• Anderson, J. R., , Hardy E. E. , , Roach J. T. , , and Witmer R. E. , 1976: A land use and land cover classification system for use with remote sensor data. U.S. Geological Survey Professional Paper 964, 28 pp.

    • Crossref
    • Export Citation
  • Belward, A. S., , Estes J. E. , , and Kline K. D. , 1999: The IGBP-DIS global 1-km land-cover data set DISCover: A project overview. Photogram. Eng. Remote Sens, 65 , 10131020.

    • Search Google Scholar
    • Export Citation
  • Chandler, L., , and Tippets D. W. , cited. 2000: NASA satellite data used operationally to help combat fires in the west. NASA News Archive. [Available online at http://earthobservatory.nasa.gov/Newsroom/NasaNews/2000/200009124019.html.].

  • Chen, F., , and Dudhia J. , 2001: Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev., 129 , 569585.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cihlar, J., 1997: GCOS/GTOS plan for terrestrial climate related observations: Version 2. GCOS Rep. 32, World Meteorological Organization, 130 pp.

  • Colle, B. A., , Westrick K. J. , , and Mass C. F. , 1999: Evaluation of MM5 and Eta-10 precipitation forecasts over the Pacific Northwest during the cool season. Wea. Forecasting, 14 , 137154.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , Mass C. F. , , and Westrick K. W. , 2000: MM5 precipitation verification over the Pacific Northwest during the 1997–1999 cool seasons. Wea. Forecasting, 15 , 730744.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crawford, T. M., , Stensrud D. J. , , Mora F. , , Merchant J. W. , , and Wetzel P. J. , 2001: Value of incorporating satellite-derived land cover data in MM5/PLACE for simulating surface temperatures. J. Hydrometeor., 2 , 453468.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crook, N. A., 1996: Sensitivity of moist convection forced by boundary layer processes to low-level thermodynamic fields. Mon. Wea. Rev., 124 , 17671785.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dickinson, R. E., , and Henderson-Sellers A. , 1988: Modeling tropical deforestation: A study of GCM land-parameterization. Quart. J. Roy. Meteor. Soc., 114 , 439462.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1989: Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46 , 30773107.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1993: A nonhydrostatic version of the Penn State–NCAR mesoscale model: Validation tests and simulation of an Atlantic cyclone and cold front. Mon. Wea. Rev., 121 , 14931513.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Friedl, M. A., and Coauthors, 2002: Global land cover from MODIS: Algorithms and early results. Remote Sens. Environ., 83 , 287302.

  • Grell, G. A., , Dudhia J. , , and Stauffer D. R. , 1995: A description of the fifth generation Penn State/NCAR mesoscale model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 138 pp.

  • Hansen, M. C., , DeFries R. S. , , Townshend J. R. G. , , Sohlberg R. , , Dimiceli C. , , and Carroll M. , 2002: Towards an operational MODIS continuous field of percent tree cover algorithm: Examples using AVHRR and MODIS data. Remote Sens. Environ., 83 , 304320.

    • Search Google Scholar
    • Export Citation
  • Hong, S-Y., , and Pan H-L. , 1996: Non-local boundary layer vertical diffusion in a medium-range forecast model. Mon. Wea. Rev., 124 , 23222339.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Justice, C. O., , Townshend J. R. G. , , Vermote E. F. , , Masuoka E. , , Wolfe R. E. , , Saleous N. , , Roy D. P. , , and Morisette J. T. , 2002: An overview of MODIS land data processing and product status. Remote Sens. Environ., 83 , 315.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kain, J. S., , and Fritsch J. M. , 1992: Convective parameterization for mesoscale models: The Kain–Fritsch scheme. The Representation of Cumulus Convection in Numerical Models, Meteor. Monogr., No. 46, Amer. Meteor. Soc., 165–170.

    • Crossref
    • Export Citation
  • Kurkowski, N. P., , and Stensrud D. J. , 2003: Assesment of implementing satellite-derived land cover data in the Eta Model. Wea. Forecasting, 18 , 404416.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Loveland, T. R., , and Belward A. S. , 1997: The IGBP-DIS global 1-km land cover data set, DISCover: First results. Int. J. Remote Sens., 18 , 32913295.

    • Search Google Scholar
    • Export Citation
  • Miller, D. A., , and White R. A. , 1998: A conterminous United States multilayer soil characteristics data set for regional climate and hydrology modeling. Earth Interactions, 2 .[Available online at http://EarthInteractions.org.].

    • Search Google Scholar
    • Export Citation
  • Nobre, C. A., , Sellers P. J. , , and Shukla J. , 1991: Amazonian deforestation and regional climate change. J. Climate, 4 , 957988.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pan, Z., , Takle E. , , Segal M. , , and Arritt R. , 1999: Simulation of potential impacts of man-made land use changes on U.S. summer climate under various synoptic regimes. J. Geophys. Res., 104 , 65156528.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., 2001: Influence of the spatial distribution of vegetation and soils on the prediction of cumulus convective rainfall. Rev. Geophys., 39 , 151177.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., , Marland G. , , Betts R. A. , , Chase T. N. , , Eastman J. L. , , Niles J. O. , , Niyogi D. S. , , and Running S. W. , 2002: The influence of land-use change and landscape dynamics on the climate system: Relevance to climate-change policy beyond the radiative effect of greenhouse gases. Philos. Trans. Roy. Soc. London, A360 , 17051719.

    • Search Google Scholar
    • Export Citation
  • Pineda, N., , Jorba O. , , Jorge J. , , and Baldasano J. M. , 2004: Using NOAA AVHRR and SPOT VGT data to estimate surface parameters: Application to a mesoscale meteorological model. Int. J. Remote Sens., 25 , 129143.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Scepan, J., 1999: Thematic validataion of high-resolution global land-cover data sets. Photogram. Eng. Remote Sens., 65 , 10511060.

  • Schaaf, C. B., and Coauthors, 2002: First operational BRDF, albedo and nadir reflectance products from MODIS. Remote Sens. Environ., 83 , 135148.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Segal, M., , Pan Z. , , Turner R. W. , , and Takle E. S. , 1998: On the potential impact of irrigated areas in North America on summer rainfall caused by large-scale systems. J. Appl. Meteor., 37 , 325331.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sen, O. L., , Wang Y. , , and Wang B. , 2004: Impact of Indochina deforestation on the East Asian summer monsoon. J. Climate, 17 , 13661380.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Strahler, A., , Muchoney D. , , Borak J. , , Friedl M. , , Gopal S. , , Lambin E. , , and Moody A. , cited. 1999: MODIS land cover product algorithm theoretical basis document (ATBD). Version 5. [Available online at http://modis.gsfc.nasa.gov/data/atbd/atbd_mod12.pdf.].

  • Tao, W-K., , and Simpson J. , 1993: The Goddard cumulus ensemble model. Part I: Model description. Terr. Atmos. Oceanic Sci., 4 , 3572.

  • Thompson, E. H., , Gutro R. , , and Bettwy M. , cited. 2004: Land cover changes affect U.S. summer climate. NASA News Archive. [Available online at http://earthobservatory.nasa.gov/Newsroom/NasaNews/2004/2004032416710.html.].

  • Townshend, J. R. G., , and Justice C. O. , 1998: Selecting the spatial resolution of satellite sensors required for global monitoring of land transformations. Int. J. Remote Sens., 9 , 187236.

    • Search Google Scholar
    • Export Citation
  • Vermote, E. F., , El Saleous N. , , and Justice C. O. , 2002: Atmospheric correction of MODIS data in the visible to middle infrared: First results. Remote Sens. Environ., 83 , 97111.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wang, Z., , Zeng X. , , and Barlage M. , 2004: Using MODIS BRDF and albedo data to evaluate global model land surface albedo. J. Hydrometeor., 5 , 314.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wei, X., , Hahmann A. N. , , Dickinson R. E. , , Yang Z-L. , , Zeng X. , , Schaudt K. J. , , Schaaf C. B. , , and Strugnell N. , 2001: Comparison of albedos computed by land surface models and evolution against remotely sensed data. J. Geophys. Res., 106 , 2068720702.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wetzel, P. J., , and Chang J-T. , 1988: Evapotranspiration from non-uniform surfaces: A first approach for short-term numerical weather prediction. Mon. Wea. Rev., 116 , 600621.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Xue, Y., 1996: The impact of desertification in the Mongolian and the inner Mongolian grassland on the regional climate. J. Climate, 9 , 21732189.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yucel, I., , Shuttleworth W. J. , , Gao X. , , and Sorooshian S. , 2003: Short-term performance of MM5 with cloud-cover assimilation from satellite observations. Mon. Wea. Rev., 131 , 17971810.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhou, L., and Coauthors, 2003: Comparison of seasonal and spatial variations of albedos from Moderate-Resolution Imaging Spectroradiometer (MODIS) and common land model. J. Geophys. Res., 108 .4488, doi:10.1029/2002JD003326.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Nested configurations of MM5 for (a) the mid-Atlantic region and (b) the southwestern United States. Locations of the observational arrays are overlain in these maps.

