G. Land-Surface/Atmospheric Interactions and their Modeling

[Background] [Impact of Land-Atmospheric Interactions]
[Land Surface Model Development for Mesoscale Weather Models]
[High Resolution Land Data Assimilation System Development]
[Urban Heat Island Modeling]

1. Background

RAP's objectives in this area are to understand, through theoretical and observational studies, the complex interactions (including biophysical, hydrological, and bio-geochemical) between the land-surface and the atmosphere at micro- and mesoscales. The ultimate goal is to integrate such knowledge into numerical mesoscale weather prediction and regional climate models to improve prediction of the impacts of land-surface processes on regional weather, climate, and hydrology.


2. Impact of Land-Atmospheric Interactions on Quantitative Precipitation Forecast (QPF)

S. Trier, K. Manning (MMM), M. LeMone, and F. Chen (RAP) used high resolution Penn State/NCAR numerical mesoscale model (MM5) simulations, initialized with soil moisture and temperature fields obtained from high-resolution land data assimilation system, to investigate how both surface processes and other aspects of mesoscale forcing influence the initiation of deep convection for a single diurnal cycle (0600 CST 19 June - 0600 CST 20 June 1998) of the dryline case.

Deep convection formed along or slightly ahead of the NE-SW oriented surface moisture gradient (dryline) from west Texas into central Oklahoma by mid to late afternoon on 19 June (Figure 2.1 ). The dryline was particularly intense across west Texas. By contrast, the dryline was slower to develop and somewhat weaker across Oklahoma, but did eventually intensify by late afternoon (Figure 2.1b). Comparison of infrared satellite data with radar-rain gauge precipitation estimates (not shown) revealed timing and location (with respect to the dryline) of the onset of deep convection similar to results from the control simulation.

Analysis of convection-resolving MM5 simulations, which utilized HRLDAS soil states as initial conditions, revealed the following:

• Several scales of motion appear critical to convection initiation in the vicinity of the dryline.
  
  
• Deep convection was later initiated from energetic small-scale (~ 10 km) horizontal convective roll circulations within a region of enhanced surface sensible heat flux located along and slightly east of the dryline.
  
  
• Roll-like circulations confined to the PBL appear to be the primary agent in directly initiating the late-afternoon deep convection in Texas. These rolls occurred within a large-scale (100-200 km) region of increased late-morning to mid-afternoon PBL height, which appeared to be associated with thermally-induced mesoscale circulations.
  
  
• Accurate specification of fine-scale gradients of soil moisture on the diurnal evolution of the PBL and the location and timing of storm initiation in the vicinity of an observed Southern Great Plains dryline is critical.
  
  
• Initialization of fine-scale gradients of soil moisture is clearly important in the simulated evolution of the dryline and convective initiation for the current case. When MM5 was initialized with the relatively coarse and somewhat too-wet soil moisture from the NCEP Eta model, the intensity of precipitation over Texas was greatly reduced (Figure 2.2b). Also, deep convection was initiated over 100 km east of the dryline, compared to that at the vicinity of dryline in both the control simulation (Figure 2.2a) and observations.

Figure 2.1. Surface winds (barbed symbols), water vapor mixing ratio (g/kg, contoured every 2 g/kg), and hourly accumulated precipitation (mm, color scale at right) on Domain 3 of the control simulation valid at (a) t=10h (1600 CST 19 June 1998), and (b) t=11h (1700 CST 19 June 1998). Tick marks on the border have a spacing of one horizontal grid length (3.3 km) within the 558 x 558 km² innermost domain.

Figure 2.2. MM5 Domain 3 (3-km) water vapor mixing ratio (g/kg, contoured every 2 g/kg) valid at t=9h (1500 CST 19 June 1998) and 3-h accumulated precipitation (color scale at right) valid for the period of t=9-12 h (1500-1800 CST 19 June 1998) for simulations (a) initialized with soil moisture from HRLDAS; control simulation, and (b) initialized with the coarser resolution soil moisture field used in ETA.

3. Land Surface Model Development for Mesoscale Weather Models

Supported by the Air Force Weather Agency (AFWA), M. Tewari and F. Chen collaborated with NCEP/NOAA and AFWA to develop and implement an advanced land-surface modeling (LSM) system in the Weather Research and Forecast (WRF) model. RAP organized a WRF/LSM working group workshop held at NCAR in June 2003 with participants from NCEP, AFWA, FSL, NCAR, and universities to define the implementation and test strategy for the coupled unified Noah LSM system. The unified Noah LSM has been implemented in the MM5 3.6, available since February 2003, and in a test version of WRF. It is expected to be released in the research version of WRF by the end of 2003.

