On predictive models of forage crops productivity by using weather variables: an application in the province of Ontario, Canada
MetadataShow full item record
The objective of this thesis is to establish predictive models for forage yield (productivity, tons per acre) using most relevant weather variables, such as precipitation and temperature from April to June. The outcome of this study is expected to be utilized to design indices and/or to set triggers for CAT bonds on forage crops in Ontario for the Government of Canada as it is exposed to tremendous agriculture risk exposures. We use forage crops data in Ontario, Canada, as an example. We propose to apply a single predictive model on a whole region, which is a vast area consisting of eight to ten counties which have a similar geographical environment. Seven models are tested for five regions with variables such as monthly rainfall, three months cumulative rainfall, average temperature, CDD (cooling degree days), etc. A new approach called weighted average temperature adjustment (WATA) is employed to deal with temperature data. The results demonstrate that the selected predictive model(s) consistently and considerably better explain the relationship between forage yield and weather variables for regions.