• Arakawa, A., , and V. R. Lamb, 1977: Computational design of the basic dynamical processes of the UCLA general circulation model. Methods Comput. Phys, 17 , 174264.

    • Search Google Scholar
    • Export Citation
  • Avissar, R., , and Y. Mahrer, 1988: Mapping frost-sensitive areas with a three-dimensional local-scale numerical model. Part I: Physical and numerical aspects. J. Appl. Meteor, 27 , 400413.

    • Search Google Scholar
    • Export Citation
  • Briegleb, B. P., , P. Minnis, , V. Ramanathan, , and E. Harrison, 1986: Comparison of regional clear-sky albedos inferred from satellite observations and model calculations. J. Climate Appl. Meteor, 25 , 214226.

    • Search Google Scholar
    • Export Citation
  • Chen, C., , and W. R. Cotton, 1983: A one-dimensional simulation of the stratocumuluscapped mixed layer. Bound.-Layer Meteor, 25 , 289321.

    • Search Google Scholar
    • Export Citation
  • Clark, T. L., 1977: A small-scale dynamic model using a terrain-following coordinate transformation. J. Comput. Phys, 24 , 186215.

  • Clark, T. L., , and R. D. Farley, 1984: Severe downslope windstorm calculations in two and three spatial dimensions using anelastic interactive grid nesting: A possible mechanism for gustiness. J. Atmos. Sci, 41 , 329350.

    • Search Google Scholar
    • Export Citation
  • Cotton, W. R., , and G. J. Tripoli, 1978: Cumulus convection in shear flow—three-dimensional numerical experiments. J. Atmos. Sci, 35 , 5031521.

    • Search Google Scholar
    • Export Citation
  • Cotton, W. R., , G. J. Tripoli, , R. M. Rauber, , and E. A. Mulvihill, 1986: Numerical simulation of the effects of varying ice crystal nucleation rates and aggregation processes on orographic snowfall. J. Climate Appl. Meteor, 25 , 16581680.

    • Search Google Scholar
    • Export Citation
  • Cotton, W. R., , J. F. Weaver, , and B. A. Beitler, 1995: An unusual summertime downslope wind event in Fort Collins, Colorado, on July 1993. Wea. Forecasting, 10 , 786797.

    • Search Google Scholar
    • Export Citation
  • Deardorf, J. W., 1980: Stratocumulus-capped mixed layers derived from a three-dimensional model. Bound.-Layer Meteor, 18 , 495527.

  • Garatuza-Payan, J., , R. T. Pinker, , and W. J. Shuttleworth, 2001: High-resolution cloud observations for northwestern Mexico from GOES-7 satellite observations. J. Atmos. Oceanic Technol, 18 , 3955.

    • Search Google Scholar
    • Export Citation
  • Harrington, J. Y., 1997: The effects of radiative and microphysical processes on simulated warm and transition season arctic stratus. Department of Atmospheric Science Paper 637, Colorado State University, Fort Collins, CO, 289 pp.

    • Search Google Scholar
    • Export Citation
  • Louis, J. F., 1979: A parametric model of vertical eddy fluxes in the atmosphere. Bound.-Layer Meteor, 17 , 187202.

  • Manabe, R., , J. Stouffer, , M. J. Spelman, , and K. Bryan, 1991: Transient responses of a coupled ocean–atmosphere model to gradual changes of atmospheric CO2. Part I: Annual mean response. J. Climate, 4 , 785818.

    • Search Google Scholar
    • Export Citation
  • Manton, M. J., , and W. R. Cotton, 1977: Parameterizations of the atmospheric surface layer. J. Atmos. Sci, 34 , 331334.

  • McNider, R. T., , and R. A. Pielke, 1981: Diurnal boundary-layer development over sloping terrain. J. Atmos. Sci, 38 , 21982212.

  • Meyers, M. P., , and W. R. Cotton, 1992: Evaluation of the potential for wintertime quantitative precipitation forecasting over mountainous terrain with an explicit cloud model. Part I: Two-dimensional sensitivity experiments. J. Appl. Meteor, 31 , 2650.

    • Search Google Scholar
    • Export Citation
  • Meyers, M. P., , P. J. Demott, , and W. R. Cotton, 1991: New primary ice-nucleation parameterizations in an explicit cloud model. J. Appl. Meteor, 30 , 708721.

    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., and and Coauthors, 1992: A comprehensive meteorological modeling system—RAMS. Meteor. Atmos. Phys, 49 , 6991.

  • Pinker, R. T., , and J. A. Ewing, 1985: Modeling surface solar radiation: Model formulation and validation. J. Climate Appl. Meteor, 24 , 389401.

    • Search Google Scholar
    • Export Citation
  • Pinker, R. T., , and I. Laszlo, 1992: Modeling surface solar irradiance for satellite applications on global scale. J. Appl. Meteor, 31 , 194212.

    • Search Google Scholar
    • Export Citation
  • Pinker, R. T., , W. P. Kustas, , I. Laszlo, , M. S. Moran, , and A. R. Huete, 1994: Basin-scale solar irradiance estimates in semiarid regions using GOES 7. Water Resour. Res, 30 , 13751386.

    • Search Google Scholar
    • Export Citation
  • Randall, D. A., , J. Coakley, , C. Fairall, , R. Kropfli, , and D. Lenschow, 1984: Outlook for research on marine subtropical stratocumulus clouds. Bull. Amer. Meteor. Soc, 65 , 12901301.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., , and L. C. Garder, 1993: Cloud detection using satellite measurements of infrared and visible radiances for ISCCP. J. Climate, 6 , 23412369.

    • Search Google Scholar
    • Export Citation
  • Slingo, A., 1990: Sensitivity of the Earth's radiation budget to changes in low clouds. Nature, 343 , 4951.

  • Stephens, G. L., 1978: Radiation profiles in extended water clouds. II: Parameterization schemes. J. Atmos. Sci, 35 , 21112122.

  • Stephens, G. L., 1984: The parameterization of radiation for numerical weather prediction and climate models. Mon. Wea. Rev, 112 , 826837.

    • Search Google Scholar
    • Export Citation
  • Tremback, C. L., , and R. Kessler, 1985: A surface temperature and moisture parameterization for use in mesoscale numerical models. Preprints, Seventh Conf. on Numerical Weather Prediction, Montreal, PQ, Canada, Amer. Meteor. Soc., 355–358.

    • Search Google Scholar
    • Export Citation
  • Tremback, C. L., , G. J. Tripoli, , and W. R. Cotton, 1985: A regional scale atmospheric numerical model including explicit moist physics and a hydrostatic time-split scheme. Preprints, Seventh Conf. on Numerical Weather Prediction, Montreal, PQ, Canada, Amer. Meteor. Soc., 433–434.

    • Search Google Scholar
    • Export Citation
  • Tripoli, G. J., , and W. R. Cotton, 1980: A numerical investigation of several factors contributing to the observed variable intensity of deep convection over south Florida. J. Appl. Meteor, 19 , 10371063.

    • Search Google Scholar
    • Export Citation
  • Tripoli, G. J., , and W. R. Cotton, 1982: The Colorado State University three-dimensional cloud/mesoscale model–1982. Part I: General theoretical framework and sensitivity experiments. J. Rech. Atmos, 16 , 185219.

    • Search Google Scholar
    • Export Citation
  • Tripoli, G. J., , and W. R. Cotton, 1986: An intense, quasi-steady thunderstorm over mountainous terrain. Part IV: Three-dimensional numerical simulation. J. Atmos. Sci, 43 , 896914.

    • Search Google Scholar
    • Export Citation
  • Walko, R. L., , C. J. Tremback, , R. A. Pielke, , and W. R. Cotton, 1995: An interactive nesting algorithm for stretched grids and variable nesting ratios. J. Appl. Meteor, 34 , 994999.

