Abstract
Here we assess the quality and in-season development of European wheat (Triticum spp.) yield forecasts during low, medium, and high-yielding years. 440 forecasts were evaluated for 75 wheat forecast years from 1993–2013 for 25 European Union (EU) Member States. By July, years with median yields were accurately forecast with errors below ~2%. Yield forecasts in years with low yields were overestimated by ~10%, while yield forecasts in high-yielding years were underestimated by ~8%. Four-fifths of the lowest yields had a drought or hot driver, a third a wet driver, while a quarter had both. Forecast accuracy of high-yielding years improved gradually during the season, and drought-driven yield reductions were anticipated with lead times of ~2 months. Single, contrasting successive in-season, as well as spatially distant dry and wet extreme synoptic weather systems affected multiple-countries in 2003, ’06, ’07, ’11 and 12’, leading to wheat losses up to 8.1 Mt (>40% of total EU loss). In these years, June forecasts (~ 1-month lead-time) underestimated these impacts by 10.4 to 78.4%. To cope with increasingly unprecedented impacts, near-real-time information fusion needs to underpin operational crop yield forecasting to benefit from improved crop modelling, more detailed and frequent earth observations, and faster computation.
Highlights
To have adequate, timely, and coherent information on expected crop production levels, the Joint Research Centre (JRC) of the European Commission (EC) forecasts crop yields and production across all European Union (EU) Member States (MS)[1,2,3]
Crop production forecasts are obtained by multiplying the crop yield forecast by the sown area, information provided by each MS to Eurostat, the statistical office of the EU
Statistical analysis is used to link past yield variability with predictors derived from interpolated meteorological observations, derived gridded agro-meteorological indicators, gridded crop model simulations2 – fed with interpolated observed meteorological data, which at the time of the yield forecast are extended by the 10-day forecast from the European Centre for Medium-range Weather Forecasting (ECMWF www.ecmwf.int), as well as remotely sensed observations
Summary
Coherent information on expected crop production levels, the Joint Research Centre (JRC) of the European Commission (EC) forecasts crop yields and production across all European Union (EU) Member States (MS)[1,2,3]. Several governmental and private sector institutes make crop production and commodity supply estimates at the national level[9]. These estimates rely on a variety or combination of qualitative and quantitative techniques for monitoring and forecasting, including field surveys, farmer inquiries, expert analysis, crop models, earth observation, or statistically based methods[10]. At the JRC, the MARS Crop Yield Forecasting System (MCYFS) facilitates monitoring of current crop conditions and forecasting of expected crop yield at harvest including country-level analyses. Statistical forecasting methods available to the analysts include trend analysis, regression analysis[20] and similarity analysis based on principal component analysis (PCA) and cluster analysis[21] (see Methods)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