  • View in gallery

    Spatial distribution of land-cover types across the 4-km model domain covering the parts of the mid-Atlantic region for (a) the USGS and (b) the MODIS-IGBP category. Black “x”s are observation sites. The dashed line shows the location of cross section used in Fig. 9.

  • View in gallery

    Spatial distribution of land-cover types across the 4-km model domain covering the parts of the southwestern United States for (a) the USGS and (b) the MODIS-IGBP category. Black “x”s are observation sites. The dashed line shows the location of cross section used in Figs. 7 and 8.

  • View in gallery

    Horizontal distributions of surface albedo at 4-km resolution for (a)–(c) the mid-Atlantic region and (d)–(f) the southwestern United States. In (a), (b), (d), and (e), the prescribed albedos are shown though in (d) and (e), the spatial correction according to the MODIS land-cover type is included. In (c) and (f), MODIS albedo is displayed for the same domains. The dashed lines in the diagrams for the mid-Atlantic region show the location of cross sections used in Fig. 9 while the ones in the diagrams for the southwestern United States show the location of cross section used in Figs. 7 and 8.

  • View in gallery

    Spatial distribution of the difference in surface skin temperature (a) between MM5-MOD and MM5 and (b) between MM5-MOD-ALB and MM5 for the southwestern United States. (c), (d) Same as (a), (b) but for the mid-Atlantic region. These products were processed from images taken on 6 Aug 2003 at 1300 local time in the southwestern United States and on 14 Nov 2002 at 1300 local time in the mid-Atlantic region. Contours of topography are also shown as background.

  • View in gallery

    From the southwestern United States on 6 Aug 2003 at 1300 local time, (a) the surface wind directions by the MM5 simulation and the surface wind differences (b) between MM5-MOD and MM5 simulations and (c) between MM5-MOD-ALB and MM5 simulations. (d)–(f) The equivalent surface wind diagrams for the mid-Atlantic region on 14 Nov 2002 at 1300 local time. In (b), (c), (e), and (f), the arrows are the wind direction differences, and the colored areas represent the difference in 10-m wind velocity. Surface topography is also shown in (a) and in (d).

  • View in gallery

    Cross section of the vertical wind field (in m s−1) and the mixing ratio (in g kg−1) taken from the semiarid region on 6 Aug 2003 at 1300 local time for the (a) MM5, (b) MM5-MOD, and (c) MM5-MOD-ALB simulations. Mixing ratio is shown as background while vertical wind is shown with contours.

  • View in gallery

    Along a cross section in the semiarid domain, (a) topography profile and the modeled PBL heights from three cases and (b) surface sensible heat and (c) latent heat fluxes from each simulation. Note that these are from the southwestern domain on 6 Aug 2003 at 1300 local time.

  • View in gallery

    (a) The contour of difference in total cloud water mixing ratio in (g kg−1) (including rainwater, hail, graupel, and snow) between MM5-MOD and MM5 and (b) surface sensible heat fluxes and (c) surface latent heat fluxes from each of the three simulations. These plots are shown along a cross section in the mid-Atlantic region on 13 Nov 2002 at 0100 local time.

  • View in gallery

    (a) The domain-averaged time series of the 2-m air temperature and (b) 2-m specific humidity for the ensemble mean, MM5-MOD, and MM5-MOD-ALB along the first 35 h of the simulation period. Standard deviation bars calculated from four members are also shown through mean ensemble simulation. Domain-averaged calculations are made in the mid-Atlantic region.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 6 6 2
PDF Downloads 5 5 2

Effects of Implementing MODIS Land Cover and Albedo in MM5 at Two Contrasting U.S. Regions

View More View Less
  • 1 Physics Department, Center for Atmospheric Sciences, Hampton University, Hampton, Virginia
© Get Permissions
Full access

Abstract

This study implements a new land-cover classification and surface albedo from the Moderate Resolution Imaging Spectroradiometer (MODIS) in the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) and investigates its effects on regional near-surface atmospheric state variables as well as the planetary boundary layer evolution for two dissimilar U.S. regions. Surface parameter datasets are determined by translating the 17-category MODIS classes into the U.S. Geological Survey (USGS) and Simple Biosphere (SiB) categories available for use in MM5. Changes in land-cover specification or associated parameters affected surface wind, temperature, and humidity fields, which, in turn, resulted in perceivable alterations in the evolving structure of the planetary boundary layer. Inclusion of the MODIS albedo into the simulations enhanced these impacts further. Area-averaged comparisons with ground measurements showed remarkable improvements in near-surface temperature and humidity at both study areas when MM5 is initialized with MODIS land-cover and albedo data. Influence of both MODIS surface datasets is more significant at a semiarid location in the southwest of the United States than it is in a humid location in the mid-Atlantic region. Intense summertime surface heating at the semiarid location creates favorable conditions for strong land surface forcing. For example, when the simulations include MODIS land cover and MODIS albedo, respective error reduction rates were 6% and 11% in temperature and 2% and 2.5% in humidity in the southwest of the United States. Error reduction rates in near-surface atmospheric fields are considered important in the design of mesoscale weather simulations.

Corresponding author address: Ismail Yucel, Center for Atmospheric Sciences, Hampton University, Hampton, VA 23668. Email: ismail.yucel@hamptonu.edu

Abstract

This study implements a new land-cover classification and surface albedo from the Moderate Resolution Imaging Spectroradiometer (MODIS) in the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) and investigates its effects on regional near-surface atmospheric state variables as well as the planetary boundary layer evolution for two dissimilar U.S. regions. Surface parameter datasets are determined by translating the 17-category MODIS classes into the U.S. Geological Survey (USGS) and Simple Biosphere (SiB) categories available for use in MM5. Changes in land-cover specification or associated parameters affected surface wind, temperature, and humidity fields, which, in turn, resulted in perceivable alterations in the evolving structure of the planetary boundary layer. Inclusion of the MODIS albedo into the simulations enhanced these impacts further. Area-averaged comparisons with ground measurements showed remarkable improvements in near-surface temperature and humidity at both study areas when MM5 is initialized with MODIS land-cover and albedo data. Influence of both MODIS surface datasets is more significant at a semiarid location in the southwest of the United States than it is in a humid location in the mid-Atlantic region. Intense summertime surface heating at the semiarid location creates favorable conditions for strong land surface forcing. For example, when the simulations include MODIS land cover and MODIS albedo, respective error reduction rates were 6% and 11% in temperature and 2% and 2.5% in humidity in the southwest of the United States. Error reduction rates in near-surface atmospheric fields are considered important in the design of mesoscale weather simulations.

Corresponding author address: Ismail Yucel, Center for Atmospheric Sciences, Hampton University, Hampton, VA 23668. Email: ismail.yucel@hamptonu.edu

1. Introduction

Regional hydrometeorological models run in a coupled land surface–hydrology and atmosphere mode are commonly used to predict near-surface and planetary boundary layer (PBL) variables at a relatively high spatial resolution. These variables are strongly influenced by the heat, moisture, and momentum exchanges between the land surface and the atmosphere. Changes in surface heat and moisture exchanges within the PBL are in turn modulated by spatial heterogeneity and mean characteristics of land cover. The reliable estimates of all the land characteristics used in the model are needed in order to produce good forecasts of surface temperature while there is no single dominant land surface characteristic to determine surface energy fluxes (Crawford et al. 2001). Further, the findings of Crawford et al. (2001) and Wetzel and Chang (1988) showed that the relative importance of five land surface characteristics indicates that soil moisture, leaf area index, and fractional green vegetation cover are most important, followed by albedo and roughness length. Other sensitivity studies have explored the impact of modifying land surface characteristics, hence the surface fluxes, at regional to global scales (Dickinson and Henderson-Sellers 1988; Nobre et al. 1991; Xue 1996; Sen et al. 2004). The impact of deforestation, a common land-cover transformation in Amazonia and tropical Asia, has been the focus of many of these and other studies. Such transformations typically increase surface albedo and decrease surface roughness, which result in a reduction in surface net radiation and, also, in surface latent heat flux. Pielke (2001) suggests that changes in surface fluxes affect weather and climate through impacts on atmospheric dynamics, clouds, and rainfall. Effectively, the spatial structure of the surface heating due to the landscape patterning alters the deep cumulus convection, which teleconnects to higher latitudes (Pielke 2001).

