A41B-01 INVITED 08:00h
Internal Variability in Nested Regional Climate Simulations
The high computational cost of regional climate simulations has meant that studies using regional climate models usually have relied on single realizations for each situation of interest. In some cases the sensitivity to different physical parameterizations, domain sizes, or other aspects of model configuration have been examined, but again there has almost always been only a single realization for each configuration. This raises the question of whether conclusions can reliably be drawn from a single realization; i.e., we must ask, What is the magnitude of internal variability in a regional climate model? Here we evaluate three realizations using the regional climate model RegCM3 to simulate the period 1986-2003 over a domain covering the continental U.S. and most of Mexico at 52 km grid spacing. Initial and lateral boundary conditions were taken from the NCEP-DOE Reanalysis 2. The three RegCM3 realizations were initialized one month apart but were otherwise identical in configuration, so that their collective behavior provides a measure of internal variability of the model. Internal variability is found to differ both by subregion and by season. Precipitation fields produced by the ensemble members tend to diverge during the warm season and then re-converge during the cool season. Spread of the ensemble precipitation is smallest for regimes that have a summer-dry (Mediterranean) climate. Reasons for internal variability are discussed, most notably the roles of soil moisture, convective precipitation, and feedbacks between moisture and precipitation.
A41B-02 INVITED 08:15h
The Development of the Eta Regional Climate Model and its Application to Warm Season Precipitation Simulation
To examine seasonal climate predictability using a regional model, in this study we developed and tested a high resolution, Eta model-based, Regional Climate Model (Eta RCM). The model was based on the NCEP operational Eta model (as of the spring of 2003, namely, the Eta model version in the NCEP 25-year Regional Reanalysis), with changes made to make the model run over a longer time period and to update the sea surface temperature (SST), sea ice, green vegetation fraction, and surface albedo fields on a daily basis. For this study, we executed the Eta RCM on the same large domain as used in both the operational Eta model and the Regional Reanalysis (RR), and using a resolution identical to that of the RR (32 km horizontally and 45 levels vertically). Presently, the model can be executed off of analyzed lateral boundary conditions of the NCEP Global Reanalysis 1 and 2, or predicted lateral boundary conditions from NCEP's global Coupled Forecast System (CFS) used for seasonal climate prediction. To examine the impact of the chosen source of initial land states and to test the skill of the Eta RCM in summer precipitation simulations, two interannual pairs of summer cases were chosen, namely, 1990 versus 1991 and 1988 versus 1993. The years 1990 and 1991 were wet and dry years, respectively, in the southwest U.S. monsoon region of interest to the North American Monsoon Experiment (NAME). The years 1988 and 1993 are known for their central U.S. drought and flood episodes, respectively. Most previous studies of RCM seasonal simulations driven by global reanalysis lateral boundary conditions and observed SST utilize one single member initialized from one single date. In contrast, we executed six members whose starting dates vary by one and a half days. The simulation period is from late May through September. In addition to executing an ensemble of realizations, we tested two different sources for the initial land states (soil moisture and soil temperature), namely, one from the NCEP/DOE Global Reanalysis 2 (GR2) and one from the NCEP Regional Reanalysis (RR). We examine the resulting Eta RCM seasonal simulations of precipitation to a) demonstrate the impact of the source of initial land states, b) illustrate whether the simulations capture reasonable interannual variability, and c) examine the extent of member-to-member variability. The results show that the use of RR rather than GR2 initial land states significantly increases the precipitation amounts in the monsoon region of the southwest U.S. and northwest Mexico. Additionally the 1990 versus 1991 simulations capture a meaningful signature of the observed interannual precipitation difference over the monsoon region. Similarly, the 1988 (drought) versus 1993 (flood) simulations reasonably capture the observed interannual precipitation difference over the central U.S. Lastly, significant spread in the precipitation patterns is evident among the six members, indicating that previous RCM studies that employed only one member may be misleading by failing to represent the inherent internal variability in the simulations of RCMs executed over large domains.
A41B-03 08:30h
OLAM: A new Generation Earth System Model
The Ocean-Land-Atmosphere Model (OLAM) is a multi-scale numerical simulation model developed at Duke University to investigate a wide range of phenomena important to weather, climate, and the ecosystem, from global to local scales. The model is based largely on the Regional Atmospheric Modeling System (RAMS) but has been extended to global coverage and contains several innovative improvements. Like RAMS, OLAM may be configured with any number of nested grids to achieve arbitrarily high resolution wherever desired. Physical parameterizations for radiative transfer, microphysics, turbulence, cumulus parameterization, and land surface energy and water budgets are adopted directly from RAMS and thus have a long history of development, testing, and verification. This overall approach to OLAM brings the best of sophisticated mesoscale modeling methods to a global model framework. This talk will highlight the major features of OLAM, focusing on new methods that were developed to achieve the design goals of the model. Examples of its performance for global and local simulations will be presented.
