MM5 Application to Modeling the Measurements from the ABLE Site

V. R. Kotamarthi, R. L. Coulter, I. Y. Lee and M. L. Wesely

Environmental Research Division, ANL

The Atmospheric Boundary Layer Experiment (ABLE) study was designed to improve the current understanding of the physics of the boundary layer and provide data for improving the boundary layer parameterizations currently in place in mesoscale meteorological models. The ABLE site is located in the Walnut River watershed in southeastern Kansas. In a region of less than 5000 km² several observation stations have been established to collect data continuously on temperature, wind fields, precipitation, surfaces fluxes and moisture at high spatial and time resolution.

We recently began simulating conditions at ABLE with the parallel version of the MM5 modeling system version 2 on the IBM SP at Argonne's Mathematics and Computer Science Division (ANL/MCS). The primary aims of these calculations are as follows:

a. Estimate the model bias for various seasons in predicting the temperature, wind fields, and precipitation in a midwestern setting.

b. Quantify the significance of model-predicted increases for a future climate.

c. Evaluate the model bias for several boundary layer parameterizations, cloud parameterizations, and land surface models currently available.

d. Integrate a land surface, vegetation and soil moisture model developed at ANL that uses satellite derived products with the meteorological model to estimate the effects of soil moisture variability on model-predicted precipitation and temperatures.

NCEP upper air sounding data sets obtained from NCAR were used to initialize the MM5 model runs. Calculations were performed from May 5 through June 20 of 1997. The parallel version of MM5 developed at ANL and running on the IBM SP2 at ANL/MCS was employed for these calculations.

The model was set up with a horizontal resolution of 9 km for the coarse grid with a total number of 45x67x23 grid points. The coarse grid covered approximately 40 N to 34 N latitude and 91 W to 100 W longitude. The model top boundary was at 100 Mb. A single nesting with a horizontal resolution of 3 km was placed to cover the ABLE site. The nested domain extended from 39 N to 36 N latitude and 95 W to 99 W longitude. The physics options chosen were as follows: IMPHYS =3 (warm rain); ICUPA =3 (Grell type cumulus parameterization); IBLTYP=5 (MRF scheme); and IFRAD=1 (simple radiation scheme). The model time step was set at 27 seconds, and the model calculations were made in the nonhydrostatic mode.

As illustrated in Fig. 1, the temperature variations predicted by the model were usually in phase with the temperature measured with an ABLE Radio Acoustic Sounding System ( RASS). The Blackader scheme produced temperatures approximately 2 C cooler at 0.5 km compared to the MRF scheme for the period tested (Fig 2). A detailed comparison between model and measured temperatures at an altitude of 0.5 km was carried out for one of the sites located near Beaumont, KS ( 37.63 N and 96.5 W). The differences in the virtual temperatures from the RASS versus the model are show in Fig. 3 ( May 1997) and Fig. 4 (June 1997). The model calculated temperatures are smaller than the RASS temperatures by 2-4 C, with a maximum probability of approximately 2 C.

The differences in the model calculated and RASS-measured profiles of virtual temperature and wind velocities are shown Figs. 5 and 6 for June 5, 1997. The model tends to produce lower temperature than the observed values below 1.2 km for this particular day. Similar results were obtained for the other days modeled. The calculated wind speeds below 0.5 km ranged 5 m/sec larger than measurements to being smaller by about 5 m/sec.

Fig. 5