The fifth-generation Pennsylvania State University–National Center for Atmospheric Research (PSU–NCAR) Mesoscale Model (MM5) used in this study relies on the U.S. Geological Survey (USGS) global 1-km land-cover map (Anderson et al. 1976) produced using data from the National Oceanic and Atmospheric Administration’s Advanced Very High Resolution Radiometer (AVHRR) (Loveland and Belward 1997) from 1992 to 1993. These 10-yr-old land-cover data used for land-use parameterization in MM5 miss current updates in patterns of land-cover change, and thus are likely to produce errors through surface feedbacks into the atmosphere. As reported in Pielke et al. (2002), North America, Europe, and Southeast Asia represent the regions of intensive human-caused land-use change, where local radiative-forcing change caused by surface albedo may actually be greater than that due to all the well-mixed anthropogenic greenhouse gases together. Pineda et al. (2004) describes a recent study that incorporates the last decade land-cover changes using AVHRR and Satellite pour l’Observation de la Terre (SPOT) data in MM5. However, this study does not provide a new global land-use map, which is available to apply for other regions of the globe.

In the present study, a new global land-cover classification map produced using data from the Moderate Resolution Imaging Spectroradiometer (MODIS) of the National Aeronautics and Space Administration (NASA)’s Terra satellite is used within MM5. The MODIS instrument provides an important advance in moderate-resolution remote sensing with new spatial and spectral information for land surface characterization (Justice et al. 2002). Crawford et al. (2001) pointed out that the enhanced spatial and spectral resolution afforded by MODIS can provide land-cover products considerably superior to those derived from AVHRR data. The research objectives in this study are to investigate not only the impact on MM5-derived near-surface weather variables but also the impact on model-calculated PBL evolution due to the MODIS-derived land-cover parameters at two dissimilar U.S. regions. One of the areas used in this study is located in the semiarid climate of the southwestern United States, while the second one is located in the humid continental climate of the mid-Atlantic region. Results are compared with observed values at two multisite observational networks for the periods of 12–14 November 2002 in the mid-Atlantic region and 3–7 August 2003 in the southwestern United States. MODIS land-use categories were mapped onto the existing USGS and Simple Biosphere (SiB) categories. Measured MODIS surface albedo was also used to replace the prescribed albedo.

2. Description of model, study areas, and data

a. Hydrometeorological model

Version 3.4 of the PSU–NCAR MM5 model (Dudhia 1993; Grell et al. 1995) was used in this study. Development and verification of the MM5 model has been carried out in many applications, including Colle et al. (1999, 2000) and Yucel et al. (2003), which are recent studies focused on the prediction of precipitation and cloud microphysics at high spatial resolution. MM5 was employed in two-way interacting nested configurations with 60 × 60 grid points at 12-km resolution and 100 × 100 grid-points at 4-km resolution in the mid-Atlantic region of the United States and with 70 × 70 grid points at 12-km resolution and 130 × 130 grid points at 4-km resolution in the southwestern United States (see Fig. 1 for simulation domains). The model is nonhydrostatic, with a terrain-following 23 sigma layers, and includes a two-way interactive nested grid. Cloud and precipitation processes were parameterized using the Tao–Simpson (1993) microphysical scheme, with the Kain–Fritsch (1992) cumulus parameterization used on the coarse domain. Cloud-interactive radiative fluxes were calculated using a single broadband radiation model developed by Dudhia (1989). The Medium-Range Forecast (MRF) (Hong and Pan 1996) is selected for the PBL.

Land surface coupling in MM5 is done through the Ohio State University land surface model (OSULSM) of Chen and Dudhia (2001). OSULSM calculates the surface moisture and heat fluxes as lower boundary conditions to interact with other model physics throughout the PBL. Soil texture is based on multilayer, 16-category soil characteristics developed by Miller and White (1998) while land-cover data are derived from AVHRR data obtained from April 1992 through March 1993. There are optional 16-type SiB and 25-type USGS land-cover classifications available in the model. Each land-cover type relies on parameter sets of surface albedo, roughness length, minimal stomatal resistance, and two parameters used in radiation stress and vapor pressure deficit functions (see Table 1). This OSULSM has one canopy layer and four soil layers employed at 0–10, 10–40, 40–100, and 100–200 cm from the ground surface. Soil moisture initialization is made based on surface analysis from the National Centers for Environmental Prediction (NCEP)’s Eta Model. The model initiation and time-varying lateral boundaries for the coarse domain are provided using 40-km analysis fields from NCEP’s Eta Model.

b. Study areas and data

1) Mid-Atlantic region

The first study area covers the central portion of the mid-Atlantic region in the United States with a focus in the Chesapeake watershed in the proximity of Virginia (Fig. 1a). The climate in this region is generally dominated by midlatitude transient weather systems though moderate variations exist in response to the physiographically diverse landscape (e.g., the Appalachian Mountains and Atlantic coast). The complex pattern of rivers and streams, and high relief of the Appalachian and Blue Ridge mountain systems with the Atlantic Ocean are the unique landscape features, and together play a dominant role in the region’s precipitation regime.

Observations from 19 individual weather stations in this region are taken from the National Climatic Data Center (NCDC). The distribution of stations on the MM5 model domain is shown Fig. 1a. Hourly measurements of near-surface air temperature and relative humidity (converted to specific humidity) were used.

2) Southwestern United States

The second study area comprised the region between 31.5° and 36°N and 108° and 115°W in southern Arizona. The semiarid environment at this study site is due both to its location within the subtropics and, to a lesser degree, to the orographic drying effect of mountain ranges in the western United States. The region is characterized by rugged isolated mountain ranges separated by wide flat valley floors. The limited annual rainfall of less than 400 mm occurs mainly as convective thunderstorms during the summer monsoon season and as frontal storms during the winter.

Meteorological observations are obtained from the Arizona Meteorological Network (AZMET) system. These measurement stations are often installed adjacent to regions of irrigated agriculture, and the data may, in part, be influenced by this fact. In this study, data from 22 individual stations were used, concentrated near the center of the study area (Fig. 1b). The data used in this study were hourly measurements of near-surface air temperature and relative humidity. Detailed information on the AZMET observational network is available online at http://ag.arizona.edu/azmet/.

3. Satellite-based land surface data

NASA’s Earth Observing System (EOS) is composed of a series of polar-orbiting satellites. MODIS is an imaging instrument aboard the Terra platform orbiting the earth from north to south across the equator in the morning. It views the entire earth’s surface every 1 to 2 days and acquires data in 36 spectral bands. Products from MODIS include global datasets derived from new moderate-resolution spectral bands with spatial resolutions of 250 m to 1 km. The MODIS land products enable regional and global research applications associated with biogeochemical cycling, energy balance, land-cover change and ecosystems, for which AVHRR’s reliance and spectral and geometric constraints have limitations (Justice et al. 2002; Cihlar 1997). The two MODIS land products, land cover and surface albedo, used in this research are described in the following subsections.

a. MODIS land-cover map

MODIS global land-cover classification at 1-km resolution (Friedl et al. 2002), created using data between July and December 2000, was used to describe earth’s ecosystems in the MM5 land surface parameterization scheme. This classification is thought to be composed of higher-quality data than previous sensors and the most detailed look at the land-use map to date. Validation of MODIS land products is being undertaken over a range of conditions. For example, a study of Hansen et al. (2002) shows that MODIS data will be a substantial improvement over AVHRR in mapping tree cover. The MODIS land-use map was produced based on a global digital database of land-use types’ images (e.g., Landsat TM) updated every 16 days. The MODIS land cover supplies an International Geosphere–Biosphere Program (IGBP) land-cover classification (Belward et al. 1999; Scepan 1999) with 17 different land-cover types including major natural vegetation types, agricultural land use, and several categories of land surfaces with little or no plant cover (i.e., bare ground, urban areas, and permanent snow and ice).

b. MODIS surface albedo

In addition to the MODIS land cover, the global MODIS albedo product (Schaaf et al. 2002), available every 16 days at 0.05° resolution, was used in this research. This is a level-3 climate modeling grid derived using the global 16-day 1-km albedo product, which is calculated by integration of the bidirectional reflectance distribution function (BRDF). This provides a new input for the climate modeling community and includes systematic aerosol correction at 1 km (Vermote et al. 2002). This product is perfectly suitable to use in the applications of the mesoscale grid scale (5–30 km) at which such products have long been missing as reported by Chen and Dudhia (2001). Albedo measurements were available as black-sky and white-sky values, representing a direct component at the solar zenith angle of local solar noon and an isotropic diffuse component, respectively. Broadband white-sky albedo calculated using seven different channels between 300 and 5000 nm in the spectrum was used in the study reported here.