A41B-04 08:40h
The FSU Regional Spectral Model and its Application to Seasonal Precipitation Prediction over the Southeast U.S.
We have developed a regional spectral model for weather and climate studies and prediction. The regional model is embedded in the FSU Global Spectral Coupled Model (FSUGCM), though in principle it can be coupled to other models or analyses, or nested within the regional model itself. The Florida State University Regional Spectral Model (FSUNRSM) utilizes the spectral method in the horizontal direction using Double Fourier trigonometric series. The regional model is a perturbation model, meaning that only deviations from the global, or base, solution are represented by the spectral functions. The regional model was designed to be compatible with the FSUGSCM. As a result, the regional model has available to it the same array of physical parameterizations, including six convection schemes, the FSU physics package and most of the NCAR CCM3.6 atmospheric physics package. The regional model also shares the same sigma-cooridinate vertical structure with Charney-Phillips staggering. The FSUNRSM is very similar in concept to the NCEP Regional Spectral Model, though with some significant differences in implementation. We will provide a brief overview of the model, including some recent enhancements, such as the inclusion of the Community Land Model version 2. We will also present some results for seasonal predictions of rainfall over the Southeast U.S. In these experiments, we ran 12 4-month integrations for the years 1986-1997 starting November 1 of each year. The seasonal precipitation anomalies (December-February) were reasonably well simulated by both the global and regional models, with the regional model performing somewhat better. More importantly, the regional model was better able to simulate the frequency of rainfall events than the global model, and in reasonable agreement with cooperative station data.
A41B-05 08:50h
Transferability as a Strategy for Researching the Water Cycle and Energy Budget at Regional Scales
Transferability is a research strategy designed to advance our understanding of physical processes underpinning the global water and energy cycles and their predictability through systematic intercomparisons of regional climate simulations on several continents and comparison of these simulated climates with coordinated continental-scale observations and analyses. A transferability working group is being established under the GEWEX Hydrometeorology Panel to address GEWEX science questions relating to feedbacks and natural variability, acceleration of the water cycle, seasonal to interannual predictability, and impacts on water resources. Through simulations of unique water cycle processes in several climate regimes, transferability experiments seek answers to questions such as "How portable are our models?" and "How general is our understanding of the physics underlying our models?" Transferability experiments make "meta-comparison" of individual model and model ensemble performance among domains as well as on particular domains. Anchored by coordinated observations from continental scale experiments, transferability studies will examine influences of parameterization choices, resolution and nesting dependencies, boundary influences, and internal model variability on the quality of predictions. The emphasis of the transferability experiments is to apply several models on several domains or single models on multiple domains as opposed to multiple models on single domains. The presentation will use an initial set of simulations to illustrate how transferability experiments seek to answer science questions relating to the water cycle and energy budget. We have performed multi-year simulations focusing on regions encompassing GEWEX Continental Scale Experiments (e.g., GAPP, BALTEX, LBA), using the ECPC Regional Spectral Model, the GKSS Lokallmodel and RegCM3. Particular attention will be given to how model accuracy changes with location.
A41B-06 09:00h
Initial results from the North American Regional Climate Change Assessment Program
The North American Regional Climate Change Assessment Program (NARCCAP) is using an ensemble of global and regional climate models (GCMs and RCMs) to produce downscaled climate change scenarios. The ensemble will provide opportunity to estimate regional climate changes as well as their uncertainty. This talk presents results from initial runs by NARCCAP modelers, focusing on sensitivity to domain size. The RCM domains must cover most of North America to satisfy needs of associated climate change impacts programs, raising questions about the influence of domain size on results. NARCCAP RCM modelers are thus conducting a systematic sequence of domain-size tests with each of their participating models. The collective testing also provides a prototype framework for NARCCAP's overall simulation and assessment program. Tests include several models using NCEP/DOE Reanalysis 2 to simulate 1979 for domains covering most of North America or larger. We assess domain sensitivity and its significance through comparisons with several observed fields (e.g., temperature, precipitation, circulation). These tests form an important foundation for estimating uncertainty in regional climate change scenarios.
A41B-07 09:10h
Study of Air-Sea Interaction over Hudson Bay using Canadian Regional Climate Model Coupled with Regional Ocean Model
The Canadian Regional Climate Model (CRCM) has been coupled with Regional Ocean Model (ROM) over the Hudson Bay. The CRCM is developed in the University of Quebec at Montreal and ROM in the Institute of Maurice Lamontagne, Fisheries and Oceans Canada. The two models are coupled with Fortran Pipe to exchange forcing data in every 30 minutes. After 4 years' simulation, the coupled model proves to be able to reproduce right ice concentration and ice position, which are two very important factors to influence the regional climate in northern America. The annual change of sea ice in Hudson Bay is also well simulated by the coupled model. The coupled model shows the cyclonic surface current in Hudson Bay during the period of summer and fall when ice is free. In winter and spring, the surface current shows a quite complicated structure. The surface current in winter and spring is different from year to year. Especially in spring, the surface current is much more complicated due to the presence of ice, fresh water from melting ice and very stable atmospheric boundary layer condition which makes surface wind stress uncertain.
