Variation in reference crop evapotranspiration caused by the Ångström–Prescott coefficient: Locally calibrated versus the FAO recommended

Accurate estimation of the reference crop evapotranspiration ( ET 0 ) is investigated due to its critical role in affecting calculation of crop water use and efficiency in agricultural ecosystems. The main emphasis in this paper is to clarify the possible uncertainty in the estimation of ET 0 associated with using un-calibrated Angstrom–Prescott ( A – P ) coefficients. We first calibrated the coefficients using long-term data records from 34 sites in the Yellow River basin in China, and then applied these coefficients to estimate short wave irradiance ( R s ) and ET 0 at 16 sites to evaluate the difference in ET 0 between the FAO recommended and the locally calibrated. We found that the direct use of the FAO recommended coefficients significantly affected the estimation of ET 0 at most sites, which differed from −3% to 15% at daily scale and from −4% to 16% at monthly scale from the locally calibrated ones. These differences are comparable with or larger than those caused by some alternatives of the FAO recommended algorithms for net irradiance or vapor pressure, which further highlights the importance of using the locally calibrated coefficients. The degree of difference in ET 0 showed a significant threshold relation with altitude and longitude in such a way that relatively small impact lies around 2233 m and 98°E, and away from these, the effect begins to increase. Given the large overestimation in water use as a consequence of the significant overestimation in ET 0 associated with the direct use of the FAO coefficients, especially in those high yield production areas with altitude A – P coefficient a , b , ( a + b ) and other easily obtainable factors (altitude, longitude and air temperature). A three-step procedure was recommended in applying these relations, which was (1) determine if calibration is needed or not for a given location; (2) estimate one of the A – P coefficients, either a or b if calibration is needed; (3) estimate the remaining coefficient using relations of ( a + b ) due to its higher coefficient of determination. In summary, we have revealed the errors and areas that are most affected when using the un-calibrated coefficients, and discussed the consequence of such error on agricultural production, and proposed practical solutions to avoid large errors. These results are intended to make the research community aware of such errors so that more appropriate choice of these coefficients is made. We hope that similar assessment will be done in other climates, contributing to managing water resources efficiently in water basins.