4. Land-cover map replacement in MM5

The existing land-cover classification maps in MM5 lack the last decade updates in land-use patterns because they were created using the AVHRR data from 1992 to 1993. During this last decade, the Earth’s terrestrial ecosystems have undergone changes such as urbanization, conversion to agriculture, desertification, deforestation, and degradation due to the natural disasters. Anthropogenic land-cover changes have also caused many spatially complex areas within an ecosystem (Hansen et al. 2002). Influence of the changes in land-cover pattern on weather and climate is comparatively stronger at regional and local scales as these are the scales of ecosystems and human communities. Studies from Hansen et al. (2002) and Townshend and Justice (1998) point out that the continuous fields from MODIS data may yield a usable land-cover-change product as they found the 250-m band to be the resolution necessary to depict human-induced land-cover change. To consider subsequent responses of land-cover change in the coupled land–atmosphere system, a new land-cover classification from the MODIS instrument was implemented in MM5.

Since the MODIS land-cover map supplies a 17-type IGBP classification, which is different from existing SiB and USGS categories in MM5, there is a need to determine the parameter sets for each IGBP class that the OSULSM in MM5 accepts. Strahler et al. (1999) provided an example for translating the IGBP classes to other classification schemes for compatibility with systems used by the modeling community. They pointed out that for nearly all classes in these schemes, there is a direct mapping of one or more IGBP classes to their equivalents. The approach used in this study follows the similar relabeling (“cross-walking”) method but translates the IGBP classes to the USGS and SiB classifications, for which the parameter sets were already established in MM5. In this approach, 15 of the 17 IGBP categories matched with categories in the USGS land cover while the other two, permanent wetlands and croplands, were related to the equivalent ones in the SiB classification. There was a direct mapping of all IGBP classes except woody savanna and cropland mosaics, which were approximated to the closest land-cover classes available in the USGS scheme. As in Strahler et al. (1999), woody savanna was assumed to be savanna class. Because cropland mosaics in the IGBP class are described as lands with mosaics of croplands, forests, shrubland, and grassland in which no one component comprises more than 60% of landscape, they have been translated to the cropland/grassland mosaic available in the USGS category. MODIS land ecosystem types were also used to update the existing land/water mask between water body and land surface in MM5. Table 1 shows application of this approach together with the corresponding parameter sets for each land-cover class. Since both MODIS and raw USGS data are in the equal area sinusoidal projection, MM5’s Terrain program takes MODIS data as input and uses its own computer codes to interpolate the data onto the MM5 Lambert conformal horizontal grid. If the horizontal grid space in MM5 is larger than 1 km, the dominant vegetation type in each grid box is selected to represent the land-cover class for that pixel.

a. Land-cover changes in mid-Atlantic region

Only the land-cover changes through the fine model domain are displayed for analysis because the finer the spatial resolution (4 km), the more detailed land-cover change can be observed. Figures 2a and 2b show land-cover maps across the 4-km model domain, which covers the vicinity of Virginia and parts in West Virginia, North Carolina, and Maryland for the AVHRR and MODIS data, respectively. It is clear that significant changes occurred during the last decade, particularly with update from mixed forest to deciduous broadleaf forest and expansion of cropland mosaics. The differences in the land classification algorithm used by the AVHRR and MODIS teams may also contribute to some of these land-cover changes. However, Strahler et al. (1999) states that the neural network classifiers used in the MODIS/IGBP classification have proven superior to conventional classifiers, often recording overall accuracy improvements in the range of 10%–20%. Land-cover changes occurred across this comparison region is equal to 44% of the total area used in this domain. In the description of the mixed forest type, the trees are in the growing stage, the update from mixed forest to broadleaf forest, therefore, explains the significant amount of tree growth that took place during this last decade. This update represents 19% of all land-cover change observed across the area. This supports the finding of increased forest cover in the last 100 yr mainly due to farm abandonment in the east of the United States reported by Thompson et al. (2004). Alternatively, croplands replaced forests mostly along the Atlantic coast and Chesapeake Bay area. The least amount of change occurred in the northwest part of the domain, the northeast of the Chesapeake Bay, and urban areas. The new, more detailed land–water mask derived from MODIS’s fine spatial resolutions at 250 and 500 m also provides the correction and updates on the delineation between water and land surfaces along the Chesapeake Bay area in the model domain.

b. Land-cover changes in the southwest United States

Land-cover maps derived from the AVHRR and MODIS instruments for the southwestern domain are shown in Figs. 3a and 3b, respectively. Twenty-six percent of pixels in the study area exhibited a land-cover change. The open shrubland region shown in Fig. 3a was originally mapped as shrubland on the USGS map. Such translation from shrubland in USGS to open shrubland is made because most of this region is described by open shrubland in MODIS data. Therefore, it is assumed that both AVHRR and MODIS data see the same land-cover type at this specified region. Most of the grassland areas seen near forest areas in AVHRR data have been replaced with open shrubland during the last decade. MODIS data also show a 28% reduction in the cover of evergreen needleleaf forest across the study area. Part of this loss may be due to fires, which burn an average 2.4 million acres per year in the United States (Chandler and Tippets 2000). In areas where reduction in forest cover occurs, savanna/woody savanna communities have become established. MODIS data also show that cropland that previously formed along the border between California and Arizona has been retired to shrubland. Another significant land-cover change or new classification is associated with the barren or sparsely vegetated surfaces that mostly cover the north of the Gulf of California in northern Mexico on the MODIS map.

5. Updating prescribed albedo values in MM5 with MODIS

As a modulator of the surface energy budget, PBL processes and cloud formation, land surface albedo is an important parameter for describing the radiative properties of the earth. Surface albedo in Table 1 is prescribed using a broadband value, which does not represent the detailed spatial variation within the model and pixels. The grid cell value may be a composite of areas of differing land use. Even the sophisticated surface albedo scheme in the Common Land Model (CLM) produces substantial biases from those derived from the MODIS broadband albedo measurements (Zhou et al. 2003). To provide consistency with the MODIS land cover, the 0.05° MODIS surface albedo products are interpolated to 12- and 4-km MM5 domains and replace the default albedo values based on land-use classification.

The first and second columns in Fig. 4 show the horizontal distribution of surface albedo at 4-km resolution in the mid-Atlantic region and in the southwestern United States, respectively. Figures 4a, 4b, 4d and 4e are the prescribed albedo values, but Figs. 4d and 4e include the spatial correction according to the MODIS land-cover map while Figs. 4c and 4f display MODIS albedo for the same domains. The spatial heterogeneity in surface albedo exhibited by MODIS data is very significant at both study areas. Spatial heterogeneity for prescribed albedo values is in better agreement with the MODIS albedo after tabulated values are updated based on new land-cover type from MODIS. This is more obvious in the mid-Atlantic region than the in southwest of United States. Prescribed albedos are generally higher, by up to 6%, than MODIS observations in the mid-Atlantic region. Compared with the mid-Atlantic region, substantially higher differences between prescribed and MODIS albedo values are produced across the semiarid region of the southwestern United States (see Figs. 4c and 4f). The prescribed versus MODIS albedo difference approaches 20%, particularly over the northern Gulf of California where barren or sparsely vegetated surfaces are apparent with MODIS land cover. The high bias results over the semiarid region are consistent with the previous studies (Wei et al. 2001; Zhou et al., 2003; Wang et al. 2004), which involved validation of albedo models using MODIS albedo measurements over the globe. Spatial heterogeneity of MODIS albedo is also more evident in the semiarid region than that in the humid and more vegetated region of the mid-Atlantic because soil albedo over desert and semidesert regions varies spatially and seasonally, depending largely on the soil mineralogical composition and wetness.

