How is crop growth model calibration performed? Results of a survey.

Abstract

Crop growth model calibration, or parameter estimation, is a demanding and critical step of a crop modeling project: Projections from a model are determined by the parameter values used in the model, and parameter uncertainty can play a major role in projection uncertainty. Despite its importance little attention has been paid to the calibration approaches and methods used. An open web-based online survey with 39 questions about crop model calibration was conducted in autumn 2016 aiming to record the current practices in crop model calibration across the crop modeling community.The sections of the survey related to the data used for calibration, the parameters calibrated, the calibration method and the software used, as well as estimation of the parameter uncertainty. In addition, there were questions providing background information about the model user and model, time required for calibration and challenges faced. Overall, 211 survey submissions were analyzed to examine the common practices in crop model calibration. The respondents covered both more and less experienced modellers with a wide range of models and calibration approaches used.This talk will present the results of the survey, and what they imply concerning the major choices faced during calibration. An important question asked respondents what they thought was the major challenge of crop model calibration. Only two (out of 211) respondents answered that they saw no major problems. This emphasizes the need for progress in calibration and setting out guidelines of good practices.