Global sensitivity analysis of yield output from the water productivity model

  • Authors:

  • Eline Vanuytrecht;Dirk Raes;Patrick Willems

  • Affiliations:

  • Division of Soil & Water Management, Department of Earth & Environmental Sciences, KU Leuven, Celestijnenlaan 200E, BE-3001 Heverlee, Belgium;Division of Soil & Water Management, Department of Earth & Environmental Sciences, KU Leuven, Celestijnenlaan 200E, BE-3001 Heverlee, Belgium;Hydraulics Laboratory, Department of Civil Engineering, KU Leuven, Kasteelpark 40, BE-3001 Heverlee, Belgium

  • Venue:
  • Environmental Modelling & Software
  • Year:
  • 2014

Quantified Score

Hi-index 0.00

Visualization

Abstract

This study includes a global sensitivity analysis of the water productivity model AquaCrop. The study rationale consisted in a comprehensive evaluation of the model and the formulation of guidelines for model simplification and efficient calibration. The global analysis comprehended a Morris screening followed by a variance-based Extended Fourier Amplitude Sensitivity Test (EFAST) under diverse environmental conditions for maize, winter wheat and rice. The analysis involved twenty-two different climate-crop-soil-meteorology combinations. The main objectives were to distinguish the model's influential and non-influential parameters, and to examine the yield output sensitivity. For the AquaCrop model, a number of non-influential parameters could be identified. Making these parameters fixed would be a step towards model simplification. Also, a list of influential parameters was identified. Despite the dependence of parameter ranking on environmental conditions, guiding principles for priority parameters were formulated for calibration in diverse conditions, valuable to model users. For this model that focuses on modelling yield response to water, parameters describing crop responses to water stress were not often among those showing highest sensitivity. Instead, particular root and soil parameters, relevant in the determination of water availability, were influential under various conditions and merit attention during calibration. The considerations made in this study about sensitivity analysis method (Morris vs. EFAST), prior parameter ranges, target functions and ranking variation according to environmental conditions can be extrapolated to other conditions and models, if done with the necessary precaution.