3.3 Comparison of daily ET between reference evapotranspiration
models and lysimeter measurements during the growing season
During the growing season, the daily ET0 calculated by
13 reference evapotranspiration equations was significantly correlated
with the lysimeter measurements (P <0.01), withR 2 ranging from 0.32 to 0.60 (Fig. 4 and Table
3). Of the combination models, FAO-24 Pen obtained the highestR 2, followed by Pen-63 and FAO-56 PM. Of the
radiation-based models, PT and DK obtained the highestR 2, followed by Makkink, Makkink (1967),
Makkink (1957), IRMAK1 and IRMAK2. It should be noted that the Makkink,
Makkink (1967) and Makkink (1957) have the sameR 2. Of the temperature-based models, HAR and
HAR2 obtained the highest R 2, followed by HAR1.
Interestingly, all models (except for HAR1) generally underestimated ET
during the growing season: values of MBE ranged from −0.88 to -0.03 mm
d−1 and averaged −0.63 mm d−1 over
all 13 models, with FAO-56 PM having the largest underestimate (by
39.75%) and HAR2 the minimum underestimate (by 0.68%) (Table 3).
The RMSE for combination models ranged from 1.31 to 1.67 mm
d−1 and averaged 1.48 mm d−1, the
RMSE for radiation-based models ranged from 1.19 to 1.56 mm
d−1 and averaged 1.40 mm d−1, and
the RMSE for temperature-based models ranged from 1.38 to 1.41 mm
d−1 and averaged 1.40 mm d−1 (Table
3). Based on the RMSE values, the performances of the 13 reference
evapotranspiration models follow the order: DK > PT
> Makkink (1967) > IRMAK2 >
Pen-63 > HAR2 > HAR > HAR1
> Makkink > FAO-24 Pen > IRMAK1
> FAO-56. Evidently, the best (DK) was 40% more accurate
than the poorest (FAO-56). Overall, the radiation-based models
demonstrated the best performance during the growing season, followed by
temperature-based models and finally combination models.