A Basis for Achieving Economic, Societal and Environmental Goals in Illinois
Xin-Zhong Liang
Regional Climate Model (RCM)
The CWRF is the Climate Extension of the Weather Research and Forecast model (WRF), incorporating all its functionalities
for numerical weather predictions while enhancing the capability for climate
applications (see Liang et al. 2004b for an introductory description). The WRF was originally designed mainly for
numerical weather prediction (NWP) and not expressly for climate studies. To
extend its capability for applications on regional climate scales, we have developed the CWRF with four
crucial characteristics to improve: [1] planetary-mesoscale
interaction by including an optimal buffer zone treatment that integrates
realistic energy and mass fluxes across the lateral boundaries of the regional
climate model (RCM) domains (Liang et al. 2001); [2] surface-atmosphere interaction by incorporating new physics modules
for planetary boundary layer, land surface and terrestrial hydrology as well as
observed variations or dynamic predictions of vegetation, ocean and sea ice;
[3] convection-cloud-radiation interaction
by implementing fully-coupled, new physical parameterizations for cumulus,
cloud microphysics, cloud formation, and radiative transfer; and [4] system consistency throughout all process modules by utilizing unified
water vapor saturation and solar zenith angle functions, common physical
constants and coherent tunable parameters. These improvements have been
accomplished through iterative, extensive model refinements, sensitivity
experiments, and rigorous validations over the past 3 years. As a result, the
CWRF has demonstrated greater capability and better performance in simulating
the U.S.
regional climate than the CMM5 (Liang
et al. 2001, 2004a).
A series of papers are being prepared to
document details of the CWRF formulations and skills in weather forecasts and
climate predictions. The first of the series depicts the construction and
implementation of surface boundary conditions, where the CWRF concept and its
major modules representing surface-atmosphere interactions are briefly
described (Liang et al. 2004b). The latest updates are being currently carried out to further
improve surface-atmosphere [2] and convection-cloud-radiation [3] interactions,
and will be completed before the proposed project starts. Most relevant
to this proposed research is the land surface representation, which is
described in more detail below.
We have
so far conducted various sensitivity and validation experiments of the initial CWRF
(without the latest updates) for the 1993 U.S. Midwest flood case, where the
LBCs were constructed from the NCEP-DOE AMIP-II reanalysis (Kanamitsu et al.
2002). The grahic, left,
compares CWRF simulated regional mean rainfall daily variations during May
2-July 31, with observations and CMM5 outputs. For the
Midwest
major flood area, the CMM5 reproduces different climate regimes, where observed
rainfall was identified with the periodic (5-day) passage of mid-latitude
cyclones in June and persistent synoptic circulations in July. The CWRF
realistically simulates the June variations, but has less skill in July with
insufficient rainfall between days 4-18
.
This deficiency likely results from incomplete representation of regional water
recycling processes, which is anticipated to be largely eliminated by the
latest updates. In particular, the incorporation of more realistic
surface boundary conditions (Liang et al. 2004b), including new surface albedo
parameterization, bedrock depth, soil and vegetation properties, will
significantly improve regional climate simulations.
On the other hand, both CWRF and CMM5 produce an excellent simulation of
observed rainfall variations during the whole period over Cascade, where the
orographic effect dominates. Over the U.S. Northeast, the observed
temporal evolution is generally well simulated by both RCMs, with somewhat
overestimated (underestimated) rainfall intensity in CWRF (CMM5).
Note that the rainfall associated with the
North American summer monsoon over Mexico is totally missed in the CMM5,
while the geographic distributions and daily variations are very realistically
reproduced by the CWRF. The rainfall intensity seems to be overestimated in the
CWRF, though the observed data has a coarse resolution and large uncertainty.
RCM Mother Domain
The RCM mother domain, centered at (37.5°N, 95.5°W),
covers the entire continental U.S. and
adjacent oceans using the Lambert conformal map projection
with
a 30-km grid spacing. This domain has been fully tested (Liang et al.
2001) and extensively used in our ongoing research projects (Liang et al.
2004a, b). The RCM is driven by lateral boundary conditions (LBCs)
within four narrow buffer zones at the mother domain edges (Liang et al. 2001).
These LBCs are provided, at a coarse-resolution, by global reanalysis data for
validation/evaluation purposes or by GCM simulations for prediction/projection.
The RCM integrates the LBCs and generates its own mesoscale circulation within
the domain interior.
It may further
include a nested domain at a 10-km grid spacing over the Midwest,
where interaction with the mother domain can be one-way, two-way or
completely deactivated
. When needed, a refined inner domain can be nested to facilitate the CRM
capability with a 2.5-km grid spacing to more closely depict local processes
(such as hydrology, air and water quality) that are keenly important for
societal impact studies.
CAQIMS Components
CAQIMS Home