6. Results

The influence of MODIS land cover and MODIS albedo in MM5 was investigated through analysis of PBL development and comparisons between observed and modeled air temperature and humidity fields. Only the results obtained from the fine model domain are presented here due to the interest of performing a detailed land-cover impact. Three successive simulations at each study area, the standard MM5 with USGS classification, the MM5 with MODIS land-cover data, and the MM5 with MODIS land-cover and albedo data, were performed. Hereafter, the first, second, and third simulations will be named MM5, MM5-MOD, and MM5-MOD-ALB, respectively.

a. Spatial differences in surface variables

1) Skin temperature

Skin temperature is one of the surface variables affected directly by the land-cover changes due to its dependence on soil physical parameters assigned for each land-cover type. The spatial distribution of the difference in surface skin temperature between MM5-MOD and MM5 and between MM5-MOD-ALB and MM5 across the fine model domain covering the southwestern United States (Figs. 5a and 5b) on 6 August 2003 at 1300 local time and the mid-Atlantic region (Figs. 5c and 5d) on 14 November 2002 at 1300 local time are shown.

MODIS-derived land-cover changes affect the modeled ground temperature from 9° to –9°C in both study regions. Such differences are more significant across both study areas with the MM5-MOD-ALB simulations. In the southwestern United States, the regions with a new reclassification of the land use to barren or sparsely vegetated surfaces have a cooling effect (up to −1.5°C) in the MM5-MOD simulation while the conversion from grassland to open shrubland around the forest areas at high terrain causes a warming effect in the MM5-MOD simulation. Extreme warming (∼9°C) was observed at some locations of high terrain across the southern part of the domain, where woodlands replaced some grassland and shrubland. Increased stomatal resistance when converting grass to woodland may also contribute to this extreme warming consistent with the slight warming effect across the Southwest, where woodlands replaced some deserts shown by Thompson et al. (2004). Such cooling and warming effects are more enhanced with the inclusion of MODIS albedo in MM5-MOD-ALB simulation. The enhanced heterogeneity obtained by MODIS surface albedo plays an important role by affecting the partitioning of energy in the surface radiation budget. The warming effect found primarily along the high elevation areas (see Figs. 5a and 5b) potentially amplifies elevated heat sources, which develops active convection areas (see Figs. 6b and 6c) in this MM5 simulation. Albedo differences seen in Fig. 4 mostly contribute to these cooling and warming effects as surface albedo is being a dominant parameter in deriving surface processes at this semiarid region where summertime surface heating is intensive.

In the mid-Atlantic region (see Figs. 5c and 5d), where the conversion from deciduous broadleaf forest to cropland mosaics dominate, MM5-MOD tends to show a cooling effect up to −2°C while MM5-MOD-ALB exhibits a warming effect between 2° and 4°C. The different surface feedback occurs because MODIS albedo values are generally lower than prescribed values in this region (see Figs. 4b and 4c), causing the land model in MM5 to absorb more solar radiation at the earth’s surface. The alteration from forest to agriculture changes the partitioning of solar insolation between sensible and latent heat. The cooling effect of the MM5-MOD is caused by higher evaporative flux, which is due to lower stomatal resistance over cropland areas. Pan et al. (1999) state that the most prominent impact of crops on the energy budget is their lower stomatal resistance. Over longer time scales, such land-cover transformation can cause a warming effect because croplands are less efficient in transpiration than forests. For example, Thompson et al. (2004) found a warming effect along the Atlantic coast where croplands replaced forest when they were investigating the effect that land-cover changes had on the U.S. summer climate over a 300-yr period. Moreover, reduced transpiration associated with the conversion of forests to agricultural regions results in less thunderstorm activity over this landscape (Pielke et al. 2002). The most significant ground temperature changes, on the order of ±9°C occur when the delineation between water and land surfaces are updated. New water bodies observed by the new land-cover classification cause a cooling effect while updates from water to land surfaces cause a warming effect. The cumulative effect of such differences in ground temperature modulates the changes in near surface air temperature and humidity through surface heat and moisture fluxes. Crook (1996) shows that PBL differences of ±1°C in air temperature and ±1 g kg−1 in mixing ratio potentially result in significant changes in thunderstorm activity in cloud model simulations.

2) Surface wind

Variations in surface heat budgets induced by MODIS land cover and albedo produce changes in the surface and lower tropospheric wind fields over both geographical domains. Figures 6a and 6d show MM5-derived wind directions over the southwest of the United States and the U.S. mid-Atlantic regions, respectively. Changes in the surface wind, due to land-cover specification, over the southwest are shown in Figs. 6b and 6c while changes for the mid-Atlantic region are shown in Figs. 6e and 6f. The arrows are the wind vector differences, and the colored areas represent the differences in 10-m wind velocity. In the original MM5 simulation (Fig. 6a), well-established southerly surface winds bringing warm and moist air from the Gulf of California are common over the southwest and northwest parts of the domain. These prevailing surface winds are weaker over the center of the domain due to the high topography of the Mojollon Rim. Near-surface winds over high terrain (1500–2000 m) in the southeast part of the domain are also observed in Fig. 6a. With the modifications introduced to MM5, the general pattern of the surface wind vectors over the entire domain are similar but the originally simulated near-surface winds are significantly modified (Figs. 6b and 6c). For example, magnitudes in near-surface wind velocity increased from 5 to 15 m s−1. Furthermore, the areal extent and magnitude of the wind field are more enhanced with the MM5-MOD-ALB simulation (see Fig. 6c), pointing out that surface albedo is an important factor in placing the vertical mixing by thermals in this semiarid region.

In the mid-Atlantic region (Fig. 6d), the prevailing surface winds are southwesterly and westerly over the entire domain and are notably modified over places where land-cover transformation from forest to cropland occurred (Fig. 6e). For example, southwesterly surface winds are stronger over the new cropland areas where the differences in magnitude increased up to 2 m s−1. This is consistent with results of Pan et al. (1999) where wind speed increases went up to 0.3– 0.5 m s−1 because a decrease in surface roughness length occurred when forest replaced with crops. For example, in this study, surface roughness length dropped from 0.8 to 0.07 m due to the replacement of forest with crops. It is important to note that in Fig. 6f, the inclusion of MODIS albedo in MM5 did not alter much of these modifications in near-surface winds (Fig. 6e). Rather than surface albedo, other soil and vegetation parameters define each land-use type, which appear to affect near-surface wind fields in this humid environment where the earth surface is densely vegetated. The changes in the magnitude of the surface winds in this humid region are roughly 7 times lower than in the semiarid region. The intense summertime land surface heating in the semiarid region indicates that surface forcing is dominant relative to the synoptic-scale forcing.

b. Impacts on model boundary layer

A cross section of the vertical wind field and the mixing ratio taken from the semiarid region is displayed in Figs. 7a–c for the MM5, MM5-MOD, and MM5-MOD-ALB simulations, respectively. The location of this cross section is shown in land cover and albedo plots in Fig. 3 and Figs. 4d, 4e and 4f, respectively. The differences in vertical wind are clearly evident, with upward motion of 2 to 3 m s−1 occurring within the cross section between the longitudes −110.2° and −110.0°. Between these longitudes at this cross section, observed change from grassland to shrubland and a decrease in albedo by about 8% with MODIS data create favorable conditions for this updraft. Such upward vertical motion inputs moist air into the boundary layer in the MM5-MOD and MM5-MOD-ALB simulations (see Figs. 7b and 7c). This updraft is developed further with MM5-MOD-ALB simulation but is not evolved in the MM5 case (Fig. 7a), where the PBL stays warmer and drier along this layer compared to the MM5-MOD and MM5-MOD-ALB cases. Furthermore, the lower-level downward and upper-level upward motions initiated in the MM5 case around the longitude of −109.7° through the PBL are enhanced in areal extent in the MM5-MOD and MM5-MOD-ALB cases.