    • Search Google Scholar
    • Export Citation
  • Whitlock, C. H., and and Coauthors, 1995: First global WCRP shortwave surface radiation budget dataset. Bull. Amer. Meteor. Soc, 76 , 905922.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    The study area modeled in the RAMS model and the location of the AZMET field sites and the single experimental sites near Tombstone, AZ, where the data used in this study were collected

  • View in gallery

    (a) The clear-sky composite image for 16 Jul 1999, and frequency distribution of observed radiance in the clear-sky composite image for the whole image area at 1600 UTC and at 2200 UTC. (b) The GOES visible image together with the derived cloudy-sky and clear-sky radiance and cloud fraction derived by the algorithm at 2200 UTC 14 Jul 1999

  • View in gallery

    Cloud-cover distributions at three half-hourly intervals starting 1700 14 Jul 1999. Columns 1–3 show the modeled cloud cover in RAMS immediately before a new ingestion of cloud-cover data with previous cloud data having been ingested at 30, 15, and 1-min intervals, respectively. Column 4 shows the observed cloud-cover image that will be ingested, while column 5 shows the modeled cloud field given by the RAMS model without any cloud-cover ingestion

  • View in gallery

    Incoming surface solar radiation observed at the AZMET sites compared with the equivalent modeled values with and without cloud-cover ingestion during the period 14–16 Jul 1999

  • View in gallery

    Incoming surface solar radiation observed at the AZMET sites compared with the equivalent modeled values with and without cloud-cover ingestion during the period 22–23 Jun 2000

  • View in gallery

    Modeled hourly average daytime surface solar radiation for model simulations with and without cloud ingestion: (a) hours that are cloudy, and (b) hours that are clear on the basis of satellite observations for 14–16 Jul 1999; (c), (d) Equivalent to (a) and (b), respectively, but for the period 22–23 Jun 2000. In each case, the correlation coefficient, root-mean-square error, and mean bias are given

  • View in gallery

    Modeled hourly average incoming radiation fluxes observed at the experimental field site near Tombstone, AZ, during the period 22–23 Jun 2000, together with modeled values with and without cloud-cover ingestion. (a) The incoming shortwave radiation, and (b) the longwave radiation when the Harrington radiation scheme was used to calculate both shortwave and longwave radiation. (c) Longwave radiation calculated when the Chen–Cotton longwave radiation scheme is used. The fractional cloud cover derived from satellite observations is also shown in (b) and (c)

  • View in gallery

    Precipitation observed at the AZMET sites compared with the equivalent modeled values with and without cloud-cover ingestion during the period 14–16 Jul 1999

  • View in gallery

    Precipitation observed at the AZMET sites compared with the equivalent modeled values with and without cloud-cover ingestion during the period 22–23 Jun 2000

  • View in gallery

    Site-average modeled and observed precipitation are compared for 14–16 Jul 1999 and 22–23 Jun 2000

  • View in gallery

    Forecast fractional cloud cover compared with observed cloud fraction derived from the GOES visible images at 15-min intervals during the 4.5-h forecast period

  • View in gallery

    Success rate of the RAMS model, following cloud ingestion, calculated at 15-min intervals during a 6.5-h forecast period for the whole domain and for the cloudy-only and clear-only subdomains

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 7 7 2
PDF Downloads 5 5 2

Impact of Ingesting Satellite-Derived Cloud Cover into the Regional Atmospheric Modeling System

View More View Less
  • 1 Department of Hydrology and Water Resources, The University of Arizona, Tucson, Arizona
  • 2 Department of Meteorology, University of Maryland, College Park, College Park, Maryland
  • 3 Department of Hydrology and Water Resources, The University of Arizona, Tucson, Arizona, and Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado
  • 4 Department of Hydrology and Water Resources, The University of Arizona, Tucson, Arizona
© Get Permissions
Full access

Abstract

This study investigates the extent to which assimilating high-resolution remotely sensed cloud cover into the Regional Atmospheric Modeling System (RAMS) provides an improved regional diagnosis of downward short- and longwave surface radiation fluxes and precipitation. An automatic procedure was developed to derive high-resolution (4 km × 4 km) fields of fractional cloud cover from visible band Geostationary Operational Environmental Satellite (GOES) data using a tracking procedure to determine the clear-sky composite image. Initial studies, in which RAMS surface shortwave radiation fluxes were replaced by estimates obtained by applying satellite-derived cloud cover in the University of Maryland Global Energy and Water Cycle Experiment's Surface Radiation Budget (UMD GEWEX/SRB) model, revealed problems associated with inconsistencies between the revised solar radiation fields and the RAMS-calculated incoming longwave radiation and precipitation fields. Consequently, in this study, the relationship between cloud albedo, optical depth, and water/ice content used in the UMD GEWEX/SRB model was applied instead to provide estimates of whole-column cloud water/ice that were ingested into RAMS. This potentially enhances the realism of the modeled short- and longwave radiation and precipitation. The ingested cloud image took the horizontal distribution of clouds from the satellite image but derives its vertical distribution from the fields simulated by RAMS in the time step immediately prior to assimilation. The resulting image was ingested every minute, with linear interpolation used to derive the 1-min cloud images between 15-min GOES samples. Comparisons were made between modeled and observed data taken from the Arizona Meteorological Network (AZMET) weather station network in southern Arizona for model runs with and without cloud ingestion. Cloud ingestion was found to substantially improve the ability of the RAMS model to capture temporal and spatial variations in surface fields associated with cloud cover. An initial test suggests that cloud ingestion enhanced RAMS short-term forecast ability.

Corresponding author address: Ismail Yucel, Department of Hydrology and Water Resources, The University of Arizona, Harshbarger Building 11, Tucson, AZ 85721. Email: yucel@hwr.arizona.edu

Abstract

This study investigates the extent to which assimilating high-resolution remotely sensed cloud cover into the Regional Atmospheric Modeling System (RAMS) provides an improved regional diagnosis of downward short- and longwave surface radiation fluxes and precipitation. An automatic procedure was developed to derive high-resolution (4 km × 4 km) fields of fractional cloud cover from visible band Geostationary Operational Environmental Satellite (GOES) data using a tracking procedure to determine the clear-sky composite image. Initial studies, in which RAMS surface shortwave radiation fluxes were replaced by estimates obtained by applying satellite-derived cloud cover in the University of Maryland Global Energy and Water Cycle Experiment's Surface Radiation Budget (UMD GEWEX/SRB) model, revealed problems associated with inconsistencies between the revised solar radiation fields and the RAMS-calculated incoming longwave radiation and precipitation fields. Consequently, in this study, the relationship between cloud albedo, optical depth, and water/ice content used in the UMD GEWEX/SRB model was applied instead to provide estimates of whole-column cloud water/ice that were ingested into RAMS. This potentially enhances the realism of the modeled short- and longwave radiation and precipitation. The ingested cloud image took the horizontal distribution of clouds from the satellite image but derives its vertical distribution from the fields simulated by RAMS in the time step immediately prior to assimilation. The resulting image was ingested every minute, with linear interpolation used to derive the 1-min cloud images between 15-min GOES samples. Comparisons were made between modeled and observed data taken from the Arizona Meteorological Network (AZMET) weather station network in southern Arizona for model runs with and without cloud ingestion. Cloud ingestion was found to substantially improve the ability of the RAMS model to capture temporal and spatial variations in surface fields associated with cloud cover. An initial test suggests that cloud ingestion enhanced RAMS short-term forecast ability.

Corresponding author address: Ismail Yucel, Department of Hydrology and Water Resources, The University of Arizona, Harshbarger Building 11, Tucson, AZ 85721. Email: yucel@hwr.arizona.edu

1. Introduction

Mesoscale meteorological models contribute to the understanding of lower atmosphere processes and are capable of diagnosing and, to a lesser extent, predicting weather variables from the synoptic scale to the microscale. High spatial resolution short-term predictions with such models are particularly important in regions where weather conditions can vary greatly over short distances and where local weather-related hazards, such as flash floods, can cause loss of life and/or damage. Clouds play a critical role in the atmospheric energy balance and the prediction of weather. They have a strong effect on solar heating by reflecting solar radiation back into space, and on the thermal cooling by intercepting infrared radiation emitted by the underlying earth and atmosphere and re-emitting part of this energy back to the surface. The presence of overlying clouds is, of course, also linked strongly to precipitation. Thus, clouds provide the primary link between the two surface energy exchange processes that most influence regional climate, namely the exchange of shortwave and longwave radiation and the exchange of water.

