000: The temperature dependence of the liquid water path of low clouds in the southern Great Plains. J. Climate, 13 , 34653486.

    • Search Google Scholar
    • Export Citation
  • Del Genio, A. D., , M. S. Yao, , W. Kovari, , and K. K. W. Lo, 1996: A prognostic cloud water parameterization for global climate models. J. Climate, 9 , 345357.

    • Search Google Scholar
    • Export Citation
  • Deng, A., , N. L. Seaman, , and J. S. Kain, 2003: A shallow-convection parameterization for mesoscale models. Part I: Submodel description and preliminary analysis. J. Atmos. Sci, 60 , 3456.

    • Search Google Scholar
    • Export Citation
  • Geleyn, J-F., , and A. Hollingsworth, 1979: An economical analytical method for the computation of the interaction between scattering and line absorption of radiation. Beitr. Phys. Atmos, 52 , 116.

    • Search Google Scholar
    • Export Citation
  • Ghan, S. J., , L. R. Leung, , and J. McCaa, 1999: A comparison of three different modeling strategies for evaluating cloud and radiation parameterizations. Mon. Wea. Rev, 127 , 19671984.

    • Search Google Scholar
    • Export Citation
  • Ghan, S. J., , R. C. Easter, , J. Hudson, , and F. M. Brèon, 2001: Evaluation of aerosol indirect radiative forcing in MIRAGE. J. Geophys. Res, 106 , 53175334.

    • Search Google Scholar
    • Export Citation
  • Huber, P. J., 1981: Regression. Robust Statistics, Wiley Series in Probability Mathematics and Statistics, John Wiley and Sons, 153–198.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., , R. Pincus, , C. Hannay, , and K-M. Xu, 2004: Use of cloud radar observations for model evaluation: A probabilistic approach. J. Geophys. Res.,109, 3202, doi:10.1029/ 2003JD003473.

    • Search Google Scholar
    • Export Citation
  • Kato, S., , T. P. Ackerman, , E. E. Clothiaux, , J. H. Mather, , G. G. Mace, , M. L. Wesely, , F. Murcray, , and J. Michalsky, 1997: Uncertainties in modeled and measured clear-sky surface shortwave irradiances. J. Geophys. Res, 102 , 2588125898.

    • Search Google Scholar
    • Export Citation
  • Liljegren, J. C., 1994: Two-channel microwave radiometer for observations of total column precipitable water vapor and cloud liquid water path. Preprints, Fifth Symp. on Global Change Studies, Nashville, TN, Amer. Meteor. Soc., 262–269.

    • Search Google Scholar
    • Export Citation
  • Liljegren, J. C., , E. E. Clothiaux, , G. G. Mace, , S. Kato, , and X. Dong, 2001: A new retrieval for cloud liquid water path using a ground-based microwave radiometer and measurements of cloud temperature. J. Geophys. Res, 106 , 1448514500.

    • Search Google Scholar
    • Export Citation
  • Long, C. N., , and T. P. Ackerman, 2000: Identification of clear skies from broadband pyranometer measurements and calculation of downwelling shortwave cloud effects. J. Geophys. Res, 105 , 1560915626.

    • Search Google Scholar
    • Export Citation
  • MacVean, M. K., , and P. J. Mason, 1990: Cloud-top entrainment instability through small-scale mixing and its parameterization in numerical models. J. Atmos. Sci, 47 , 10121030.

    • Search Google Scholar
    • Export Citation
  • Miller, M. A., , M. P. Jensen, , and E. E. Clothiaux, 1998: Diurnal cloud and thermodynamic variations in the stratocumulus transition regime: A case study using in situ and remote sensors. J. Atmos. Sci, 55 , 22942310.

    • Search Google Scholar
    • Export Citation
  • Min, Q., , and L. C. Harrison, 1996: Cloud properties derived from surface MFRSR measurements and comparison with GOES results at the ARM SGP site. Geophys. Res. Lett, 23 , 16411644.

    • Search Google Scholar
    • Export Citation
  • Morcrette, J-J., , and C. Jakob, 2000: The response of the ECMWF model to changes in the cloud overlap assumption. Mon. Wea. Rev, 128 , 17071732.

    • Search Google Scholar
    • Export Citation
  • Randall, D. A., 1980: Conditional instability of the first kind upside-down. J. Atmos. Sci, 37 , 125130.

  • Sengupta, M., 2002: Radiative impact of continental stratus at the Southern Great Plains: A climatological perspective. Ph.D. thesis, The Pennsylvania State University, 130 pp. [Available online at http://etda.libraries.psu.edu/theses/approved/WorldWideFiles/ ETD-179/phdthesis.pdf.].

    • Search Google Scholar
    • Export Citation
  • Sengupta, M., , E. E. Clothiaux, , T. P. Ackerman, , S. Kato, , and Q. Min, 2003: Importance of accurate liquid water path for estimation of solar radiation in warm boundary layer clouds: An observational study. J. Climate, 16 , 29973009.

