An evaluation of irrigation-yield repsonse models for use with wheat & barley in Canterbury, New Zealand

Abstract

This study evaluated the ability of nine models to describe and predict the response of grain yield to flood irrigation of wheat and barley crops. The models may be divided into four groups: 1). Input-Output models 2). Potential Deficit models 3). Non-phasic Actual Deficit models 4). Phasic Actual Deficit models. Potential evapotranspiration, adjusted for crop cover, was calculated daily using the Penman formula. The timing of the different stages of development required for the phasic models was based on elapsed photothermal time from sowing. The input-output models were least successful in describing the response of yield to irrigation. There was little difference between the other models in the amount of variation explained. On average, both crop species produced about 300 kg (grain)/ha per irrigation or 8 kg (grain)/ha per mm of net water received. The potential and actual deficit models showed early sown crops to produce a response of 5-12 kg (grain)/ha per mm of irrigation applied when needed. The phasic models showed that the sensitivity of different developmental phases to drought varied erratically with sowing date. The predictive ability of the models was tested on four independent experiments. The potential deficit and non-phasic actual deficit models accurately predicted small responses to irrigation of less than 20% but were less accurate at predicting larger responses of up to 100%. These models adequately describe the response of wheat and barley yield to irrigation in Canterbury.