Seasonal assessment of the grass reference evapotranspiration estimation from limited inputs using different calibrating time windows and lysimeter benchmarks

Abstract

[eng] Models relying on limited inputs are very valuable for estimating reference evapotranspiration, and subsequently irrigation doses, but their accuracy can be very dependent from calibration. This study assessed three versions of the Hargreaves-Samani (HS) and the FAO Penman-Monteith (PM) equations to estimate reference evapotranspiration (ETo), relying respectively on three input combinations. Further the six models were adjusted each using different time windows for calculating the calibrating constants, namely global, annual, monthly, fortnightly, and weekly constants, while all the models were calibrated and tested using calculated and lysimeter benchmarks. The models relying on mean air temperature and solar radiation tended to be more accurate than those relying on mean air temperature and relative humidity, while these tended to be more accurate than those relying on air temperature difference, but there might be intra annual exceptions according to the monthly indicators. The errors of the PM estimations were just slightly higher than those of the corresponding HS estimations. The accuracy improvement in the calibrated versions was higher the shorter the time window used for averaging the calibrating parameters. Thus, the application of monthly or, at least, seasonal calibrating constant might be recommended for a suitable correction of the bias. During the year, the estimations presented markedly lower errors and lower differences within models during the summer. The error decrease in the calibrated versions was more marked during the winter. The assessment relying on lysimeter benchmarks provided similar qualitative patterns than the assessment relying on calculated benchmarks, but the corresponding error ranges were higher. Finally, 6 examples were presented for visualizing the effect of the method used to estimate ETo on the corresponding resulting average annual crop water requirements. If irrigation scheduling is based on a soil water balance using crop evapotranspiration estimates, at least, a monthly bias assessment of the ETo estimates in combination with the crop cycle lengths and dates might contribute to infer if crop water requirement infra-estimation trends are identified during crop sensitive stages to water deficit.