However, the parameterization of clouds is one of the largest sources of uncertainty in coupled atmosphere–land surface models. This uncertainty is exacerbated by the great disparity between the spatial scale of clouds, typically 100 meters to a few kilometers, and model grid scale, typically a few tens to a few hundred kilometers. Mesoscale atmospheric models do predict clouds, but model estimates of their spatial distribution and radiative characteristics may have considerable error. Manabe et al. (1991) pointed out that cloud-related weakness in modeled surface radiation produces markedly unrealistic surface fluxes of energy and water. Small perturbations in the amount or radiative properties of clouds can strongly affect climate (Randall et al. 1984; Slingo 1990), and weakness in simulating these cloud properties is currently a major problem in numerical weather prediction models. On the other hand, measurements of the extent and optical properties of clouds using earth satellites have the potential to supplement surface observations and to improve weather simulation by models.

The goals of the research reported here were (a) to investigate the feasibility of deriving high-resolution, frequent estimates of cloud cover from Geostationary Operational Environmental Satellite (GOES) data and ingesting these estimates into a typical mesoscale model, and (b) by making comparison with field data, to study the effect of ingesting remotely sensed cloud cover into the quality of analyzed fields given by the mesoscale model. An automatic procedure was developed to derive high-resolution fields of cloud properties from visible band GOES data using a tracking procedure to determine the clear-sky composite image. Using multichannel cloud retrieval algorithms based on International Satellite Cloud Climatology Project (ISCCP) or National Oceanic and Atmospheric Administration's National Environmental Satellite, Data, and Information Service (NOAA/NESDIS) data were deemed to be too coarse for this application. The cloudy- and clear-sky radiances and cloud fraction derived from this algorithm were used in the University of Maryland (UMD) Global Energy and Water Cycle Experiment's Surface Radiation Budget (GEWEX/SRB) model to retrieve cloud properties of optical depth and cloud water/ice. Vertically integrated cloud water/ice fields were ingested directly into the Regional Atmospheric Modeling System (RAMS) to update modeled cloud cover. To demonstrate the impact on the model's ability to diagnose surface radiation and precipitation fields, comparisons were made between modeled and observed data for model runs with and without cloud ingestion for 14–16 July 1999 and 22–23 June 2000. These periods were selected because they had significant, changing cloud cover. Finally, an initial test was made of the improvement in short-term predictions of cloud cover given by ingesting cloud data into RAMS prior to predictive runs.

2. Model description

a. The RAMS model

Version 3b of the Colorado State University RAMS model (Pielke et al. 1992) was used in this study. The model is nonhydrostatic (Tripoli and Cotton 1980) with a terrain-following coordinate system (Clark 1977; Tripoli and Cotton 1982) and includes telescoping, interactive, nested grids (Clark and Farley 1984; Walko et al. 1995) and several alternative turbulent closure schemes (Deardorf 1980; McNider and Pielke 1981; Tripoli and Cotton 1986). Comprehensive testing, tuning, and development of the RAMS representation of surface–atmosphere interactions and of atmospheric convection and radiation processes have been carried out (Tripoli and Cotton 1982; Chen and Cotton 1983; Tremback et al. 1985; Pielke et al. 1992).

The RAMS was operated at a 4-km resolution with a single 400 km × 400 km domain that covered the semiarid region of interest in southern Arizona. The model was initiated and its lateral boundaries then nudged every 6 h by atmospheric fields synthesized by the National Centers for Environmental Prediction's (NCEP)s Eta Model.1 The Earth Resources Observation Systems (EROS) Data Center Distributed Active Archive Center (EDC DAAC) 1-km surface cover dataset was used to characterize vegetation across the modeled domain. The primary, relevant features of RAMS as applied during this study are given in Table 1.

The explicit parameterization of cloud microphysics was originally developed by Cotton et al. (1986) and applied in a range of weather and geographical conditions (Manton and Cotton 1977; Cotton and Tripoli 1978; Meyers et al. 1991; Meyers and Cotton 1992). In the model, the mixing ratios of seven different types of hydrometeors are described within clouds—specifically, cloud water, rainwater, pristine ice, aggregate ice, snow, graupel, and hail. Cloud water and water vapor, along with potential temperature, air temperature, and pressure, are derived diagnostically from the prognostic values of ice–liquid water potential temperature (θil), dry air density, total water, rainwater, and ice water (Tripoli and Cotton 1982). At temperatures warmer than the homogeneous ice nucleation temperature (i.e., TH = 233.16 K), cloud water and water vapor mixing ratios are diagnosed as
i1520-0493-130-3-610-e1
where rt, rr, rice, and rvs are the mixing ratio of total water, rainwater, ice water, and the saturated water vapor, respectively. At temperatures below TH, all diagnosed cloud water freezes to ice (rice), thus
i1520-0493-130-3-610-e3
In the previous equations, the ice mixing ratio and rainwater mixing ratio are predicted by applying time-dependent mass continuity equations, which include advective, diffusive, and source and sink terms, the latter describing the rates of conversion of water between phases and hydrometeor types.

In this study, radiative fluxes were initially calculated using the (optional) three-band scheme developed by Chen and Cotton (1983). However, with this scheme, the calculated incoming surface solar radiation was found to be unrealistically low whenever RAMS diagnosed the presence of clouds. For this reason, the improved scheme of Harrington (1997) that uses an eight-band, two-stream radiation transfer code was adopted and used in preference. This two-stream model recognizes scattering and absorption in the atmosphere in terms of an asymmetry factor, single scattering albedo, cloud optical depth, and the geometric thickness of atmospheric layers. These are applied in both the shortwave and longwave parameterizations using appropriate reflectances and emittances. The total hydrometeor optical depth at each level is calculated by summing the extinction coefficients for each microphysical cloud water/ice field, these being based on hydrometeor concentration, a characteristic diameter, the thickness of the level, and fitted extinction coefficients. However, unlike in the Harrington scheme, the Chen–Cotton scheme parameterizes optical characteristics of clouds based on an empirical relationship derived by Stephens (1978, 1984).

b. The UMD GEWEX/SRB model

In this study, the UMD GEWEX/SRB algorithm (Whitlock et al. 1995) was used to infer the downward surface solar radiation and cloud water/ice (via optical depth) from satellite observations. The Delta-Eddington two-stream radiative transfer model (Pinker and Ewing 1985; Pinker and Laszlo 1992) was applied to generate lookup tables that contain atmospheric reflectance and transmittance over a plane-parallel, nonreflecting surface in five spectral intervals (0.2–0.4, 0.4–0.5, 0.5–0.6, 0.6–0.7, and 0.7–4.0 μm). It is assumed that there is no water vapor absorption at wavelengths less than 0.7 μm, and that clouds are therefore nonabsorbent for such wavelengths. Optical thickness parameterization is based on calculations made using a set of eight different “standard” cloud types. Stephens (1978, 1984) derived the least-squares-fitted lines describing the broadband optical thickness of these eight cloud types for two spectral regions (0.3–0.75 μm and 0.75–4 μm) as a function of total liquid water path (LWP).

The satellite algorithm is based on deriving a relationship between the broadband, top of the atmosphere albedo, and atmospheric transmissivity taken from precalculated lookup tables derived for discrete values of the solar zenith angle, water vapor, ozone, and aerosol loading, and most importantly, for different cloud optical thickness. The surface spectral albedo is retrieved as a function of surface type and satellite-derived clear-sky composite radiance by combining reference albedo models (Briegleb et al. 1986). The required atmospheric precipitable water estimates are taken from the Eta Model, with an adjustment for the topography used in the RAMS model. In each spectral interval, upward and downward fluxes at the top of the atmosphere and surface, and the optical depth of the atmosphere are determined from the spectral transmittance. The latter is obtained from the top-of-the-atmosphere shortwave reflectance, which is estimated from the radiance in the (narrow) wavelength bands at which satellite observations are made using a transformation between narrow waveband radiance and shortwave reflectance.

3. Study area and data

a. Field data

The study area comprises the region 31.5°–36°N and 108°–115°W in southern Arizona. The semiarid environment that typifies this study site is due both to its location relative to the equator and the orographic drying effect of mountain ranges in the western United States. The region is characterized by rugged mountain ranges separated by flat valley floors. The limited annual rainfall of less than 400 mm occurs mainly as convective thunderstorms in a summer monsoon season and as frontal storms during the winter. Daily temperature changes can be high (about 20°C).