Along the same cross section, the topography profile and the modeled boundary layers from the three cases are shown in Fig. 8a while surface sensible heat and latent heat fluxes from each simulation are displayed in Figs. 8b and 8c, respectively. PBL profiles obtained from each model run along this cross section generally follow the variations in surface heat and moisture fluxes, which are directly affected by land surface characteristics. Sensible heat is the dominant energy flux for forming PBL profiles because latent heat flux is easily curtailed by soil moisture stress over this semiarid terrain. The location of the shallow and moist PBL profile seen along the cross section between −110.0° and −109.5° longitudes in the MM5 case is extended westward by each MM5-MOD and MM5-MOD-ALB case due to the distributions of vapor mixing ratio across this layer (see Figs. 7a–c). Along the section where PBL heights are the shallowest and wettest, the latent heat becomes the dominant flux for surface feedbacks as seen in Fig. 8c. The strength of latent heat flux is about 50–100 W m−2 smaller for the MM5-MOD and MM5-MOD-ALB between the longitudes −110.6° and −110.0° along the cross section, where grassland is converted to shrubland. This lower latent heat flux is attributable to the substantial increase (from 40 to 300 in Table 1) in stomatal resistance along with this conversion.

The importance of surface forcing on cloud and precipitation formation is investigated along a cross section (see Figs. 2 and 4a–c for its location) from the mid-Atlantic domain. Figure 9a shows differences in total cloud water (including rainwater, hail, graupel, and snow) between MM5-MOD and MM5 while surface sensible heat fluxes and surface latent heat fluxes from each of the three simulations are shown in Figs. 9b and 9c, respectively. The positive contour values (0.01 and 0.002 kg kg−1) represent new cloud formation occurring in MM5-MOD due to changes in land surface feedbacks with the MODIS land-cover implementation. Beneath the layer where positive contour values are seen, sensible heat fluxes are greater in the MM5-MOD run, which cause the modeled atmosphere to become more buoyant creating favorable conditions for extra cloud formation. Along this cross section, changes from deciduous forest to cropland mosaics attribute to an increase in sensible heat flux and a decrease in latent heat flux. However, latent heat fluxes are still greater than sensible heat fluxes at this cross section and at all the domain of the mid-Atlantic region. The negative contours shown at the western edge of the cross section where some updates from deciduous forest to cropland mosaics and from water to land appear, there are also significant reductions in surface latent and sensible heat fluxes in MM5-MOD case, which causes a removal of clouds originally created in the MM5 simulation. From Figs. 9b and 9c, it is apparent that there are no substantial differences between MM5-MOD and MM5-MOD-ALB cases, indicating that such feedbacks are solely derived by MODIS land-cover replacement at this humid environment.

It is noteworthy to mention that surface wetness due to irrigated agricultural areas has a weak impact on the resolvable rainfall systems and hardly generates any new rainfall areas as reported by Segal et al. (1998) when they investigated the impact of irrigated areas in North America on summer rainfall with MM5. Their study also points out that the overall change in land use has a larger impact on rainfall compared with the change solely related to changes in irrigation. The net effect of surface wetness on moist convective processes, which is indicative of potential for convective rainfall is uncertain because a decreased Bowen ratio (a ratio of sensible heat flux to latent heat flux) will tend to promote convective precipitation thermodynamically, but in various cases it may reduce the intensity of thermally forced circulations that may be necessary to provide the trigger for release of convective instability (Pan et al. 1999).

c. Comparison with observations

The observational arrays overlain on land-cover patterns from USGS and MODIS in Figs. 2 and 3 are used to calculate error statistics from the model simulations. Generally, improvements (positive error reductions) are obtained in MM5-derived near-surface air temperature and humidity when MODIS land cover and albedo were used in MM5. Tables 2 and 3 show the calculated root-mean-square (rms) error reduction for these fields obtained with MM5-MOD and MM5-MOD-ALB cases for each day along the single continuous simulation period and for study areas in the southwestern United States and in the mid-Atlantic region, respectively. Because model performance shows variability throughout the simulation period, improvements in near-surface weather variables have day-to-day variations. For example, this day-to-day variability is more significant for 2-m temperature in the southwest United States in Table 2.

There are observational arrays where land-cover updates from one to another do not occur (see Figs. 2 and 3: this is especially more apparent in the southwestern United States). The air above these locations, however, is affected by their nearby ecosystem changes through aerodynamic mixing in near-surface weather. In the southwestern United States, many observational locations mapped as shrubland in both land-cover schemes show significant improvements in 2-m temperature and humidity due to land-cover change from forest to shrubland at their surrounding pixels (see large change in roughness length and stomatal resistance with this conversion in Table 1). Similarly, there is also improvement in near-surface weather variables due to remote ecosystem changes from cropland to shrubland. In the mid-Atlantic region, the significant improvements in 2-m air temperature and humidity are obtained at comparison pixels where there are conversions from deciduous forest to agriculture and from water to land.

Improvements in near-surface weather variables are much greater in the dry environment of southwestern United States than in humid environment of the mid-Atlantic region, emphasizing the larger impact of land surface features in semiarid regions. Use of surface albedo measurements yields additional improvements in near-surface meteorological variables in the southwestern United States while it degrades (negative error reduction) them in the mid-Atlantic region with an exception in the first-day and all-day statistics in 2-m humidity. There are a total of six and three stations in the southwestern United States and in the mid-Atlantic region, respectively, that showed degradation in near-surface weather variables throughout the simulation period. Interestingly, in both study areas, the highest rms error reduction is obtained on the last day of the simulation period, indicating that there are accumulative effects of surface fluxes on these simulations. For example, the accuracy in air temperature is increased to 21% over the semiarid sites in the MM5-MOD-ALB case by day 4.

The statistics of the rms error (rmse) and bias in air temperature and humidity are given for each simulation in Table 4. Error statistics obtained from MM5-MOD and MM5-MOD-ALB cases are consistently better than the original MM5 case. The results here are similar to those of Kurkowski and Stensrud (2003) who showed that the rmse improvements in 2-m temperature varied between 0.01° and 0.06°C clearly improved temperature biases in the Eta Model forecast when they investigated the impact of using real-time vegetation fraction derived from satellite data in the Eta model. Compared to their findings, the current study exhibits a higher rmse improvement in 2-m temperature, which reachs 0.33°C for the southwestern United States. This is because all surface parameters (e.g., albedo, roughness length, minimal stomatal resistance, radiation stress, and vapor pressure deficit functions) become effective in the surface feedbacks with an updated land-cover type.

Since the above differences are quite small and given the fact that such differences can also be obtained with internal model variability, it is important to perform some ensemble simulations along which the significance of these differences can be highlighted. Three ensemble runs initiated at 0600, 1200, and 1800 UTC on 12 November 2002 were made for a comparison period in the mid-Atlantic region. Figure 10 shows the domain-averaged time series of the 2-m air temperature in Fig. 10a and 2-m specific humidity in Fig. 10b for the ensemble mean, MM5-MOD and MM5-MOD-ALB along the first 35 h of simulation period. The ensemble mean is calculated from four members including the MM5 simulation. The standard deviation bars along with ensemble mean simulation are plotted to indicate the spread of the ensemble. These error bars show that model variability due to initial condition errors become less important toward the end of the simulation. In these figures, domain-averaged plots of temperature and specific humidity from MM5-MOD and MM5-MOD-ALB sometimes lie outside of the model variability. This indicates that model runs with surface parameter changes are higher than the model variation due to initial condition errors. Therefore, it implies that the results over some locations in the model domain are statistically significant. It is likely that results in Fig. 10 would be more statistically significant in the semiarid location because surface feedbacks through parameter changes were stronger in that region than in the mid-Atlantic.

