HR: 1340h
AN: GC33B-1254    [Abstracts]
TI: Statistical Significance of the Trends in the Extremes of Precipitation Over the US
AU: * Mahajan, S
EM: salilmahajan@tamu.edu
AF: Department of Atmospheric Sciences, Texas A&M University, College Station, TX 77843 United States
AU: North, G R
EM: g-north@tamu.edu
AF: Department of Atmospheric Sciences, Texas A&M University, College Station, TX 77843 United States
AU: Saravanan, R
EM: sarava@ariel.met.tamu.edu
AF: Department of Atmospheric Sciences, Texas A&M University, College Station, TX 77843 United States
AU: Genton, M G
EM: genton@stat.tamu.edu
AF: Department of Statistics, Texas A&M University, College Station, TX 77843 United States
AB: Extremes of precipitation have significant social and economic impacts. Recently, it has been postulated that the extremes of precipitation are on the increase over the past decades, and that this is due to anthropogenic climate change. Any systematic increase in the extremes of precipitation is a matter of great concern as extreme events can significantly disrupt human lives. Observational studies also suggest that there has been an increase in floods and droughts in the past decades, but it is difficult to assess their statistical significance as the real world provides just one realization of the stochastic behavior associated with precipitation variability. In this study, we attempt to test the statistical significance of increasing trends in the extremes of precipitation over the United States over the past few decades using the Monte Carlo technique. We use monthly precipitation data from various stations spread out across the US to look for a statistically significant trend. It is assumed that the distribution of monthly precipitation over a station can be approximated by the log-normal distribution. Multivariate analysis is used to generate synthetic data for the Monte Carlo tests, taking into account spatial correlations between different stations. Trends in precipitation extremes are computed for different realizations of the synthetic data, and this ensemble of trends is used to establish the 95% confidence interval for the trends. The observed trend is tested for statistical significance under these confidence intervals. A positive trend is observed but is found not to be statistically significant. We also apply the Monte Carlo test to precipitation data obtained from various coupled and uncoupled general circulation models for different climate change scenario integrations. No significant trends are found in most of the control integrations for the climate of the 20th century, thus establishing the robustness of our test. Climate change projections for the 21st century, however, display a significant positive trend, implying a role for anthropogenic forcing in the increasing extremes of precipitation in the future.
DE: 1616 Climate variability (1635, 3305, 3309, 4215, 4513)
DE: 1626 Global climate models (3337, 4928)
DE: 1817 Extreme events
DE: 1847 Modeling
DE: 1854 Precipitation (3354)
SC: Global Climate Change [GC]
MN: Fall Meeting 2005