Observations are taken from the Arizona Meteorological Network (AZMET) system. The measurement stations are often installed adjacent to regions of irrigated agriculture and the data may, in part, be influenced by this fact. This study used data from 17 individual stations that are concentrated near the center of the southern Arizona study area (Fig. 1). Information on the location and elevation of these sites is provided in Table 2. The data in this study were hourly measurements of downward surface solar radiation and precipitation. Solar radiation is measured using a Licor LI200S pyronometer, while precipitation is recorded using RG2501 tipping bucket rain gauges. More detailed information on the AZMET observational network is available at http://ag.arizona.edu/azmet/.

A field experiment was set up to provide additional measurements of downward longwave radiation fluxes for comparison with model-generated fluxes because these were not available from the AZMET network. The experiment site is located at 32.05°N, 110.50°W, near Tombstone in southern Arizona (Fig. 1). A Kipp and Zonen radiometer comprising two pyrogeometers and two pyronometers was deployed at the site, which is capable of simultaneously measuring all four components of the radiation (upwelling and downwelling solar and upwelling and downwelling longwave radiation). Because data collection with this instrument started on 15 June 2000, data were only available for the study period of 22–23 June 2000.

b. Satellite data

The primary satellite data used for this study were the reflected visible (0.55–0.75 μm) radiances observed by the imaging visible/infrared spin-scan radiometer (VISSR) on GOES-10 (or GOES-West) positioned at longitude 135°W above the equator. The VISSR provides visible images with a resolution of about 1-km every 15 min. The raw data were originally stored by NOAA/NESDIS, but these data were subsequently converted to calibrated “albedo,” unadjusted for direction angle by the National Aeronautics and Space Administration (NASA). A computer code was developed and applied to analyze and quality control the data by filtering any random noise in the satellite images (e.g., noisy lines in the images are filtered and replaced by the nearby good lines).

4. Cloud screening algorithm

In this study, an automatic procedure was developed to derive high-resolution (4 km × 4 km) fields of cloud cover and associated cloudy- and clear-sky radiance maps from visible band GOES data. The method is an extension of that used by Garatuza et al. (2000), but it differs in that it uses a new real-time tracking procedure to determine the required clear-sky composite images rather than relying on retrospective derivation of monthly average threshold indexes. Cloud detection involves partitioning radiance samples from within an array of 4 × 4 contiguous pixels into three categories. The decision rules used to discriminate between these groups are threshold statements. Thus, an individual pixel was designated to be “cloudy” if the radiance is greater than a threshold value (e.g., 35% reflectance). Alternatively, an individual pixel may be designated “clear” if the radiance is less than a specified value (in the present study, the radiance of a running mean plus the standard deviation of clear-sky values was used). Pixels that are neither cloud-covered nor clear were designated “mixed”. Using this approach minimizes false cloud detection, but thin clouds may be missed if their observed radiances closely resemble those for pixels with no clouds.

a. Deriving clear-sky composite images

The automatic procedure required specification of acceptance windows, values of clear-sky radiance, and the standard deviation of the clear-sky radiance, which were initially determined from among the preexisting clearest GOES observations. The value of the acceptance window (here called window) defines the number of radiance units above the running mean for which some cloudy sky is assumed. This value determines whether the currently observed value of the radiance is accepted and used to update the running estimate of clear-sky radiance, and it was set to the estimated random noise given by the satellite sensor. If the current value is deemed acceptable, the real-time tracking procedure then updates the mean clear-sky radiance and standard deviation of each target area used to specify the location and time-specific optimum thresholds (Rossow and Garder 1993).

Corrections for satellite-viewing geometry were first applied to allow for the angular dependence of clear-sky radiance for each individual pixel. A decision was then made whether to update the value of clear-sky radiance from that used at the current time on the previous day. The following logical expressions were used to replace the previously assumed values of Rclear and the standard deviation, σclear of clear-sky radiance (typical value of 255 W m−2) for the target area. If R < (window + Rclear and σ < (4 σclear), then
RclearRclearR
and, if σ < σclear, then
σclearσ
where
i1520-0493-130-3-610-e6
In the above expressions, R is the currently measured radiance, σ is the newly calculated standard deviation of radiance, weight is the transient value of the weight assigned to the old running mean of clear-sky radiance, Wmax (=0.7) is the maximum weight, ndays is the number of days through which running mean has not been updated, and td (=10) is an assumed time constant (in days).

This decision/replacement procedure was carried out for each 4 × 4-pixel target area. The daily average clear-sky composite image derived from the 15-min composite images using the above-described algorithm is illustrated in Fig. 2a for 16 July 1999 (Note: In practice, the algorithm was run for the period 14–16 July 1999). This figure shows that any contamination which may result from the presence of unusually persistent clouds is minimized and/or removed. The frequency distribution of observed radiance in the clear-sky composite image for the whole image area is also shown in Fig. 2a for 16 July 1999 at 1600 UTC and at 2200 UTC. In these diagrams, the narrower peak at lower values corresponds to water surfaces, while the broader peak at higher radiances corresponds to land surfaces. The two peaks for water and land surfaces are reasonably well defined relative to the full range of radiance, and the observed differences in the frequency distribution between morning and afternoon are significant.

b. Deriving fractional cloud cover

For each GOES target area, the analysis procedure provides the number of cloudy and clear pixels and the value of their radiance. The observed value of radiance for the pixel is compared against Rclear, the clear-sky composite value for the target area in which the pixel falls; if the observed value is less than (Rclear + 15σclear), the pixel is classified as being a clear-sky pixel. (Note: The factor 15 in this expression may vary with application and is sometimes called the contrast threshold. Typically, the contrast threshold corresponds to a reflectance of about 10%.) If the observed raw data count is greater than 358 times the cosine of the solar zenith angle, the pixel is classified as being totally cloud covered. (In raw data terms, the value 358 corresponds to the radiance prescribed to bright clouds in this study.) These prescribed values were originated from the NOAA/NESDIS cloud detection method. If the pixel does not fall into either the clear-sky or totally cloud-covered categories, it is classified as having mixed cloud cover and is reallocated between the clear-sky and totally cloud-covered categories. If the observed radiance for the mixed pixel is Rmixed, a fractional reallocation of (RmixedRclear)/(358 − Rclear) is made to the cloud-covered category, the remainder to the clear-sky category. The final step is to assign a fractional cloud cover to each target area on the basis of how many of the 16 contributing pixels have finally been designated as having complete cloud cover.

The cloudy and clear radiances, and cloud cover derived from this algorithm, along with original GOES visible images, are illustrated in Fig. 2b for 14 July, 1999 at 2200 UTC. This figure shows that the algorithm does indeed successfully separate cloudy from clear regions in the GOES images and that it shows an ability to successfully capture the small-scale cloud features discernible in the satellite image.

5. Updating cloud cover in RAMS

Data assimilation is an important tool for providing initial conditions and improving diagnosed fields in numerical prediction models. Observations are routinely incorporated into the meteorological models at weather forecasting centers using a variety of data-assimilation methods including, for instance, the variational technique, which recognizes the errors in both the observed and prior-model values of variables. However, in this study, we opted to directly replace the modeled cloud-cover field with that derived from satellite observations. This reflects the fact that errors in the horizontal position of clouds in the model-calculated cloud fields are likely to be much greater than those in the satellite-observed fields. In this exploratory study, direct replacement also allows a clearer appreciation of the role and value of the remotely sensed data, and it enhances simplicity of description.

