07121-DW1801

Goal
The goal of this project is to evaluate the impacts of climate variability on North Atlantic hurricane activity through a nested climate modeling approach that combines existing global climate simulations, with more detailed regional simulations of individual storms and hurricanes, and advanced statistical techniques. Project results will be used to assess likely changes in hurricane activity under a global warming scenario and to project likely changes in hurricane intensity and frequency in the Gulf of Mexico over the next fifty years.

Performers
National Center for Atmospheric Research (NCAR), Boulder, CO 80305
Georgia Institute of Technology (GA Tech), Atlanta, GA 30332

Background
Since the disastrous 2004 and 2005 hurricane seasons, there has been considerable debate on whether climate change is affecting the frequency and intensity of hurricanes in the Gulf of Mexico. This project aims to provide some realistic climate simulation results that may help to resolve this debate.

Climate simulations are being carried out for the North Atlantic, with a focus on the Gulf of Mexico. The large-scale climate component is provided by existing global climate simulations from the NCAR Community Climate System Model (CCSM) archive of simulations. This is one of the leading global climate models in existence today and has undergone careful review and testing by the scientific community. In addition, it has been extensively documented and provides one of the best available future projections of ocean and atmosphere changes in the North Atlantic under greenhouse warming scenarios. These global-scale simulations are too coarse in their spatial resolution to model hurricane activity, so the research team is also considering multiple and complementary downscaling methods for the entire North Atlantic basin with focus on the Gulf of Mexico. Dynamical downscaling will be conducted using the Nested Regional Climate Model (NRCM) by nesting the regional model into the global model. Statistical downscaling will combine existing methods and new developments and a combined statistical-dynamical approach to downscaling is also being explored. The project team will thus be conducting a series of downscaling simulations for current and future climate and hurricane activity in the Gulf of Mexico using a variety of techniques. All results will be archived by the NCAR for use by the industry and other researchers.

The project will be led by NCAR’s Mesoscale and Microscale Meteorology Division in partnership with their Climate and Global Dynamics Division and the Georgia Institute of Technology. NCAR is a Federally-Funded Research and Development Center administered by the National Science Foundation. NCAR’s new Blue Fire supercomputer and the DOE Argonne National Laboratory supercomputing system are being used for the simulations.

Deliverables for this project will include a final report detailing results and analyses for each technical task, as well as an archive of all project data and simulations, to be made available to the research community through an NCAR database. These simulations are likely to serve as the basis for more sophisticated modeling in the future, as the global and regional climate knowledge base grows and supercomputing capabilities become more advanced.

Potential Impacts
This project will result in combined global climate and regional weather models that will allow global climate projections as well as detailed snapshots of potential hurricane activity in the Gulf of Mexico. Advanced statistical methods will also be used to supplement this dynamical downscaling. This should result in a better understanding of the potential impacts of climate variability and change on hurricane activity in the North Atlantic and the Gulf of Mexico. These projections will document any expected change in hurricane frequency and/or intensity with a greater degree of accuracy than previously possible, thus helping to inform decision-making by energy companies active in the Gulf of Mexico.

Accomplishments

Tracking Tropical Cyclones
An automated tropical cyclone tracking algorithm is necessary to detect tropical cyclones in the NRCM simulations. This algorithm has been developed to detect all tropical cyclones but importantly also to exclude any weather systems that do not quite fit the definition of a tropical cyclone. The resulting cyclone database has been used to complete many of the deliverables. Many of the outputs for Task 1 have been completed and plots of tropical cyclone parameters from the existing NRCM simulations for the period 2000-2005 are available. We note that this channel configuration of the Nested Regional Climate Model driven by a global analysis dataset produces a reasonable temporal and spatial distribution of tropical cyclones globally but has a tendency to overproduce the total number of tropical cyclones.

Revising the Global Model Climate
A technique has been developed to remove a high shear bias seen in the original CCSM climate. The technique involves replacing the mean climate base-state in the global model (CCSM) with the mean climate base-state from analysis data (i.e. our best estimate of the observed base-state) while retaining day-to-day weather, climate variability and climate change signals from the global model. Comparison of the shear between this new simulation and the original simulation show that the shear is much reduced and is now within the range typically observed. This simulation shows a sensible spatial distribution and frequency of tropical cyclone genesis and storm tracks.

Statistical Downscaling
Statistical downscaling of the revised global model datasets using a Genesis Potential Index has been performed to extract tropical cyclone climatologies from future climates for Task 2. Extensive testing on reanalysis data has helped to identify the valid scales and regions for which the Genesis Potential Index gives a reasonable estimate of tropical cyclone frequency. This statistical downscaling has since been applied to the full ensemble of future climates generated by CCSM to determine tropical cyclone trends and variability out to 2100. Work is currently underway to formulate a new statistical approach suitable over a full range of spatial scales with particular focus on the Gulf of Mexico.