    • Search Google Scholar
    • Export Citation
  • Tiedtke, M., 1989: A comprehensive mass flux scheme of cumulus parameterization in large scale models. Mon. Wea. Rev, 117 , 17791800.

  • Tiedtke, M., 1993: Representation of clouds in large-scale models. Mon. Wea. Rev, 121 , 30403061.

  • Zhao, Q., , and F. H. Carr, 1997: A prognostic scheme for operational NWP models. Mon. Wea. Rev, 125 , 19311953.

  • Zhao, Q., , T. L. Black, , and M. E. Baldwin, 1997: Implementation of the cloud prediction scheme in the Eta Model at NCEP. Wea. Forecasting, 12 , 697712.

    • Search Google Scholar
    • Export Citation
  • Zhong, S., , and J. C. Doran, 1997: A study of the effects of spatially varying fluxes on cloud formation and boundary layer properties using data from the Southern Great Plains Cloud and Radiation Testbed. J. Climate, 10 , 327341.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Histograms of daytime, nighttime, and all cloud-base heights from observations (left). The consolidated 4-yr dataset, while (right) data are divided into years

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 2.
Fig. 2.

Histograms of seasonal cloud-base height from observations

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 3.
Fig. 3.

Histograms of daytime, nighttime, and all cloud liquid water paths from observations. (left) The consolidated 4-yr dataset, while (right) the data are divided into years

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 4.
Fig. 4.

Histograms of seasonal cloud liquid water path from observations

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 5.
Fig. 5.

Histograms of 4-yr consolidated, annual, and seasonal normalized cloud forcing obtained from observations

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 6.
Fig. 6.

Histograms of cloud liquid water path occurrence from observations and the ECMWF and MOLTS models for daytime, nighttime, and all cases

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 7.
Fig. 7.

Histograms of cloud-thickness occurrence from observations and the ECMWF and MOLTS models for daytime, nighttime, and all cases

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 8.
Fig. 8.

Histograms of cloud liquid water content occurrence from observations and the ECMWF and MOLTS models for daytime, nighttime, and all cases

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 9.
Fig. 9.

Histograms of seasonal cloud liquid water path occurrence from observations and the ECMWF and MOLTS models for all cases

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 10.
Fig. 10.

Histograms of seasonal cloud-thickness occurrence from observations and the ECMWF and MOLTS models for all cases

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 11.
Fig. 11.

Histograms of seasonal cloud liquid water content occurrence from observations and the ECMWF and MOLTS models for all cases

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 12.
Fig. 12.

Histograms of cloud-base height occurrence from observations and the ECMWF and MOLTS models for daytime, nighttime, and all cases

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 13.
Fig. 13.

Histograms of seasonal cloud-base height occurrence from observations and the ECMWF and MOLTS models for all cases

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Fig. 14.
Fig. 14.

Histograms of normalized cloud forcing occurrence from observations and the ECMWF and MOLTS models

Citation: Journal of Climate 17, 24; 10.1175/JCLI-3231.1

Table 1.

The annual distribution of cloud liquid water path, cloud-base height, and cloud forcing from all cases of single-layer warm clouds for the period 1997–2000. Note that the actual daytime LWP points (197 145) are reduced to 166 286 points for cloud forcing results due to intermittent missing flux measurements

Table 1.
Table 2.

The seasonal distribution of cloud liquid water path, cloud-base height, and cloud forcing from all cases of single-layer warm boundary layer clouds for the period 1997–2000

Table 2.
Table 3.

Regression coefficients for the three different equations that we used to cover the range of observed cloud liquid water paths

Table 3.
Table 4.

Comparison of observed cloud liquid water path, normalized cloud forcing, cloud thickness, cloud-base height, and cloud liquid water content with output from the ECMWF and MOLTS models for all cases. Observed normalized cloud forcing is reported both from flux measurements and the parameterization using obs cloud liquid water paths.

Table 4.
Table 5.

Comparison of observed cloud liquid water path, normalized cloud forcing, cloud thickness, cloud-base height, and cloud liquid water content with output from the ECMWF and MOLTS models for DJF (winter) season

Table 5.
Table 6.

Comparison of observed cloud liquid water path, normalized cloud forcing, cloud thickness, cloud-base height, and cloud liquid water content with output from the ECMWF and MOLTS models for MAM (spring) season

Table 6.
Table 7.

Comparison of observed cloud liquid water path, normalized cloud forcing, cloud thickness, cloud-base height, and cloud liquid water content with output from the ECMWF and MOLTS models for JJA (summer) season

Table 7.
Table 8.

Comparison of observed cloud liquid water path, normalized cloud forcing, cloud thickness, cloud-base height, and cloud liquid water content with output from the ECMWF and MOLTS models for SON (fall) season

Table 8.
Save