Initially, the high-resolution, derived fields of cloudy- and clear-sky radiance and cloud fraction were used in the UMD GEWEX/SRB model to produce estimates of the surface solar radiation fluxes at 4 km × 4 km pixel scale. These estimated fluxes agreed quite well with observations and, as Pinker et al. (1994) pointed out, are superior to many model-calculated estimates. The resulting surface solar radiation flux fields were used to successfully update RAMS values and, not surprisingly, they greatly improved the accuracy of the model-calculated net shortwave radiation at the surface. However, because the simulated cloud cover remained unchanged, the incoming longwave radiation fluxes retained their (generally lower) clear-sky values and precipitation was incorrectly located. However, the radiances and cloud fraction provided by the cloud-screening algorithm are intimately related via the cloud optical depth to the cloud water/ice simulated in the RAMS model. This relationship can be used to derive an estimate of the vertically integrated cloud water/ice in any grid square that can be directly ingested into the RAMS model to update the modeled cloud cover and, in this way, the shortwave, longwave, and precipitation fields are simultaneously and consistently improved. Modifying the cloud cover in RAMS was therefore adopted as a superior option.

a. Updating modeled cloud cover from satellite observations

Details of the methods used to update modeled cloud cover are given in the appendix; here, we provide only an overview of the strategy used. Because satellite images only provide information on the horizontal position of cloud water/ice in the modeled domain, it is necessary to define a procedure for distributing the cloud water/ice vertically in each modeled grid square. In this study, four different methods were used, depending on the difference between the observed and modeled cloud field, as follows.

  1. In the case of a 4 km × 4 km grid square in which both the satellite image and RAMS indicate no cloud is present, no alteration was made to RAMS variables for this grid square.
  2. In the case of a 4 km × 4 km grid square in which the satellite image indicates cloud and RAMS also simulates cloud, it was assumed that RAMS calculates the relative vertical position of clouds and the relative proportions of different cloud species correctly, even if it miscalculates the total amount of cloud water/ice. Thus, the procedure was to increase/decrease the modeled mass of each component species of cloud water/ice simulated in RAMS at levels by calculated fractions until the whole-column mass of cloud water/ice simulated in the model was equal to that estimated from a satellite.
  3. In the case of a 4 km × 4 km grid square in which the satellite image indicates there is no cloud but RAMS has simulated cloud, the procedure was to set the cloud water/ice in RAMS to zero for all cloud-water species at all levels in the grid square.
  4. In the case of a 4 km × 4 km grid square in which the satellite image indicates cloud but RAMS does not simulate cloud, a more complex, speculative procedure was used. The modeled temperature profile can be used to calculate a synthetic water mass profile corresponding to having a height-invariant relative humidity at all levels. An incremental procedure was used to specify a value of height-invariant relative humidity for all levels above level 3 in the model. Specification was such that the whole-column water mass content produced by the incremental procedure was equal to the whole-column mass of cloud water/ice estimated from a satellite. The vertical profile of cloud water/ice was then assumed to be equal to the difference between these two profiles. Although this procedure can assign some cloud water/ice at all levels (above 120 m), the ingested cloud water is preferentially assigned to levels where the modeled air column was closest to saturation. During the next model time step, and in the absence of better information, this calculated mass of cloud water/ice was all assigned to the liquid water species of cloud in RAMS; the other cloud-water species remained zero. However, in practice, RAMS quickly reassigns the water mass between different modeled cloud-water species and model levels.
Further, with frequent (1-min, see next section) cloud-cover ingestion, RAMS responds to the assimilated image and locates clouds where they are observed so that the more plausible of the above procedures, that is, (a) or (b) rapidly becomes the procedure most commonly used.

b. Updating cloud cover at different frequencies

Tests were made of the impact of ingesting GOES-derived cloud fields at different frequencies. Figure 3 shows example comparisons between the modeled fields given by RAMS with and without cloud ingestion once every 30, 15, and 1 min on 14 July 1999. It is obvious that, without cloud ingestion, RAMS generally tends to underestimate cloud cover significantly and (although not evident in this figure) when it does predict clouds, it is often wrongly distributed horizontally. Further, it seems RAMS is predisposed to introduce “lumping” of cloud cover between assimilation cycles. This is most noticeable when GOES images are ingested every 30 min. Nonetheless, the quality of the modeled cloud fields is significantly improved by 30- and 15-min ingestion. When GOES-derived cloud fields are assimilated at a higher frequency, that is, at 1-min intervals, with the assimilated cloud fields being derived by linear interpolation between 15-min GOES images, the tendency of RAMS to evolve lumpy clouds is controlled, and there is substantial improvement in modeled horizontal distribution and the overall cloud amount. On the basis of these results, it was concluded that 1-min assimilation of images linearly interpolated between 15-min GOES samples was advisable, at least for the version of the RAMS model used in this study. Perhaps less-frequent assimilation might be sufficient with a different model or a different convective scheme.

6. Results

a. Impact of cloud ingestion on the surface energy and water balance analysis

1) Incoming shortwave radiation

Figure 4 shows the hourly average surface solar radiation modeled by RAMS on 14–16 July 1999 with and without (1-min) cloud ingestion, compared with values observed at the AZMET sites. Figure 5 shows equivalent diagrams on 22–23 June 2000. Certain features are clear and consistent in these figures. Whenever the sky is clear, RAMS calculates reasonably good estimates of surface radiation. However, when warm-season daytime convective clouds are present, the RAMS model operating without cloud ingestion underestimates cloud cover in this environment, and thus consistently overestimates surface solar radiation relative to observations.

This weakness is mitigated by cloud ingestion. The modeled atmosphere becomes optically thicker because of the ingested clouds, and the surface radiation fields calculated by RAMS are improved relative to observations. The improvement is most obvious at sites 6, 7, 9, 10, 11, 12, and 26 on 14 July 1999, and at sites 1, 5, 6, 10, 12, 15, 21, and 26 on 22 June 2000, when dense clouds are simulated. However, the impact of relatively thin clouds on the modeled surface radiation is sometimes missed, even when such clouds are being ingested into RAMS. This is apparent at sites 23 and 26 during the period 14–16 July 1999, and at sites 4, 7, 13, 21, and 23 during the period 22–23 June 2000, due to the radiation physics in RAMS assuming that the atmosphere is largely transparent with relatively thin high clouds. In general, with no cloud ingestion, RAMS systematically estimates very low values of surface radiation early in the morning over most of the sites because the model tends to create an inversion (and fog) layer near the surface at these times. Cloud-cover ingestion removes this layer.

Figure 6 shows the modeled hourly average daytime surface solar radiation for model simulations without and with cloud ingestion. Figures 6a and 6b correspond to the period 14–16 July 1999, Fig. 6a is for hours which are cloudy and Fig. 6b is for hours which are clear on the basis of satellite observations. Figures 6c and 6d are equivalent to Figs. 6a,b, respectively, but correspond to the period 22–23 June 2000. In each case, the correlation coefficient, root-mean-square error, and mean bias are given in the figure. In cloudy-sky conditions, the scatter is substantially less with ingestion of cloud cover, and the correlation coefficient, root-mean-square error, and mean bias are improved accordingly. The improvement during the period 22–23 June 2000, is less than during the period 14–16 July 1999, because, as mentioned above, thin high clouds were more common in this later period. When there is no ingestion of clouds, RAMS calculates incoming surface solar radiation quite well in clear-sky conditions, but with some uncertainty and evidence of significant positive bias. This bias is partly offset by periods in which there is an (likely unrealistic) early morning, near-surface inversion layer modeled in the runs without cloud ingestion. In both study periods, the runs with cloud ingestion give a slightly reduced rmse due, in part, to the removal of the periods with a near-surface inversion layer, but the (fortuitous) reduction of bias offset is consequently also removed.

2) Incoming longwave radiation

Figure 7 shows modeled hourly average incoming radiation fluxes observed at the experimental field site near Tombstone, Arizona, during the period 22–23 June 2000, together with modeled values with and without cloud ingestion. Figures 7a and 7b show modeled incoming shortwave and longwave radiation, respectively, when the Harrington radiation scheme was used to calculate those types of radiation. The RAMS model has options for shortwave and longwave radiation schemes, which can be used independently. Figure 7c shows the results obtained when the Chen–Cotton longwave radiation scheme is used. The fractional cloud cover derived from satellite observations is also shown in Figs. 7b and 7c.

With the Chen–Cotton longwave radiation scheme (Fig. 7c), RAMS-simulated incoming longwave radiation responds appropriately to the variation in observed cloud cover during daylight hours when satellite observations are ingested. However, the same is not true when the Harrington longwave scheme is used (Fig. 7b). In this case, the longwave flux shows the expected response only when relatively thick clouds are ingested, for example, at local times 15 and 42 h. On the average, over the period for which comparison was made, ingesting clouds improves the rmse from 71.54 W m−2 to 38.62 W m−2 and the bias from −69.80 W m−2 to −26.84 W m−2 when the Chen–Cotton longwave radiation scheme is used. When the Harrington longwave scheme is used, the rmse improves from 73.42 W m−2 to 59.57 W m−2 and the bias from −71.76 W m−2 to −50.72 W m−2.