The coupled WRF/Noah LSM system was tested in realtime mode(with 22-km grid spacing) at NCAR to perform once-daily 48-h forecast. It was initialized with 0000 UTC EDAS (both atmospheric and soil conditions, its precipitation EQ threat and bias score were verified against NCEP Stage-II rainfall data. Figure 3.1 shows that the WRF/Noah (blue line) forecasts significantly improved the precipitation skills for the majority of rain categories, particularly for light rains, as compared to WRf/OSULSM (cyan line). Note that the only difference in these two coupled WRF models is the land surface model.

Figure 3.1. Precipitation skill scores for WRF and several operational forecasts from 1 to 30 May 2003. The forecasts are verified over the CONUS by comparison to an analysis on a Lambert Conformal grid. Grid boxes containing no observations for a given day are not included in the verification. WRFMASS (blue): WRF coupled to the OSULSM (predecessor of the unified Noah LSM); WRFNOHA: WRF coupled to the unified Noah LSM. Courtesy of NSSL/NOAA.

4. High Resolution Land Data Assimilation System Development

A High-Resolution Land Data Assimilation System (HRLDAS) is being developed by K. Manning, D. Gochis, D. Yates, and F. Chen to support application of high-resolution numerical weather prediction models. It uses observed hourly precipitation, solar radiation derived from satellite and analyzed surface wind and temperature to force a land-surface model to simulate the evolution of soil moisture. In this system, the NCEP/NOAA hourly 4-km rainfall analysis, based on NEXRAD and rain gauge observations, is used so that errors in soil moisture caused by precipitation and radiation bias in coupled modeling systems could be avoided. Long-term HRLDAS simulations were conducted over the US SGP regions to support RAP land atmospheric interaction study project for IHOP 2002.

The HRLDAS characterizes soil moisture/temperature and vegetation variability at small scales (~4km) over large areas in order to provide improved initial land and vegetation conditions for the MM5/LSM coupled model for our QPF simulations. As shown on Figure 4.1, the HRLDAS soil moisture field captured not only the general west-east soil-moisture contrast, but also fine-scale heterogeneity caused by small-scale convective rain, land use types, soil texture, and vegetation characteristics.

Figure 4.1. 4-km grid volumetric soil moisture at 5 cm below the ground surface simulated by HRLDAS for the IHOP 2002 area for 1200 UTC 31 May 2002.

5. Urban Heat Island Modeling

Correctly modeling the urban effects on the atmosphere in mesoscale models (MM5 and WRF) and its assimilation into systems such as RTFDDA is critical, because their forecasts and analysis are often used as input for atmospheric dispersion models and for planning field experiment. The current Noah land surface model in MM5 has an overly-simplified urban representation, which merely increases the roughness length and reduces surface albedo for urban land use In the long run, in collaboration with Dr. Kusaka, (a visitor from Japan), and F. Chen are coupling an advanced one-layer urban-canopy model, based on work by Dr. Kusaka, with the Noah LSM and implement them into mesoscale models.

To provide realtime support for the 2003 Joint Urban Atmospheric Dispersion Study field experiment in Oklahoma City (OKC), an intermediate approach was adopted. The current Noah LSM was enhanced to capture primary influences of urban areas from bulk heat and momentum transfer without considering urban geometry to balance complexity and data input requirements for RTFDDA. The Noah LSM urban enhancements included increasing the roughness length from 0.5 m to 0.8 m, reducing surface albedo to represent the shortwave radiation trapping in the urban canyons, using a larger volumetric heat capacity for the urban surface (walls, roofs, and roads), increasing the value of soil thermal conductivity to parameterize large heat storage in the urban surface and underlying surfaces, and reducing green vegetation fraction over urban city to decrease evaporation.

The urban heat island effects seem prominent in the 9-day average of 1-km grid RTFDDA forecasts (Figure 5.1); the OKC area is ~ 2-3^oC warmer than the rural regions. Note that on Figure 5.1b this nocturnal urban heating is able to keep the lower boundary layer slightly unstable (with heat transferred from surface to the atmosphere), while the surrounding rural areas are mostly stable because of a surface inversion. Examination of daily RTFDDA forecasts revealed even more pronounced influences of urban land use on the atmosphere. For instance, the PBL height over the core OKC urban region was about 100 m higher than over the rural regions. Due to stronger mixing in the nocturnal convective mixed layer, and hence less decoupling of the surface layer with the atmosphere, the strength of low-level jet over OKC was weaker than in the surrounding areas.

Figure 5.1. Averages of nine clear-sky day July 2003 RTFDDA forecasts for Domain 4 (1-km grid spacing) valid at 06 UTC (about local midnight). a) 2-m temperature, and b) sensible heat fluxes.

[Top]