Estimation of diurnal-provided potential evapotranspiration using GNSS and meteorological products

Abstract

In the context of global warming, extreme drought has seriously endangered the safety of human lives and assets, indicating an urgent need for its accurate monitoring and early detection. Since various types of drought monitoring indices are calculated from the variable of potential evapotranspiration (PET), thus the selection of models to derive PET has attracted broad attention in recent decades. Generally, the common-used Thornthwaite (TH) model possesses the features of calculation simplicity and great universality, however, its accuracy and temporal resolution (monthly) are relatively low, which greatly hinders the further uses of this model. Considering these restrictions, a novel method for estimating diurnal-provided PET values (DTH-PET) using the TH model was first proposed, in which a 31-day sliding window was adopted to improve its temporal resolution. The new method takes both the diurnal and “monthly” variation features into account. Verification results showed that, compared with the monthly TH model, the improvements in the mean root mean square (RMS) errors made by the new method was 3.1 mm. In addition, to further improve the accuracy of the DTH-PETs, a new high-order calibration function with the use of GNSS and meteorological measurements including ZTD, temperature and pressure was then constructed and fully validated. In comparison with the 29.95 mm mean RMS from the conventional TH model without calibration, the new method led to a 4.37 mm result, i.e., a 25.57 mm improvement (over 85%) was made after a thorough calibration process. In addition, the comparison of the newly-proposed method with the common-used first-order equation also further illustrates its superiority during the whole calibration process. Therefore, all these results suggest that, by using the new strategies proposed in this study, DTH-PET can be easily estimated with high accuracy using limited meteorological variables.