At local time 37 h, the fractional cloud cover was around 0.2 but, at this time, the longwave flux calculated with the Chen–Cotton scheme was 400 W m−2. This implies that, in spite of the very low amount of observed cloud cover, the model was able to capture small-scale, relatively thicker cumulus cloud at this time, which is consistent with the modeled incoming shortwave radiation (Fig. 7a). Without cloud ingestion, using the Harrington longwave scheme gives lower nighttime values of incoming longwave radiation (∼250 W m−2), less consistent with observations than when using the Chen–Cotton longwave scheme (∼350 W m−2). With both longwave radiation schemes, ingesting clouds gives some minor improvement in nighttime values of incoming longwave radiation relative to observations, because there is no cloud ingestion during the night. It seems that RAMS shows a tendency to rapidly lose ingested cloud cover data.

3) Precipitation

Figure 8 shows a comparison between the precipitation observed at the AZMET sites and that calculated by RAMS with and without cloud ingestion for 14–16 July 1999, while Fig. 9 shows equivalent diagrams for 22–23 June 2000. It is apparent that the RAMS model consistently underestimates surface precipitation at the AZMET sites when no satellite observations of cloud are ingested. However, this feature is improved to some extent when clouds are ingested. The simulated storms are approximately in the right place and occur at about the right time because the modeled cloud cover is more realistic during the day. This is not, of course, the case for nighttime storms. RAMS was not, for instance, able to capture the observed precipitation (∼20 mm) that occurred at night 12 h after the last cloud ingestion at sites 7 and 11 (Fig. 8).

However, with cloud ingestion, RAMS does tend to overestimate precipitation relative to that observed at the AZMET sites, perhaps because modeled precipitation is highly sensitive to the concentration of microphysical variables that is being updated frequently (every minute) by satellite observations. There is a greater tendency to overestimate precipitation during the second study period, 22–23 June 2000. At sites 5, 6, 10, 12, 15, and 21, for instance, RAMS calculates substantial precipitation while observations show little precipitation, although the model had reason to expect precipitation because observed incoming solar radiation (Fig. 5) suggests the presence of thick overlying convective clouds at these times.

The precipitation observations are point measurements and, to minimize the consequences of this on the comparison, site-average modeled and observed precipitation are compared in Fig. 10 for 14–16 July 1999 and 22–23 June 2000. Averaging enhanced the agreement and precipitation is at least modeled to occur during these periods—which is not the case without cloud ingestion—and its timing is broadly correct. However, a substantial discrepancy in magnitude between modeled and observed precipitation remains, suggesting there is scope for improving the representation of the relationship between clouds and precipitation used in the model.

b. The short-term forecast

As an initial test of the impact of cloud ingestion on the predictive ability of RAMS, a 6.5-h predictive model run was made following 3 h of 1-min cloud-cover ingestion. Every 15 min, the forecast fractional cloud cover across a 4 km × 4 km modeled domain was compared with the observed cloud fraction derived from the GOES satellite image (derived as described earlier) as shown in Fig. 11 for a 4.5-h forecast period. The modeled fractional cloud-cover fields were calculated using the time- and location-specific relationship between satellite-derived total cloud mass and fractional cloud cover for the period of the study.

As a test of the evolution in model-predictive skill following initiation of the free-running predictive run, a skill score was defined, based on the match between modeled and observed cloud cover over the model domain, the level of success being specified by the equation
i1520-0493-130-3-610-e7
where fmi and fsi are modeled and satellite-derived fractional cloud cover, respectively. This skill assessment was applied for the whole domain and also applied separately for the grid squares that were observed to be both clear and cloudy (because images with many clear pixels that are also modeled as clear may favorably bias the estimated skill). Figure 12 shows the success rate following cloud ingestion calculated at 15-min intervals during the 6.5-h forecast period for the whole domain and for the cloudy-only and clear-only subdomains. In this comparison, we did not include the forecast run without cloud ingestion because RAMS mostly simulates “no clouds” over the modeled domain. In general, the success rate is 63% after 2 h, and the forecast ability of RAMS is better than 50% up to 4 h over the whole domain and the cloud-covered subdomain. The success rate rises for the subdomain of clear pixels as the run proceeds and observed and modeled cloud cover falls.

7. Summary and conclusions

This paper describes research in which techniques were developed to acquire fine-scale information on cloud cover from the GOES visible imager and to assimilate these data into the RAMS model to yield improved diagnoses of the fields of incoming short- and longwave radiation and precipitation at the earth's surface. By improving the initial analysis of the cloud field in model simulations, application of these methods enhances the model's potential to make more accurate short-term forecasts.

The primary conclusions of the present research are as follows:

  1. It is possible to create an automatic cloud-screening procedure that uses real-time tracking (as opposed to retrospective analysis) as the basis of a cloud-cover retrieval algorithm that is robust, functions correctly, and provides appropriate, fine-resolution detection of clouds.
  2. It is possible to derive three-dimensional fields of cloud cover for assimilation into a mesoscale meteorological model (RAMS) from retrieved horizontal cloud patterns by assuming that the model computes realistic vertical profiles during the previous time step; these fields can be ingested without disrupting model stability.
  3. On the average, there is significant improvement in the ability of RAMS to model both the overall amount and horizontal distribution of cloud fields when satellite observations are ingested, especially when cloud fields—linearly interpolated between the 15-min GOES images—are ingested every minute.
  4. Cloud ingestion substantially increases the quality of the comparison between RAMS-calculated estimates of downward surface solar radiation and AZMET observations by extending the adequate simulation in clear-sky conditions to give improved simulation during cloudy-sky conditions.
  5. A single-point comparison between model-derived and observed downward longwave radiation suggests that the ability of RAMS to simulate this field is also improved by ingesting cloud cover, most significantly during the day (when cloud-cover information is available) but also to some extent during the night.
  6. Comparison between model-calculated precipitation and AZMET observations demonstrated that ingesting cloud-cover assimilation gives an improvement in the timing and, to some extent, magnitude of modeled precipitation, but the agreement was not perfect.
  7. Prior cloud ingestion enhances the ability of RAMS to forecast cloud cover (with a skill score greater than 50% for up to 4 h).

Acknowledgments

Primary support for the research described in this paper came from NASA Grant NAG5-3640-5. Additional support came from NASA Grant NAG8-1531. We are grateful to James Toth for his valuable help with the RAMS model during the early stages of this research, to Jaime Garatuza-Payan for help in the development of the cloud-screening algorithm, to Russ Scott for providing data from the infrared radiometer, to the AZMET program for allowing use of their data, and to Dan Braithwaite for assistance in obtaining the required satellite datasets. Thanks are also due to Corrie Thies for valuable editorial suggestions.

REFERENCES

  • Arakawa, A., , and V. R. Lamb, 1977: Computational design of the basic dynamical processes of the UCLA general circulation model. Methods Comput. Phys, 17 , 174264.

    • Search Google Scholar
    • Export Citation
  • Avissar, R., , and Y. Mahrer, 1988: Mapping frost-sensitive areas with a three-dimensional local-scale numerical model. Part I: Physical and numerical aspects. J. Appl. Meteor, 27 , 400413.

    • Search Google Scholar
    • Export Citation
  • Briegleb, B. P., , P. Minnis, , V. Ramanathan, , and E. Harrison, 1986: Comparison of regional clear-sky albedos inferred from satellite observations and model calculations. J. Climate Appl. Meteor, 25 , 214226.

    • Search Google Scholar
    • Export Citation
  • Chen, C., , and W. R. Cotton, 1983: A one-dimensional simulation of the stratocumuluscapped mixed layer. Bound.-Layer Meteor, 25 , 289321.

    • Search Google Scholar
    • Export Citation
  • Clark, T. L., 1977: A small-scale dynamic model using a terrain-following coordinate transformation. J. Comput. Phys, 24 , 186215.

  • Clark, T. L., , and R. D. Farley, 1984: Severe downslope windstorm calculations in two and three spatial dimensions using anelastic interactive grid nesting: A possible mechanism for gustiness. J. Atmos. Sci, 41 , 329350.