A41B-08 09:20h
Interannual Variability of South American Climate in a Regional Climate Model and Observations
The nested climate modeling approach is increasingly being used to address the need for higher spatial and temporal resolution climate information from seasonal forecasts and climate change scenarios. To evaluate this approach, an ensemble nested model climatology for South America is being performed using a regional climate model (RegCM3) driven with lateral boundary conditions from a general circulation model (GCM; ECHAM4) and NCEP/NCAR reanalyses, and using observed sea surface temperatures (SSTs) for the period 1982-2003. We will present preliminary results from the first 10 years of model integrations. The annual cycle and interannual variability of the nested model climate will be compared against both observations and the driving CGM. In particular, we will examine the climate features of the tropical and subtropical South American sector including the ITCZ, SACZ, and South Atlantic subtropical high and their relationships to moisture transport and continental rainfall and their representation in the RegCM and GCM. Discussion will focus on the ``added value'' of this dynamical downscaling methodology.
A41B-09 09:30h
Seasonal Prediction of the Regional Eta/SSiB model in North American Climate Stu
This study aims to understand the predictability of the Regional Eta/SSiB model in the North American regional climate study. We mainly focus on model simulations of precipitation and other relevant hydro-meteorological variables. To achieve our goals, a series of sensitivity studies are designed to explore the role of a variety of factors in water cycle simulations. These factors include domain size, lateral boundary conditions, horizontal resolution, sea surface temperature, and convective parameterizations. The summer of 1998 is selected for the experiments. The results show that the domain size and lateral boundary conditions have the most significant impacts on the simulations. In the domain size experiment, a set of domains is selected for testing. The domain that covers only continental U.S. produces the best simulations in temporal evolution and spatial distribution of the precipitation. When the domain is expanded to the southern oceans, although the model is still able to produce reasonable large-scale circulation, the simulations of the intensity and the spatial distribution of the precipitation deteriorate dramatically when the ocean areas become large. It is likely that there is a southern barrier for the Eta model with reanalysis data and also there could be a critical spatial scale beyond which regional climate starts interacting strongly with the global climate. The lateral boundary conditions are also crucial for regional climate studies as the regional model tends to maintain the imposed large-scale features. Because the 200mb zonal wind and low level moisture transfer play crucial roles in proper simulations of the 1998 US summer precipitation, maintaining the same boundary conditions produce similar large scale patterns and cause limitations in the climate sensitivity study. For example, when investigating the deep soil temperature effect using the Eta model, it was found that with the same lateral boundary conditions in both control Eta run and anomaly Eta run, the impact of deep soil temperature is not clear. The climate anomaly signals become evident only when deep soil temperature anomalies are imposed to both regional Eta model and the GCM, which provides the lateral boundary conditions for the Eta model, in the anomaly runs.
A41B-10 09:40h
Investigating the Climatic Effects of the NAO Over Greenland Using Polar MM5
The North Atlantic Oscillation (NAO) is the leading mode of variability of the extratropical atmospheric circulation in the Atlantic region. In the current work, we employ the Polar MM5 mesoscale atmospheric model to study NAO-related climate variability over the Greenland ice sheet. Emphasis is placed on assessing the relative importance of ice sheet topography, sea surface temperature and sea ice in generating the climatic signature of the NAO in this region. Implications for future climate change are also discussed.
A41B-11 09:50h
Impact of anthropogenic land-use/land-cover change on climate & hydrologic cycle of a semi-arid region
Recent modeling work with the Regional Atmospheric Modeling System (RAMS), validated with observations, has shown that the impact of land-use/land-cover (LULC) change over a span of twenty years in the Greater Phoenix area has had important consequences on the dynamic and thermodynamic characteristics of the regional Arizona climate. In particular, we investigate the impact of anthropogenic LULC change, derived with remotely sensed observations from NASA satellites, on the climate and hydrologic cycle of the Greater Phoenix area. Numerical modeling results using LULC data derived from 1970's Landsat data were compared with simulations making use of the National Land Cover Dataset for 2001 (NLCD 2001). We find that simulation of event-scale to seasonal precipitation in the Greater Phoenix area during the wettest part of the year, that of the North American Monsoon System (NAMS), is sensitive to these land-surface changes. These results, in the context of the facts that Phoenix is one of the most rapidly expanding U.S. cities and that, being in an arid region, its water management challenges are significant, have implications for the future climate, environment, and quality of life in the region.