7. Summary and conclusions

This paper describes research in which a new land-cover classification map and surface albedo data from the MODIS instrument were incorporated into MM5. Land surface cover parameter sets derived from emerging MODIS products map well to existing classifications, such as those from the USGS and SiB, within MM5. The impact of this incorporation on model near-surface meteorological variables through model PBL layer was investigated at two dissimilar U.S. regions. The following conclusions can be derived from this research:

  • The 44% and 26% of the total pixels used in the mid-Atlantic region and in the SW of the U.S. domains, respectively, showed ecosystem changes. These changes, for example, indicate an increase in the amount of forest cover in the mid-Atlantic region but show a decrease in forest cover occurred in the southwestern United States.
  • Weak spatial variability in surface albedo due to the prescribed values in MM5 is substantially improved by using continuous surface albedo data from MODIS. High biases were found in prescribed albedos that are particularly more evident across the semiarid region. Spatial adjustment of albedo performed according to the MODIS land-cover type yielded some improvement in the variability of prescribed albedo values but it did not alter their magnitudes.
  • Land-cover changes directly affect MM5-calculated surface skin temperatures with the differences ranging between −9° and 9°C in both study areas. Cooling and warming effects are more enhanced with the inclusion of MODIS albedo due to its important role in the partition of the surface radiation budget. Over the regions where there are large heat modifications, remarkable alterations in near-surface wind are found. Modifications in surface wind are more significant over the semiarid region where the summertime land surface forcing reaches its maximum. Inclusion of MODIS surface albedo is also more sensitive in the simulation of wind fields over this semiarid location.
  • Surface heat and moisture fluxes altered by the changes in land cover and surface albedo affected modeled PBL evolution through which mesoscale cloud and precipitation systems are triggered. Both MODIS surface datasets, particularly in the southwestern United States, enhanced and created more local vertical motions within the modeled PBL.
  • Soil and vegetation parameters yielded larger changes in land surface feedbacks over the mid-Atlantic region where the environment is humid and vegetated while surface albedo has significant effects in land surface forcing over the semiarid land of the southwestern United States.
  • Based on the comparison with observations at two study regions, errors in air temperature and humidity are appreciably reduced with the implementation of MODIS land cover and albedo data. Improvements are substantially higher in the semiarid region than those in the humid region. In both study regions, the inclusion of MODIS surface albedo in MM5 resulted in further improvements, which were particularly evident in the semiarid sites. Additionally, the reduction in error was greatest during the last day of the MM5-MOD and MM5-MOD-ALB simulations in both study areas. This is indicative of beneficial cumulative surface feedbacks, which reach their largest influence on the last day.
  • Rmse improvements in near-surface meteorological variables are considered important in this study as, for example, a study from Kurkowski and Stensrud (2003) showed a modification in Eta Model temperature biases due to a slight improvement in such state variables.

Recent efforts have produced a number of high-quality remote sensing products from EOS to be used in earth system modeling applications with the aim of improving model performance. This study contributes to this effort by exhibiting evidence that MODIS-derived land cover and surface albedo products yield significant and consistent beneficial impacts in a coupled hydrometeorological model. This study also contributes to ongoing efforts in mesoscale model developments at various institutions—for example, the National Center for Atmospheric Research (NCAR)—by replacing the 10-yr-old land-cover classification scheme in MM5. It is recommended that MODIS-derived land-cover products be implemented within the Weather Research and Forecasting (WRF) modeling framework.

Acknowledgments

Primary support for the research described in this paper came from NASA-Vaccess Grant NAG13-03019. Some partial support was also obtained from NOAA-CREST Grant 219115/219116. The author is grateful to the EOS group at Boston University for providing the MODIS land-cover products and to NASA Gateway for the quick access to MODIS albedo products, and to three anonymous reviewers whose comments improved the quality of this manuscript. The author also thanks NCDC and AZMET for allowing use of their data. Thanks are also due to David J. Gochis and Tom Kovacs for valuable editorial suggestions.

REFERENCES

  • Anderson, J. R., , Hardy E. E. , , Roach J. T. , , and Witmer R. E. , 1976: A land use and land cover classification system for use with remote sensor data. U.S. Geological Survey Professional Paper 964, 28 pp.

    • Crossref
    • Export Citation
  • Belward, A. S., , Estes J. E. , , and Kline K. D. , 1999: The IGBP-DIS global 1-km land-cover data set DISCover: A project overview. Photogram. Eng. Remote Sens, 65 , 10131020.

    • Search Google Scholar
    • Export Citation
  • Chandler, L., , and Tippets D. W. , cited. 2000: NASA satellite data used operationally to help combat fires in the west. NASA News Archive. [Available online at http://earthobservatory.nasa.gov/Newsroom/NasaNews/2000/200009124019.html.].

  • Chen, F., , and Dudhia J. , 2001: Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev., 129 , 569585.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cihlar, J., 1997: GCOS/GTOS plan for terrestrial climate related observations: Version 2. GCOS Rep. 32, World Meteorological Organization, 130 pp.

  • Colle, B. A., , Westrick K. J. , , and Mass C. F. , 1999: Evaluation of MM5 and Eta-10 precipitation forecasts over the Pacific Northwest during the cool season. Wea. Forecasting, 14 , 137154.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Colle, B. A., , Mass C. F. , , and Westrick K. W. , 2000: MM5 precipitation verification over the Pacific Northwest during the 1997–1999 cool seasons. Wea. Forecasting, 15 , 730744.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crawford, T. M., , Stensrud D. J. , , Mora F. , , Merchant J. W. , , and Wetzel P. J. , 2001: Value of incorporating satellite-derived land cover data in MM5/PLACE for simulating surface temperatures. J. Hydrometeor., 2 , 453468.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Crook, N. A., 1996: Sensitivity of moist convection forced by boundary layer processes to low-level thermodynamic fields. Mon. Wea. Rev., 124 , 17671785.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dickinson, R. E., , and Henderson-Sellers A. , 1988: Modeling tropical deforestation: A study of GCM land-parameterization. Quart. J. Roy. Meteor. Soc., 114 , 439462.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1989: Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46 , 30773107.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dudhia, J., 1993: A nonhydrostatic version of the Penn State–NCAR mesoscale model: Validation tests and simulation of an Atlantic cyclone and cold front. Mon. Wea. Rev., 121 , 14931513.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Friedl, M. A., and Coauthors, 2002: Global land cover from MODIS: Algorithms and early results. Remote Sens. Environ., 83 , 287302.

  • Grell, G. A., , Dudhia J. , , and Stauffer D. R. , 1995: A description of the fifth generation Penn State/NCAR mesoscale model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 138 pp.

  • Hansen, M. C., , DeFries R. S. , , Townshend J. R. G. , , Sohlberg R. , , Dimiceli C. , , and Carroll M. , 2002: Towards an operational MODIS continuous field of percent tree cover algorithm: Examples using AVHRR and MODIS data. Remote Sens. Environ., 83 , 304320.

    • Search Google Scholar
    • Export Citation
  • Hong, S-Y., , and Pan H-L. , 1996: Non-local boundary layer vertical diffusion in a medium-range forecast model. Mon. Wea. Rev., 124 , 23222339.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Justice, C. O., , Townshend J. R. G. , , Vermote E. F. , , Masuoka E. , , Wolfe R. E. , , Saleous N. , , Roy D. P. , , and Morisette J. T. , 2002: An overview of MODIS land data processing and product status. Remote Sens. Environ., 83 , 315.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kain, J. S., , and Fritsch J. M. , 1992: Convective parameterization for mesoscale models: The Kain–Fritsch scheme. The Representation of Cumulus Convection in Numerical Models, Meteor. Monogr., No. 46, Amer. Meteor. Soc., 165–170.

    • Crossref
    • Export Citation
  • Kurkowski, N. P., , and Stensrud D. J. , 2003: Assesment of implementing satellite-derived land cover data in the Eta Model. Wea. Forecasting, 18 , 404416.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Loveland, T. R., , and Belward A. S. , 1997: The IGBP-DIS global 1-km land cover data set, DISCover: First results. Int. J. Remote Sens., 18 , 32913295.

    • Search Google Scholar
    • Export Citation
  • Miller, D. A., , and White R. A. , 1998: A conterminous United States multilayer soil characteristics data set for regional climate and hydrology modeling. Earth Interactions, 2 .[Available online at http://EarthInteractions.org.].

    • Search Google Scholar
    • Export Citation
  • Nobre, C. A., , Sellers P. J. , , and Shukla J. , 1991: Amazonian deforestation and regional climate change. J. Climate, 4 , 957988.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pan, Z., , Takle E. , , Segal M. , , and Arritt R. , 1999: Simulation of potential impacts of man-made land use changes on U.S. summer climate under various synoptic regimes. J. Geophys. Res., 104 , 65156528.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., 2001: Influence of the spatial distribution of vegetation and soils on the prediction of cumulus convective rainfall. Rev. Geophys., 39 , 151177.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., , Marland G. , , Betts R. A. , , Chase T. N. , , Eastman J. L. , , Niles J. O. , , Niyogi D. S. , , and Running S. W. , 2002: The influence of land-use change and landscape dynamics on the climate system: Relevance to climate-change policy beyond the radiative effect of greenhouse gases. Philos. Trans. Roy. Soc. London, A360 , 17051719.

    • Search Google Scholar
    • Export Citation
  • Pineda, N., , Jorba O. , , Jorge J. , , and Baldasano J. M. , 2004: Using NOAA AVHRR and SPOT VGT data to estimate surface parameters: Application to a mesoscale meteorological model. Int. J. Remote Sens., 25 , 129143.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Scepan, J., 1999: Thematic validataion of high-resolution global land-cover data sets. Photogram. Eng. Remote Sens., 65 , 10511060.