    • Search Google Scholar
    • Export Citation
  • Cotton, W. R., , and G. J. Tripoli, 1978: Cumulus convection in shear flow—three-dimensional numerical experiments. J. Atmos. Sci, 35 , 5031521.

    • Search Google Scholar
    • Export Citation
  • Cotton, W. R., , G. J. Tripoli, , R. M. Rauber, , and E. A. Mulvihill, 1986: Numerical simulation of the effects of varying ice crystal nucleation rates and aggregation processes on orographic snowfall. J. Climate Appl. Meteor, 25 , 16581680.

    • Search Google Scholar
    • Export Citation
  • Cotton, W. R., , J. F. Weaver, , and B. A. Beitler, 1995: An unusual summertime downslope wind event in Fort Collins, Colorado, on July 1993. Wea. Forecasting, 10 , 786797.

    • Search Google Scholar
    • Export Citation
  • Deardorf, J. W., 1980: Stratocumulus-capped mixed layers derived from a three-dimensional model. Bound.-Layer Meteor, 18 , 495527.

  • Garatuza-Payan, J., , R. T. Pinker, , and W. J. Shuttleworth, 2001: High-resolution cloud observations for northwestern Mexico from GOES-7 satellite observations. J. Atmos. Oceanic Technol, 18 , 3955.

    • Search Google Scholar
    • Export Citation
  • Harrington, J. Y., 1997: The effects of radiative and microphysical processes on simulated warm and transition season arctic stratus. Department of Atmospheric Science Paper 637, Colorado State University, Fort Collins, CO, 289 pp.

    • Search Google Scholar
    • Export Citation
  • Louis, J. F., 1979: A parametric model of vertical eddy fluxes in the atmosphere. Bound.-Layer Meteor, 17 , 187202.

  • Manabe, R., , J. Stouffer, , M. J. Spelman, , and K. Bryan, 1991: Transient responses of a coupled ocean–atmosphere model to gradual changes of atmospheric CO2. Part I: Annual mean response. J. Climate, 4 , 785818.

    • Search Google Scholar
    • Export Citation
  • Manton, M. J., , and W. R. Cotton, 1977: Parameterizations of the atmospheric surface layer. J. Atmos. Sci, 34 , 331334.

  • McNider, R. T., , and R. A. Pielke, 1981: Diurnal boundary-layer development over sloping terrain. J. Atmos. Sci, 38 , 21982212.

  • Meyers, M. P., , and W. R. Cotton, 1992: Evaluation of the potential for wintertime quantitative precipitation forecasting over mountainous terrain with an explicit cloud model. Part I: Two-dimensional sensitivity experiments. J. Appl. Meteor, 31 , 2650.

    • Search Google Scholar
    • Export Citation
  • Meyers, M. P., , P. J. Demott, , and W. R. Cotton, 1991: New primary ice-nucleation parameterizations in an explicit cloud model. J. Appl. Meteor, 30 , 708721.

    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., and and Coauthors, 1992: A comprehensive meteorological modeling system—RAMS. Meteor. Atmos. Phys, 49 , 6991.

  • Pinker, R. T., , and J. A. Ewing, 1985: Modeling surface solar radiation: Model formulation and validation. J. Climate Appl. Meteor, 24 , 389401.

    • Search Google Scholar
    • Export Citation
  • Pinker, R. T., , and I. Laszlo, 1992: Modeling surface solar irradiance for satellite applications on global scale. J. Appl. Meteor, 31 , 194212.

    • Search Google Scholar
    • Export Citation
  • Pinker, R. T., , W. P. Kustas, , I. Laszlo, , M. S. Moran, , and A. R. Huete, 1994: Basin-scale solar irradiance estimates in semiarid regions using GOES 7. Water Resour. Res, 30 , 13751386.

    • Search Google Scholar
    • Export Citation
  • Randall, D. A., , J. Coakley, , C. Fairall, , R. Kropfli, , and D. Lenschow, 1984: Outlook for research on marine subtropical stratocumulus clouds. Bull. Amer. Meteor. Soc, 65 , 12901301.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., , and L. C. Garder, 1993: Cloud detection using satellite measurements of infrared and visible radiances for ISCCP. J. Climate, 6 , 23412369.

    • Search Google Scholar
    • Export Citation
  • Slingo, A., 1990: Sensitivity of the Earth's radiation budget to changes in low clouds. Nature, 343 , 4951.

  • Stephens, G. L., 1978: Radiation profiles in extended water clouds. II: Parameterization schemes. J. Atmos. Sci, 35 , 21112122.

  • Stephens, G. L., 1984: The parameterization of radiation for numerical weather prediction and climate models. Mon. Wea. Rev, 112 , 826837.

    • Search Google Scholar
    • Export Citation
  • Tremback, C. L., , and R. Kessler, 1985: A surface temperature and moisture parameterization for use in mesoscale numerical models. Preprints, Seventh Conf. on Numerical Weather Prediction, Montreal, PQ, Canada, Amer. Meteor. Soc., 355–358.

    • Search Google Scholar
    • Export Citation
  • Tremback, C. L., , G. J. Tripoli, , and W. R. Cotton, 1985: A regional scale atmospheric numerical model including explicit moist physics and a hydrostatic time-split scheme. Preprints, Seventh Conf. on Numerical Weather Prediction, Montreal, PQ, Canada, Amer. Meteor. Soc., 433–434.

    • Search Google Scholar
    • Export Citation
  • Tripoli, G. J., , and W. R. Cotton, 1980: A numerical investigation of several factors contributing to the observed variable intensity of deep convection over south Florida. J. Appl. Meteor, 19 , 10371063.

    • Search Google Scholar
    • Export Citation
  • Tripoli, G. J., , and W. R. Cotton, 1982: The Colorado State University three-dimensional cloud/mesoscale model–1982. Part I: General theoretical framework and sensitivity experiments. J. Rech. Atmos, 16 , 185219.

    • Search Google Scholar
    • Export Citation
  • Tripoli, G. J., , and W. R. Cotton, 1986: An intense, quasi-steady thunderstorm over mountainous terrain. Part IV: Three-dimensional numerical simulation. J. Atmos. Sci, 43 , 896914.

    • Search Google Scholar
    • Export Citation
  • Walko, R. L., , C. J. Tremback, , R. A. Pielke, , and W. R. Cotton, 1995: An interactive nesting algorithm for stretched grids and variable nesting ratios. J. Appl. Meteor, 34 , 994999.

    • Search Google Scholar
    • Export Citation
  • Whitlock, C. H., and and Coauthors, 1995: First global WCRP shortwave surface radiation budget dataset. Bull. Amer. Meteor. Soc, 76 , 905922.

    • Search Google Scholar
    • Export Citation

APPENDIX

Updating RAMS-Calculated Clouds from Satellite Observations

Whole-column cloud water and ice estimated from a satellite

The UMD GEWEX/SRB model provides estimates of cloud optical depth, τi,j, for the Ni × Nj (4 km × 4 km) target areas within the study area; this being determined as a function of cloudy-sky radiance, Li,j, thus
τi,jfLi,j
Each satellite target area is collocated with a 4 km × 4 km grid used in RAMS by applying a weighted interpolation between the two grid projections. A relationship is assumed between mCL,SATi,j, the mass (in kg) of cloud water and ice in the atmosphere above the model grid square (i, j), and the satellite-derived values of cloud optical depth, with the form
mCL,SATi,jexp{[log(τi,j)−0.26]/1.71}ds
where ds is the area (in m2) of a horizontal grid square. The right-hand side of Equation (A2) is taken from the equation derived by Stephens (1978, 1984).

