Probabilistic forecasting of crop yield across Canada under environmental uncertainty
Citation
Newlands, N.K. (2017) Probabilistic forecasting of crop yield across Canada under environmental uncertainty. 37th International Symposium on Remote Sensing of Environment (ISRSE 37): Earth Observation for Development and Adaption to a Changing World. May 8-12, 2017. Tshwane (Pretoria), South Africa.
Plain language summary
There is increasing concern over the destructive impacts of climate change, extreme weather events and crop disease on global food security. Reliable crop forecasts have the potential to address this challenge by aiding agricultural stakeholders a way to identify potential risks and benefits, with sufficient lead-time to adapt, adjust and improve their crop production plans. This is especially needed during times where crop production is at risk or decisions are more uncertain. This paper provides an overview of the Integrated Canadian Crop Yield Forecaster (ICCYF) - a probabilistic approach that uses machine-learning to predict regional-scale yield distributions, integrating satellite remote-sensing, meteorological, and survey data. The paper also discusses current inter-linked research work that is seeking to improve our ability to forecast changes and losses in yield across complex agricultural landscapes and the multi-scale effects of slower, interannual shifts (e.g., ENSO teleconnection forcing) and faster, sudden shocks (extreme weather events and disease outbreaks).
Abstract
There is increasing concern over the destructive impacts of climate change, extreme weather events and crop disease on global food security. Reliable crop forecasts have the potential to address this challenge by aiding agricultural stakeholders a way to identify potential risks and benefits, with sufficient lead-time to adapt, adjust and improve their crop production plans. This is especially needed during times where crop production is at risk or decisions are more uncertain. This paper provides an overview of the Integrated Canadian Crop Yield Forecaster (ICCYF) - a probabilistic approach that uses machine-learning to predict regional-scale yield distributions, integrating satellite remote-sensing, meteorological, and survey data. The paper also discusses current inter-linked research work that is seeking to improve our ability to forecast changes and losses in yield across complex agricultural landscapes and the multi-scale effects of slower, interannual shifts (e.g., ENSO teleconnection forcing) and faster, sudden shocks (extreme weather events and disease outbreaks).