Empirical analysis on crop-weather relationships

Behzad Sharif, David Mankowski, Kurt Christian Kersebaum, Mirek Trnka, Kirsten Schelde, Jørgen Eivind Olsesen


There have been several studies, where process-based crop models are developed, used and compared in order to project crop production and corresponding model uncertainties under climate change. Despite many advances in this field, there are some correlations between climate variables and crop growth, such as pest and diseases, that is often absent in process-based models. Such relationships can be simulated using empirical models. 

In this study, several statistical techniques were applied on winter oilseed rape data collected in some European countries. The empirical models were then used to predict yield of winter oilseed rape in the field experiments during more than 20 years, up to 2013. Results suggest that newly developed regression techniques such as shrinkage methods work well both in yield projections and finding the influential climatic variables. Many of regression techniques agree in terms of yield prediction; however, choice of significant climate variables is rather sensitive to the choice of regression technique.


MACSUR members can access the full report (9 pp) in the file repository after login at http://macsur.eu

Previous issues and volumes can be found in the 'Archives' section.

You can refer to a paper published in this series in the following format Author (2013) Title. FACCE MACSUR Reports 2: D-C1.3, where "D-C1.3" is the article ID en lieu of page range.