HR: 0800h
AN: GC21A-0165    [Abstracts]
TI: A METHOD FOR ESTIMATING THRESHOLD CROSSING TIMES WITH AN APPLICATION TO CLIMATE CHANGE
AU: * Kushnir, Y
EM: kushnir@ldeo.columbia.edu
AF: Lamont Doherty Earth Observatory / Columbia University, 61 Route 9W, Palisades, NY 10964, United States
AU: De la Pena, V H
EM: vhd1@columbia.edu
AF: Dept of Statistics Columbia University, Room 1027 SSW, 1255 Amsterdam Ave., New York, NY 10027-6900, United States
AU: Ravindranat, A
EM: aravindranat@fordham.edu
AB: Climate projections for the 21st century show that many subtropical regions, including the North American Southwest and the Mediterranean Basin countries are facing an imminent drying tendency. In the near-term, this drying is projected to proceed at a rate that does not depend strongly on the exact greenhouse gas emission scenario. It is important for decisions makers to assess the time when a given amount of drying will occur the associated limits of uncertainty. This can be done using a multi-model ensemble. Here we compare the more common method of finding such threshold crossing time ? which is based on following a multi-model average ? against an alternative, more accurate method that makes use of the special properties of estimators of the expectation. The method consists of calculating the threshold crossing time for all models (and ensemble members) participating in the assessment and looking at the various moments of the crossing-time distribution. We show that the alternative method is superior in its prediction capabilities, generates more accurate results, is very consistent, and gives more meaningful results including a timing error estimate.
DE: 1605 Abrupt/rapid climate change (4901, 8408)
DE: 1616 Climate variability (1635, 3305, 3309, 4215, 4513)
DE: 1626 Global climate models (3337, 4928)
DE: 1637 Regional climate change
DE: 1655 Water cycles (1836)
SC: Global Environmental Change [GC]
MN: 2007 Fall Meeting