Using crop models with seasonal prediction: a new study
EUPORIAS research has led to the publication of a new scientific paper looking at how crop models might be used with seasonal weather forecasts to provide seasonal yield outlooks.
The JULES-crop model is built around the Joint UK Land Environment Simulator (JULES - https://jules.jchmr.org/). The paper first investigates what drives year to year variations in maize yields produced by the model at the global scale, to help understanding the impact of different ways of driving the model with weather data. JULES typically needs a combination of weather variables (precipitation, radiation, temperature, pressure, specific humidity and wind) to be provided, ideally with six or three hourly datasets. The new study finds that compared to using the "best" input data and initial conditions, the model performs reasonably well given fewer variables and using a disaggregation tool to calculate three hourly weather data from daily averages. This is important since only daily weather data are normally available from seasonal forecasts. The study also looked at the effect of initialising the model with long-term average weather data (rather than actual data for the year in question), on modelled yields. This method could simplify the use of the model with seasonal forecasts, since obtaining the appropriate data needed required by JULES for the start date of the seasonal forecast would present significant practical challenges.
Citation: Williams, K. E. and Falloon, P. D.: Sources of interannual yield variability in JULES-crop and implications for forcing with seasonal weather forecasts, Geosci. Model Dev. Discuss., 8, 4599-4621, doi:10.5194/gmdd-8-4599-2015, 2015.