Dynamical Downscaling
Owing to the high computational cost of running the NRCM, a time-slice approach was adopted whereby the three decades of 1995-2005, 2020-2030 and 2045-2055 were simulated to explore tropical cyclones over the next 50 years. The three time slices have been completed on a 36 km gridded domain and extensive analysis has shown the NRCM is able to capture a reasonable number and spatial and temporal distribution of tropical cyclones in the Atlantic basin and the Gulf of Mexico in current and future climate. Considering the entire North Atlantic basin, the model produces an increase in storm frequency and intensity over the next 50 years and also a shift in location to lower latitudes.

Application of extreme value theory has also been shown to be successful in aiding the downscaling of the more extreme hurricanes. A revised approach to doing this is expected to provide further detail on severe storms beyond that possible even with the highest resolution climate simulations.

Assessing Damage
An assessment of the impact of climate variability and change on damage to off-shore facilities in the Gulf has been conducted by developing a damage index known as The Willis Hurricane Index. This index has been applied to the storm data over the three decades of 1995-2005, 2020-2030 and 2045-2055 to examine trends in expected damage.

Changes in the Ocean
An analysis of ocean heat content in the global climate model data revealed a sharp upward trend for the Gulf of Mexico. This trend is evident from the late 20th century and continues to the end of the dataset in the year 2100. An observational dataset for the past few decades corroborates this upward trend evident in the model data. Work is planned to fully connect the NRCM atmospheric model to ocean heat content so the atmosphere can fully realize this reservoir of energy available for tropical cyclone development.

Communication of Results
A comprehensive workshop was conducted at UCAR on March 8 and 9th 2010 with all presentations posted to the RPSEA website.

A summary paper has been prepared for the 2010 Offshore Technology Conference Metocean special session.

A follow-up focused workshop between industry and scientists is planned for later in 2010.

Current Status
The key tasks detailed in the Project Management plan and Technology Status Assessment are mostly complete, as detailed below. Additional tasks that are natural extensions to the original plan have been proposed and are currently in progress under a no-cost project extension. These are also detailed below.

Model Set-Up and Initial Testing.
First, the research team analyzed existing NRCM simulations of North American tropical cyclones and tested existing downscaling approaches for improving their spatial resolution. The research team analyzed NRCM simulations for the 1995-2005 period and plotted the Climate Prediction Index (CPI), maximum wind speed, radius to maximum winds, forward translation speed, translation heading, and occurrence of Category 1-5 storms over the 10-year simulation period. Maps display storm tracks, starting positions and positions of storm upgrades to tropical storm and hurricane force, mean monthly wind shear, and water temperatures in the Gulf of Mexico.

Intermediate Testing and Analysis.
In this task, the research team conducted simulations of 50-year duration, using archived data, and analyzed multi-year trends in wind shear and mixed-layer ocean temperatures for the North Atlantic hurricane latitudes. For global-scale simulations, testing with the CCSM covered the historical period from 1957-20007. For the high-resolution simulations, testing was performed with the NRCM for the 10-year period from 1995-2005.

50-Year Forecast.
The aim of this task is to test the model’s ability to forecast future climate, by projecting the simulations to 2055. The research team completed three 11-year simulation periods, covering 1995-2005 for the current climate, and 2020-2030 and 2045-2055 for future climate. Climate parameters have been analyzed using a spatial resolution domain of 36 km. It is possible that the impact of climate variability and change on tropical cyclones may be sensitive to the resolution of the NRCM. To test this sensitivity, the three time-slices of 1995-2005, 2020-2030 and 2045-2055 are currently being run again, this time at a higher resolution using 12 km grid spacing. These simulations are running using time allocated on a machine at Argonne National Lab.

25-Km Climate Global Simulation.
This task was dependent on an external group achieving a stable global model simulation on a 25 km grid. This has not yet been possible. As a result this task has been put on hold.

Additional Tasks

Uncertainty
A thorough assessment of uncertainty in our methods is underway to determine confidence in our results. Investigations are underway to distinguish the reproducible climate change signal in storm activity from the noise due to uncertainty.

Statistical-Dynamical Downscaling
The revised global model data for one of the CCSM simulations has been downscaled using Kerry Emanuel’s statistical-dynamical hurricane model. Future work will focus on understanding the differences between all our downscaling methods.

Exploring the Most Intense Storms
The known limitation of model resolution on hurricane intensity has been addressed by the application of Extreme Value Theory to our modeled distributions of storm intensity. This statistical technique provides estimates of the frequencies of high intensity storms and the response to climate variability and change.

Project data, simulation results, and reports will be made available to the research community through an NCAR data portal.

Project Start: February 23, 2009
Project End: December 31, 2010

DOE Contribution: $544,085
Performer Contribution: $140,000

Contact Information:
RPSEA – Art Schroeder (aschroeder@rpsea.org or 713-372-2817)
NETL - Jay Jikich (Sinisha.Jikich@netl.doe.gov or 304-285-4320)
NCAR – Greg Holland (gholland@ucar.edu or 303 497-8949)

Additional Information

Final Project Report [PDF-1.34MB] February, 2011

Model Investigations of the Effects of Climate Variability and Change on Future Gulf of Mexico Tropical Cyclone Activity [PDF] OTC May, 2010