  • Schaaf, C. B., and Coauthors, 2002: First operational BRDF, albedo and nadir reflectance products from MODIS. Remote Sens. Environ., 83 , 135148.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Segal, M., , Pan Z. , , Turner R. W. , , and Takle E. S. , 1998: On the potential impact of irrigated areas in North America on summer rainfall caused by large-scale systems. J. Appl. Meteor., 37 , 325331.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sen, O. L., , Wang Y. , , and Wang B. , 2004: Impact of Indochina deforestation on the East Asian summer monsoon. J. Climate, 17 , 13661380.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Strahler, A., , Muchoney D. , , Borak J. , , Friedl M. , , Gopal S. , , Lambin E. , , and Moody A. , cited. 1999: MODIS land cover product algorithm theoretical basis document (ATBD). Version 5. [Available online at http://modis.gsfc.nasa.gov/data/atbd/atbd_mod12.pdf.].

  • Tao, W-K., , and Simpson J. , 1993: The Goddard cumulus ensemble model. Part I: Model description. Terr. Atmos. Oceanic Sci., 4 , 3572.

  • Thompson, E. H., , Gutro R. , , and Bettwy M. , cited. 2004: Land cover changes affect U.S. summer climate. NASA News Archive. [Available online at http://earthobservatory.nasa.gov/Newsroom/NasaNews/2004/2004032416710.html.].

  • Townshend, J. R. G., , and Justice C. O. , 1998: Selecting the spatial resolution of satellite sensors required for global monitoring of land transformations. Int. J. Remote Sens., 9 , 187236.

    • Search Google Scholar
    • Export Citation
  • Vermote, E. F., , El Saleous N. , , and Justice C. O. , 2002: Atmospheric correction of MODIS data in the visible to middle infrared: First results. Remote Sens. Environ., 83 , 97111.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wang, Z., , Zeng X. , , and Barlage M. , 2004: Using MODIS BRDF and albedo data to evaluate global model land surface albedo. J. Hydrometeor., 5 , 314.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wei, X., , Hahmann A. N. , , Dickinson R. E. , , Yang Z-L. , , Zeng X. , , Schaudt K. J. , , Schaaf C. B. , , and Strugnell N. , 2001: Comparison of albedos computed by land surface models and evolution against remotely sensed data. J. Geophys. Res., 106 , 2068720702.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wetzel, P. J., , and Chang J-T. , 1988: Evapotranspiration from non-uniform surfaces: A first approach for short-term numerical weather prediction. Mon. Wea. Rev., 116 , 600621.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Xue, Y., 1996: The impact of desertification in the Mongolian and the inner Mongolian grassland on the regional climate. J. Climate, 9 , 21732189.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yucel, I., , Shuttleworth W. J. , , Gao X. , , and Sorooshian S. , 2003: Short-term performance of MM5 with cloud-cover assimilation from satellite observations. Mon. Wea. Rev., 131 , 17971810.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhou, L., and Coauthors, 2003: Comparison of seasonal and spatial variations of albedos from Moderate-Resolution Imaging Spectroradiometer (MODIS) and common land model. J. Geophys. Res., 108 .4488, doi:10.1029/2002JD003326.

    • Search Google Scholar
    • Export Citation
Fig. 1.
Fig. 1.

Nested configurations of MM5 for (a) the mid-Atlantic region and (b) the southwestern United States. Locations of the observational arrays are overlain in these maps.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM536.1

Fig. 2.
Fig. 2.

Spatial distribution of land-cover types across the 4-km model domain covering the parts of the mid-Atlantic region for (a) the USGS and (b) the MODIS-IGBP category. Black “x”s are observation sites. The dashed line shows the location of cross section used in Fig. 9.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM536.1

Fig. 3.
Fig. 3.

Spatial distribution of land-cover types across the 4-km model domain covering the parts of the southwestern United States for (a) the USGS and (b) the MODIS-IGBP category. Black “x”s are observation sites. The dashed line shows the location of cross section used in Figs. 7 and 8.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM536.1

Fig. 4.
Fig. 4.

Horizontal distributions of surface albedo at 4-km resolution for (a)–(c) the mid-Atlantic region and (d)–(f) the southwestern United States. In (a), (b), (d), and (e), the prescribed albedos are shown though in (d) and (e), the spatial correction according to the MODIS land-cover type is included. In (c) and (f), MODIS albedo is displayed for the same domains. The dashed lines in the diagrams for the mid-Atlantic region show the location of cross sections used in Fig. 9 while the ones in the diagrams for the southwestern United States show the location of cross section used in Figs. 7 and 8.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM536.1

Fig. 5.
Fig. 5.

Spatial distribution of the difference in surface skin temperature (a) between MM5-MOD and MM5 and (b) between MM5-MOD-ALB and MM5 for the southwestern United States. (c), (d) Same as (a), (b) but for the mid-Atlantic region. These products were processed from images taken on 6 Aug 2003 at 1300 local time in the southwestern United States and on 14 Nov 2002 at 1300 local time in the mid-Atlantic region. Contours of topography are also shown as background.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM536.1

Fig. 6.
Fig. 6.

From the southwestern United States on 6 Aug 2003 at 1300 local time, (a) the surface wind directions by the MM5 simulation and the surface wind differences (b) between MM5-MOD and MM5 simulations and (c) between MM5-MOD-ALB and MM5 simulations. (d)–(f) The equivalent surface wind diagrams for the mid-Atlantic region on 14 Nov 2002 at 1300 local time. In (b), (c), (e), and (f), the arrows are the wind direction differences, and the colored areas represent the difference in 10-m wind velocity. Surface topography is also shown in (a) and in (d).

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM536.1

Fig. 7.
Fig. 7.

Cross section of the vertical wind field (in m s−1) and the mixing ratio (in g kg−1) taken from the semiarid region on 6 Aug 2003 at 1300 local time for the (a) MM5, (b) MM5-MOD, and (c) MM5-MOD-ALB simulations. Mixing ratio is shown as background while vertical wind is shown with contours.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM536.1

Fig. 8.
Fig. 8.

Along a cross section in the semiarid domain, (a) topography profile and the modeled PBL heights from three cases and (b) surface sensible heat and (c) latent heat fluxes from each simulation. Note that these are from the southwestern domain on 6 Aug 2003 at 1300 local time.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM536.1

Fig. 9.
Fig. 9.

(a) The contour of difference in total cloud water mixing ratio in (g kg−1) (including rainwater, hail, graupel, and snow) between MM5-MOD and MM5 and (b) surface sensible heat fluxes and (c) surface latent heat fluxes from each of the three simulations. These plots are shown along a cross section in the mid-Atlantic region on 13 Nov 2002 at 0100 local time.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM536.1

Fig. 10.
Fig. 10.

(a) The domain-averaged time series of the 2-m air temperature and (b) 2-m specific humidity for the ensemble mean, MM5-MOD, and MM5-MOD-ALB along the first 35 h of the simulation period. Standard deviation bars calculated from four members are also shown through mean ensemble simulation. Domain-averaged calculations are made in the mid-Atlantic region.

Citation: Journal of Hydrometeorology 7, 5; 10.1175/JHM536.1

Table 1.

Translation of IGBP land-use classes by using USGS and SiB classifications and the corresponding parameter sets for each IGBP land-cover class. A value of 999 is equivalent to “not applicable.” Open shrubland of IGBP is equivalent to mixed shrubland/grassland of USGS in the mid-Atlantic region, where the environment is dominantly vegetated. Its parameters in this case are 0.20, 0.005, 170, 100, and 39.18 for albedo, Zo, Rcmin, Rgl, and hs, respectively.

Table 1.
Table 2.

The rms error reduction for 2-m temperature and 2-m relative humidity with the MM5-MOD and the MM5-MOD-ALB runs for each day along the single continuous simulation period in the southwestern United States.

Table 2.
Table 3.

The rms error reduction for 2-m temperature and 2-m specific humidity with the MM5-MOD and the MM5-MOD-ALB runs for each day along the single continuous simulation period in the mid-Atlantic region.

Table 3.
Table 4.

On the average, over the period for which comparison was made at two study areas, the statistics of the rmse and bias in air temperature and humidity are given for each simulation in Table 4.

Table 4.
Save