Towards Attribution of Hurricane Activity Changes
We explore recent progress towards attributing observed tropical storm and hurricane activity changes in the Atlantic basin. We outline a strategy for regional climate change attribution that is relatively general, and discuss aspects of the hurricane attribution problem that make it particularly difficult to make strong attribution statements about, and recent efforts to do so. The general approach is a two part attribution, wherein an ultimate causal agent is connected to a proximate causal agent, and that proximate causal agent is connected to changes in hurricane activity. In order to attribute changes in hurricane activity, one must: i) define a measure of activity, ii) make efforts to correct historical estimates of this change in activity for data problems, iii) develop comprehensive dynamical models and theoretical understanding to connect changes in activity to large-scale environmental parameters. This methodology is applied to annual tropical storm (TS) counts in the Atlantic. In the observed record there is an unambiguous increase in TS counts over the past 25 years, but accounting for 'missed' storms the increase since the late-19th Century is not significant since late-19th. Using comprehensive dynamical models, it is found that the recent increase in Atlantic TS and hurricanes can be forced by observed sea surface temperature (SST) changes. In these models, it is found that the global-mean and tropical-mean SST changes were not the primary driver of changes in TS and hurricane frequency, and that the changes were driven by the pattern of SST change. Efforts are made to identify the most influential patterns of SST change, and connect these relevant patterns to forced and internal climate variations.
Seasonal Simulation of Atlantic Hurricanes using the FSU/COAPS Climate Model
We use the FSU/COAPS climate model to study cyclogenesis, evolution, intra- to interannual variability of tropical cyclone activity in the Atlantic Basin and possible changes in activity due to climate change. The first phase of this project involves the simulation of recent year climatic conditions using observed SSTs. The second phase, currently underway, will simulate future climate scenarios using SSTs from IPCC climate model projections. The climate model is run at moderately high resolution (T126 or approximately 1 degree) and with 27 vertical levels. The model generates hurricane-like vortices that bear remarkable resemblance to their real-world counterparts: a warm core with cyclonic inflow at lower levels, anticyclonic outflow aloft, deep central pressures and intense precipitation. The genesis location and tracks are consistent with climatological observations. While the model produces fewer very intense storms (Category 4-5 on the Saffir-Simpson scale) than observed, the total number of storms and interannual variability is very reasonable when compared to the historical record. In order assess the ability to simulate tropical cyclone activity on seasonal timescales, we run an ensemble of 4 6-month simulations for 20 years (1986-2005) using prescribed weekly sea surface temperatures. The simulation time period covers the North Atlantic hurricane season from June 1 to November 30. Each ensemble member was initialized with a different atmospheric initial condition (near June 1 of the respective year) from ECMWF analyses. We find that for the Atlantic Basin, the correlation between the simulated and observed number of storms was approximately 0.78. The seasonality of the storms was similar to the observed, with most storms occurring in August and September, though not as sharply peaked. We further found that there can be large sensitivity to cyclogenesis using different model parameterizations. Pending completion of the second phase of the project, and given the encouraging results from the first phase of simulating the current climate, we hope to be able to determine possible changes in hurricane activity in future climate conditions.
Impact of Climate Change on Salinity of the Upper Delaware Bay
We analyze a time series of salinity in the upper Delaware Bay from 1963 to 2007 with the aim of quantifying the impact of streamflow and sea level. 80% of the variability in seasonally averaged salinity can be explained with a simple, four-parameter function of streamflow. After removing the streamflow influence we find that salinity has increased 1.08 ± 0.47 over this time period, 22 ± 10% of the long-term mean of 4.8. This increase is greater than expected from sea-level rise, based on relationships between salinity and sea level derived from numerical models and short-term observations. Long-term changes in the bathymetry of the Bay may explain the discrepancy. The projected increase in salinity due to sea-level rise by 2100 is 3.3 ± 3.1 (69 ± 65% of the mean), similar in magnitude to projected streamflow impacts.
Probabilistic Downscaling of 21st Century Daily Precipitation Occurrence and Intensity in the United States
Previous studies have reported positive trends in total annual precipitation when averaged over the contiguous United States. However, regional manifestations of precipitation changes are less clear. For example, it is unclear whether precipitation changes have resulted from increases in the number of precipitation events or in the intensity of precipitation during those events, or some combination of changes in occurrence and intensity. We present a novel approach for investigating changes in precipitation occurrence and intensity using established statistical models. Precipitation occurrence is modeled using a two-state first-order Markov chain model and, for the wet days, precipitation intensities are drawn from the two-parameter gamma distribution. Parameters for both models are estimated using historical station data. Regression equations are then developed to relate the parameter estimates to variations in larger-scale climate variables using the European Center for Medium-Range Weather Forecasting (ECMWF). When applied to 20th and 21st century AOGCM output from the IPCC AR4 archive, the transfer functions facilitate examination of the evolution of the precipitation processes due to anthropogenic climate change. The downscaling method also allows stochastic generation of multiple precipitation time series for the same climate change scenario, which can be used to derive other 21st century precipitation descriptors and assess effects of sampling variability.
Capacity and Need for Providing Climate Change Scenarios for Small Islands
The emphasis on GCM outputs in the IPCC's Fourth Assessment Report (AR4) limits what can be confidently said for small island states such as those of the Caribbean. For finer resolutions, regional models with resolutions of 25 to 50 km are required, or statistical downscaling techniques, such as those using GCM outputs as predictors in regression equations derived from station data and atmospheric parameters. Such modeling results are, however, far less available and too piecemeal. Jamaica, Antigua and Dominica's second national communications to UNFCCC provide a good opportunity for a case study of current ability to provide climate change scenarios using these techniques for small islands. Results from GCMs, a regional model (PRECIS) and statistical downscaling are compared for rainfall and temperature. As a result, some recommendations for modeling needs of small islands are suggested. These include: (i) obtaining optimum results from regional models based on an intercomparison of regional model results over the islands of the Caribbean, Indian and Pacific Oceans, similar to the multi-model data set used in AR4 (ii) averaging outputs of several GCM's for input predictors into downscaling as opposed to using that obtained from a single GCM, (iii) exploring downscaling techniques which reduce data requirement limitations, and (iv) devising/adopting statistical downscaling techniques for sea-level rise. A glaring overarching need, however, is for the development of regional science centres of capacity as a means of fast tracking such efforts.
A multi-member, high-resolution, transient simulation of 20th and 21st century climate in the United States
We are currently conducting a suite of high-resolution climate model simulations for the United States. This nested experiment covers the full continental United States at 25 km horizontal resolution, with five GCM- generated ensemble members covering 1951-2100 in the A1B emissions scenario.To our knowledge, this is the first century-scale, high-resolution, multi-member climate change projection for any continental area. Our high-resolution climate model experiment uses RegCM, a hydrostatic, sigma coordinate, primitive equation nested climate model. Large-scale boundary conditions are provided to RegCM by the NCAR Community Climate System Model (CCSM3), a fully coupled atmosphere-ocean general circulation model (GCM). I will show results from this unique experiment, including the time-evolution temperature and precipitation extremes, as well as the dynamics shaping that evolution. For instance, the simulation shows that by the 2030-2039 period, extreme seasonal heat stress is common, with changes in both warm-season large-scale circulation and surface moisture balance contributing to the magnitude of the change.