Document: PET-3-13-5

Droughts, lakes and crop insurance: Ecological risk assessment using mud and math.

LEAVITT, P.R.* ¹, J.A.RUSAK ¹, O.G.OLSON ¹, S.WUNSAM ¹, G.CHEN ¹, M.CHEN ¹, J.YOU ¹, L.ZHANG ¹ and B.CUMMING ²

University of Regina ¹
Queen's University ²

Abstract:
In 1988 drought-related agricultural losses exceeded $50 billion in the USA, yet crop insurance practices are based on only short-term (<40 y) risk assessments. We used conditional probability analyses of fossil-reconstructed lake salinity to quantify drought and flood characteristics across the northern Great Plains. Canonical analysis of fossil, climatic, agricultural and insurance data (1950-1980) demonstrated that historic wheat production was significantly correlated to changes in sedimentary diatom species assemblages (r = 0.45; P < 0.05). Risk assessments based on Weibull models of sub-decadal time series (AD 1000-1997) from 3 sites showed that the probability of a catastrophic drought was only 6% by 2000, but rose rapidly to 52% by 2030. Flood probabilities were higher. Spectral analyses of censored time series showed that all extreme events exhibited significant periodicities, but that cycle period differed between drought (62-82 y) and flood events (11 y). Impacts of global warming were evident as slightly increased drought frequency, but not intensity, during the 20th century, although several decades will be required before anthropogenic influences emerge statistically from background climatic variation.

Keywords: Drought, risk assessment, agriculture, insurance

This abstract is being presented at: 11:45 AM in session:
Oral Session #54: Lake Ecology.