Fractional cloud-mass profiles represented in RAMS prior to cloud ingestion

Prior to cloud ingestion, RAMS calculates values of air temperature, Ti,j,k, air pressure, pi,j,k, and mixing ratios for total water, rTi,j,k, for cloud water, rCi,j,k, for rainwater, rRi,j,k, for hail, rHi,j,k, for graupel, rGi,j,k, for aggregate, rAi,j,k, for snow, rSi,j,k, for pristine, rPi,j,k, and for water vapor, rVi,j,k, respectively, at each vertical level, k, above each model grid square (i, j). The corresponding density of dry air, ρd (in kg m−3) and vapor pressure for water vapor, e (in kPa), are calculated by the following equations:
i1520-0493-130-3-610-ea3
where Rd(=287 J kg−1 K−1) is the dry air gas constant.
The mass of water vapor, mVi,j,k, and the mass of water as cloud water, mCi,j,k, rainwater, mRi,j,k, hail, mHi,j,k, graupel, mGi,j,k, aggregate, mAi,j,k, snow, mSi,j,k, pristine, mPi,j,k, all in kg, in a parcel of the atmosphere with volume Vi,j,k can then be calculated from their respective mixing ratios and the vapor pressure, air pressure, and temperature of the air parcel, thus
i1520-0493-130-3-610-ea5
The total mass of cloud water and ice produced in the atmospheric column above the modeled grid square is then calculated from
i1520-0493-130-3-610-ea6
where k refers to each atmospheric level in the atmospheric column and Nk(=28) is the number of vertical levels in the model. If the total mass of cloud water and ice, MCL,RAMSi,j, in each atmospheric column where satellite-observed clouds are greater than zero (meaning that RAMS also produces some clouds for this particular column after ingestion), for a given grid square, the vertical cloud fraction is simply given at each vertical level by
i1520-0493-130-3-610-ea7
where fi,j,k's are vertical cloud mass profiles for cloud water, rainwater, hail, graupel, aggregate, snow, and pristine, respectively.
If MCL,RAMSi,j is equal to zero (meaning that RAMS does not produce any clouds for this particular column but the satellite observes clouds), cloud water must be created at vertical levels. In this case, it is assumed that only cloud water (rather than other cloud components) is created (cloud water is the basis for creating the other cloud-water species in the model). Such cloud water is created at the most appropriate level in the atmospheric column, that is, at levels where the air is closest to saturation until MCL,RAMSi,j is greater than mCL,SATi,j. To do this, it is assumed that there is some critical value of relative humidity, RHc, which is constant through the vertical profile. At each level in the atmospheric column, water vapor in excess of this value of relative humidity is reassigned to cloud water to make MCL,RAMSi,j greater than mCL,SATi,j. However, because the appropriate value of RHc is not known a priori at each time step, an appropriate value must be found, as follows. Starting with RHc set to 1, the value is reduced (in steps of ΔRH) until MCL,RAMSi,j is greater than mCL,SATi,j, or until RHc is zero, whichever occurs first. As soon as one of these two criteria is satisfied, it is assumed that newly created vertical profiles of cloud water given by adding in the water vapor from the air above RHc are correct in relative terms. Once calculated, these newly created values of cloud water can then be used to calculate the fractional mass or clouds at each vertical level in the atmospheric column as above, from
i1520-0493-130-3-610-ea8
As previously mentioned, the fractional masses for other cloud species are set equal to zero at this stage, but their values quickly evolve as the model runs.

Updating the RAMS cloud field and air temperature

Once the fractional mass or cloud at each vertical level in the atmospheric column has been calculated as above, a new mass of cloud water, rainwater, hail, graupel, aggregate, snow, and pristine in each parcel (i, j, k) of the atmosphere can be calculated from
i1520-0493-130-3-610-ea9
A revised mixing ratio for cloud water, rainwater, hail, graupel, aggregate, snow, and pristine is then recalculated from
i1520-0493-130-3-610-ea10
This equation is applied separately for each new water mass (mi,j,k)new of cloud water, rainwater, hail, graupel, aggregate, snow, and pristine to calculate their corresponding revised mixing ratios (ri,j,k). Finally, the mixing ratio of total water, rTi,j,k, is updated by summing revised components of cloud water and ice and vapor mixing ratios as follows:
i1520-0493-130-3-610-ea11
If the satellite does not assign any cloud cover for a particular atmospheric column above the grid square in the modeled domain, any cloud amounts previously calculated by RAMS in that column are removed by assigning all the components of cloud water and ice to zero. In this case, air parcel (i, j, k) of the atmosphere is represented as being in the vapor phase, thus
rTi,j,krVi,j,k
Finally, the modeled air temperature is updated immediately after the mixing ratio of the cloud water and ice fields have been revised in accordance with the satellite data.
Fig. 1.
Fig. 1.

The study area modeled in the RAMS model and the location of the AZMET field sites and the single experimental sites near Tombstone, AZ, where the data used in this study were collected

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0610:IOISDC>2.0.CO;2

Fig. 2.
Fig. 2.

(a) The clear-sky composite image for 16 Jul 1999, and frequency distribution of observed radiance in the clear-sky composite image for the whole image area at 1600 UTC and at 2200 UTC. (b) The GOES visible image together with the derived cloudy-sky and clear-sky radiance and cloud fraction derived by the algorithm at 2200 UTC 14 Jul 1999

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0610:IOISDC>2.0.CO;2

Fig. 3.
Fig. 3.

Cloud-cover distributions at three half-hourly intervals starting 1700 14 Jul 1999. Columns 1–3 show the modeled cloud cover in RAMS immediately before a new ingestion of cloud-cover data with previous cloud data having been ingested at 30, 15, and 1-min intervals, respectively. Column 4 shows the observed cloud-cover image that will be ingested, while column 5 shows the modeled cloud field given by the RAMS model without any cloud-cover ingestion

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0610:IOISDC>2.0.CO;2

Fig. 4.
Fig. 4.

Incoming surface solar radiation observed at the AZMET sites compared with the equivalent modeled values with and without cloud-cover ingestion during the period 14–16 Jul 1999

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0610:IOISDC>2.0.CO;2

Fig. 5.
Fig. 5.

Incoming surface solar radiation observed at the AZMET sites compared with the equivalent modeled values with and without cloud-cover ingestion during the period 22–23 Jun 2000

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0610:IOISDC>2.0.CO;2

Fig. 6.
Fig. 6.

Modeled hourly average daytime surface solar radiation for model simulations with and without cloud ingestion: (a) hours that are cloudy, and (b) hours that are clear on the basis of satellite observations for 14–16 Jul 1999; (c), (d) Equivalent to (a) and (b), respectively, but for the period 22–23 Jun 2000. In each case, the correlation coefficient, root-mean-square error, and mean bias are given

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0610:IOISDC>2.0.CO;2

Fig. 7.
Fig. 7.

Modeled hourly average incoming radiation fluxes observed at the experimental field site near Tombstone, AZ, during the period 22–23 Jun 2000, together with modeled values with and without cloud-cover ingestion. (a) The incoming shortwave radiation, and (b) the longwave radiation when the Harrington radiation scheme was used to calculate both shortwave and longwave radiation. (c) Longwave radiation calculated when the Chen–Cotton longwave radiation scheme is used. The fractional cloud cover derived from satellite observations is also shown in (b) and (c)

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0610:IOISDC>2.0.CO;2

Fig. 8.
Fig. 8.

Precipitation observed at the AZMET sites compared with the equivalent modeled values with and without cloud-cover ingestion during the period 14–16 Jul 1999

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0610:IOISDC>2.0.CO;2

Fig. 9.
Fig. 9.

Precipitation observed at the AZMET sites compared with the equivalent modeled values with and without cloud-cover ingestion during the period 22–23 Jun 2000

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0610:IOISDC>2.0.CO;2

Fig. 10.
Fig. 10.

Site-average modeled and observed precipitation are compared for 14–16 Jul 1999 and 22–23 Jun 2000

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0610:IOISDC>2.0.CO;2

Fig. 11.
Fig. 11.

Forecast fractional cloud cover compared with observed cloud fraction derived from the GOES visible images at 15-min intervals during the 4.5-h forecast period

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0610:IOISDC>2.0.CO;2

Fig. 12.
Fig. 12.

Success rate of the RAMS model, following cloud ingestion, calculated at 15-min intervals during a 6.5-h forecast period for the whole domain and for the cloudy-only and clear-only subdomains

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0610:IOISDC>2.0.CO;2

Table 1.

The primary and relevant features of RAMS as applied in this study

Table 1.
Table 2.

Location and elevation of the observation sites in the southern Arizona (AZMET) area used in this study

Table 2.
1

Eta Model fields are readily available on the Advanced Weather Interactive Processing System (AWIPS) Grid 212 at a 40-km resolution.

Save