24 Search Results

One of the key factors in the improved understanding of climate science is the development and improvement of high fidelity climate models. These models are critical for projections of future climate scenarios, as well as for highlighting the areas where further measurement and experimentation are needed for knowledge improvement. In this paper, we focus on several computing issues associated with climate change modeling. First, we review a fully coupled global simulation and a nested regional climate model to demonstrate key design components, and then we explain the underlying restrictions associated with the temporal and spatial scale for climate change modeling. We then discuss the role of high-end computers in climate change sciences. Finally, we explain the importance of fostering regional, integrated climate impact analysis. Although we discuss the computational challenges associated with climate change modeling, and we hope those considerations can also be beneficial to many other modeling research programs involving multiscale system dynamics.

In the duration of this project, we finished the main tasks set up in the initial proposal. These tasks include: collecting needed data of regional aerosol emissions (mainly SO2); building the RICES model; conducting preliminary simulation runs on some policy scenarios. We established a unified and transparent IA modeling platform that connecting climate change and local air pollution. The RICES model is the pioneering IA model that treats climate change and local air pollution as correlated global and local stock externalities.

Regional air quality simulations are conducted for four summers (2001, 2002, 2051, and 2052) to examine the sensitivity of air quality to potential regional climate change in the U.S. In response to the predicted warmer climate in 2051/2052, emissions of isoprene and terpene increase by 20-92.1% and 20-56%, respectively, over most of the domain. Surface O3, which is sensitive to changes in temperature and solar radiation but relatively insensitive to changes in PBL height and cloud fraction, increase by up to 19-20%. PM2.5, its compositions, and visibility exhibit an overall negative sensitivity (decrease by up to 40%), resulting from the competition of the negative temperature effect and positive emission/temperature effects. While the response of dry deposition is governed by the negative sensitivity of surface resistances, that of wet deposition is either positive or negative, depending on the relative dominancy of changes in PM2.5 and precipitation. Overall the net climatic effect dominates changes in O3, PM2.5, wet and total deposition, and the net biogenic emission effect is important for isoprene, organic matter, visibility, and dry deposition over several regions. Models that do not include secondary organic aerosol formation from isoprene photooxidation may underestimate by at least 20% the air quality responses to future climate changes over many areas of the modeling domain. Both regional climate and air quality exhibit interannual variability, particularly in temperature, isoprene emissions, and PM2.5 concentrations, indicating a need for long-term simulations to predict future air quality.

The final report describes the study aimed at exploring the variable-resolution stretched-grid (SG) approach to decadal regional climate modeling using advanced numerical techniques. The obtained results have shown that variable-resolution SG-GCMs using stretched grids with fine resolution over the area(s) of interest, is a viable established approach to regional climate modeling. The developed SG-GCMs have been extensively used for regional climate experimentation. The SG-GCM simulations are aimed at studying the U.S. regional climate variability with an emphasis on studying anomalous summer climate events, the U.S. droughts and floods.

At Argonne National Laboratory we have developed a high performance regional climate modeling simulation capability based on the NCAR MM5v3.4. The regional climate simulation system at Argonne currently includes a Java-based interface to allow rapid selection and generation of initial and boundary conditions, a high-performance version of MM5v3.4 modified for long climate simulations on our 512-processor Beowulf cluster (Chiba City), an interactive Web-based analysis tool to facilitate analysis and collaboration via the Web, and an enhanced version of the CAVE5d software capable of working with large climate data sets. In this paper we describe the application of this modeling system to investigate the role of model resolution in predicting extreme events such as the ''Hurricane Huron'' event of 11-15 September 1996. We have performed a series of ''Hurricane Huron'' experiments at 80, 40, 20, and 10 km grid resolution over an identical spatiotemporal domain. We conclude that increasing model resolution leads to dramatic changes in the vertical structure of the simulated atmosphere producing significantly different representations of rainfall and other parameters critical to the assessment of impacts of climate change.

The paper addresses the need to increase the temporal and spatial resolution of meteorological data currently used in air quality simulation models, AQSMs. Transport and diffusion parameters including mixing heights and stability used in regulatory air quality dispersion models are currently computed using routinely collected twice daily (00Z and 12Z) upper air sounding data from approximately 100 locations, and hourly surface data from approximately 300 locations spread across the US. The limited resolution data base limits the accuracy, reliability and validity of the derived dispersion parameters for air quality models. The MM4-FDDA (Mesoscale Meteorological model, Version 4 with Four Dimensional Data Assimilation), a state-of-science dynamic modeling system that assimilates routinely available surface and upper air sounding data has been demonstrated to provide accurate and well characterized primary meteorological fields on hourly time intervals, with fine horizontal and vertical resolution. The MM4-FDDA can also provide as outputs, parameters such as surface heat and momentum fluxes which are necessary for computing the stability dispersion parameters. The goal of the study is to generate meteorological data with accuracy of transport, precipitation and dynamic consistency superior to both direct interpolation of synoptic scale analysis of observations and on-site meteorology, and purely predictive-mode model results. This is a demonstration project to produce a one year meteorological data base on an hourly basis for a horizontal grid resolution of 80 km and for 15 vertical layers is underway.

