General Circulation Model Linkage to Mesoscale Model
Objectives
A primary goal of SRBEX is linking global-scale climate change to
regional climate with a particular emphasis on the hydrologic
components. An important component of this effort is comparing
statistics from a global-scale general circulation model (GCM), from a
regional-scale climate model, and from actual meteorological
observations to characterize model performance and improve the linkage
between models. The initial comparisons use horizontal grid spacings of
approximately 500 km and 100 km for the GCM and regional-scale models,
respectively. These results will provide the necessary framework for
using a regional-scale model with nesting capabiities, where the
innermost nested domain would have a grid spacing below 10 km.
Current Activities
Coupled GCM/regional-scale simulations have been undertaken with the
GENESIS GCM (Thompson and
Pollard, 1994; Pollard and
Thompson, 1994)and a mesoscale model (MM Version 4) which has been
altered for climate studies. The altered version of the MM is known as
the RegCM2 (Giorgi et al., 1993a;
1993b)and has been used in a number of climate studies. RegCM2 is
driven at the lateral boundaries (outer 4 grid points) with
meteorological fields (temperature, wind components, humidity, and
surface pressure) from the GENESIS GCM every 12 hours.
The GENESIS simulation has been integrated for 10 years using observed
sea-surface temperatures (the AMIP data set) for the 1979-1988 time
period. The GENESIS GCM uses a horizontal resolution of 4.5 (latitude)
by 7.5 (longitude) degrees and 12 vertical levels. It also uses a
land-surface package (LSX) in which heat, momentum, and moisture fluxes
from the biosphere are computed on a horizontal grid with a resolution
of 2 by 2 degrees. The coupled GCM/RegCM2 simulations are being run
for the same ten-year time period.
The primary focus of the RegCM2 simulations has been on the Eastern
U.S., since the Susquehanna River Basin is the area of interest.
Accordingly, one initial set of simulations has used a domain centered
on central Pennsylvania (41 degrees north latitude, 77 degrees west
longitude), using a 30 by 30 array of 108km-square grid cells. This
places the western boundary of the RegCM2 grid in Kansas, the eastern
boundary over the western Atlantic, the northern boundary over
Hudson Bay and the southern boundary over the southern U.S..
To study the effect of lateral boundary conditions on the RegCM2
results, a second set of simulations has used an expanded domain that
includes the 48 conterminous U.S. and parts of southern Canada, the
eastern Pacific, and the Gulf of Mexico. This domain is centered near
the Kansas-Missouri border (40 degrees north latitude, 95 degree west
longitude) and uses a 60 (east-west) by 42 (north-south) array or 108 km
grid cells.
Results
To date, only the model outputs for about the first year, beginning 1
December 1979, have been examined carefully. Results for the first six
months are presented in detail in the 1994 annual report
and the references cited therein.
Highlights of the results to date include