Olson, O. G. University of Regina, olsono@uregina.ca
Leavitt, P. G. University of Regina, leavitt@uregina.ca
Chen, G. University of Regina, chen@uregina.ca
Chen, M. University of Regina, min.chen@uregina.ca
You, J. University of Regina, youj@uregina.ca
Laird, . Queen's University, lairdk@biology.queensu.ca

 
DROUGHT PREDICTION USING CONDITIONAL PROBABILITY ANALYSIS OF PALEOCLIMATE RECORDS
 
Fossil records of past climates can potentially be used for risk assessment of future droughts because environmental conditions that affect crops (high temperature, low precipitation) also alter lake chemistry and biotic composition. For example, fossil diatom inferences of past salinity (DI-Salinity) in Moon Lake are significantly correlated (r=-0.45, P<0.05) with areal crop losses in North Dakota (1921-1980). Consequently, paleoclimate data can be used to predict the occurrence of future droughts when high-resolution fossil records are modelled using conditional probability analyses. In Moon Lake, DI-salinity declined 52.5% since 0 AD, particularly since 1200 AD. This trend was removed by linear regression and droughts were defined as the years with positive DI-salinity greater than that of recent droughts (1987-1988). The number of years between drought events (inter-arrival times) were quantified, and mean (34+/-74 yr), median (7 yr) and minimum (1 yr) values calculated. Weibull models were fit to inter-arrival times and future drought probabilities estimated for the next 1-30 years. Analyses showed that the probability of severe drought by the end of the century was 7%, but that this likelihood increased to over 50% by 2025, assuming no future global warming.
 
Day: Friday, Feb. 5
Time: 11:45 - 12:00pm
Location: Hilton of Santa Fe
 
Code: SS11FR1145H