As part of the development effort of a regional climate model (RCM) for the southern Great Basin, this paper presents a validation analysis of the climatology generated by a high-resolution RCM driven by observations. Two multiyear simulations were performed over the western United States with the RCM driven by European Centre for Medium-Range Weather Forecasts analyses of observations. This validation analysis is the first phase of a project to produce simulations of future climate scenarios over a region surrounding Yucca Mountain, Nevada, the only location currently being considered as a potential high-level nuclear-waste repository site. Model-produced surface air temperatures and precipitation were compared with observations from five southern Nevada stations located in the vicinity of Yucca Mountain. The seasonal cycles of temperature and precipitation were simulated well. Monthly and seasonal temperature biases were generally negative and largely explained by differences in elevation between the observing stations and the model topography. The model-simulated precipitation captured the extreme dryness of the Great Basin. Average yearly precipitation biases were mostly negative in the summer and positive in the winter. The number of simulated daily precipitation events for various precipitation intervals was within factors of 1.5-3.5 of observed. Overall, the model tended to overestimate the number of light precipitation events and underestimate the number of heavy precipitation events. At Yucca Mountain, simulated precipitation, soil moisture content, and water infiltration below the root zone (top 1 m) were maximized in the winter. Evaporation peaked in the spring after temperatures began to increase. The conclusion drawn from this validation analysis is that this high-resolution RCM simulates the regional surface climatology of the southern Great Basin reasonably well when driven by meteorological fields derived from observations. 26 refs., 9 figs., 4 tabs.

The report presents the results of an investigation to provide insight into the sensitivity of recharge estimates to historical and synthetic climate variability and projected climate change. In the study, the climatic variability in the 87-year Ellensburg historical record (1901-87) was first analyzed, and then ground-water recharge for the Ellensburg basin was simulated using the DPM model for the 22-year period (1956-77) and the 87-year period. Previously, data from three weather stations were used to interpolate daily temperature and precipitation to the cells of the basin (Bauer and Vaccaro, 1990). Because data from only one station were going to be used in the study, the previous results were compared with the results of the 22-year single-station simulation and results using a 22-year synthetic generated climate record. The object of comparing results of the one-station and three-station simulations was not to examine the effects of spatial climate variability. The comparison was done to determine the potential error in the estimated recharge caused by the loss of information on spatial climate variability when using climatological data from only one station. The results of 87-year simulations incorporating projected climatic change derived from three general circulation models were then compared with the 87-year results calculated using the historical record.

CHAMMP is an integral part of the ESD climate modeling program and its objectives are highly complementary to the modeling activities being conducted as part of the Program for Climate Model Diagnosis and Intercomparison. CHAMMP is also closely linked to the US Global Change Research Program, especially through its interactions with the major climate modeling centers of the other agencies. Because of its need for and focus on use of forefront supercomputers, CHAMMP is also a contributing program in the High Performance Computing Program, which is a new Presidential Initiative dedicated to the grand challenge'' computational problems. Just as the ARM program was formed in recognition of the inability of present general circulation models (GCMs) to satisfactorily represent cloud formation, convection, and radiative processes, the CHAMMP program was organized in response to the need to harness greater computational power in the pursuit of regionally resolved projections of climate change. The goal of the CHAMMP Climate Modeling Program is to develop, verify, and apply a new generation of climate models within a coordinated framework that incorporates the best available scientific and numerical approaches to represent physical, biogeochemical, and ecological processes, that fully utilizes the hardware and software capabilities of new computer architectures, that probes the limits of climate predictability, and finally that can be used to address the challenging problem of understanding the greenhouse climate issue through the ability of the models to simulate time-dependent climatic changes over extended times and with regional resolution.

CHAMMP is an integral part of the ESD climate modeling program and its objectives are highly complementary to the modeling activities being conducted as part of the Program for Climate Model Diagnosis and Intercomparison. CHAMMP is also closely linked to the US Global Change Research Program, especially through its interactions with the major climate modeling centers of the other agencies. Because of its need for and focus on use of forefront supercomputers, CHAMMP is also a contributing program in the High Performance Computing Program, which is a new Presidential Initiative dedicated to the ``grand challenge`` computational problems. Just as the ARM program was formed in recognition of the inability of present general circulation models (GCMs) to satisfactorily represent cloud formation, convection, and radiative processes, the CHAMMP program was organized in response to the need to harness greater computational power in the pursuit of regionally resolved projections of climate change. The goal of the CHAMMP Climate Modeling Program is to develop, verify, and apply a new generation of climate models within a coordinated framework that incorporates the best available scientific and numerical approaches to represent physical, biogeochemical, and ecological processes, that fully utilizes the hardware and software capabilities of new computer architectures, that probes the limits of climate predictability, and finally that can be used to address the challenging problem of understanding the greenhouse climate issue through the ability of the models to simulate time-dependent climatic changes over extended times and with regional resolution.