Comparison of the three-temperature model and conventional models for estimating transpiration
Qiu G. Y,Miyamoto K,Sase S,Gao Y,Shi P,Yano T.
DOI: https://doi.org/10.6090/jarq.36.73
2002-01-01
Abstract:The three-temperature (3T) model is used for estimating transpiration from only temperature and net radiation. Comparison with data from a weighing lysimeter showed that the 3T model is accurate. The objectives of this study were to confirm the main advantages, possible field applications, and accuracy of the 3T model over conventional models through theoretical analysis and experimental verification. Four commonly used transpiration models were chosen for comparison: Peniman-Monteith (P-M), Bowen ratio, temperature difference, and ENWATBAL models. In a verification experiment conducted in a 1-ha sorghum field, microclimate, soil, and plant variables were extensively measured. The results showed that the 3T model has 4 main advantages. The major advantage is that it is theoretically sound, simple, and easily applicable, especially in developing countries. The next advantage is that aerodynamic resistance, surface resistance, and empirical parameters are not included. As a result, the transpiration process can be more easily revealed. The third advantage is that quantitative information on transpiration can be obtained with considerably fewer measurements, especially for application to remote sensing. The fourth advantage is that there is no fetch requirement. Because of these advantages, the 3T model could be applied both at small heterogeneous sites for local measurements and in large-scale fields for remote sensing. The transpiration estimated by the 3T model and the 4 conventional models was compared with lysimeter-measured data. The transpiration estimated by the 3T and P-M models agreed with the lysimeter-measured values. The mean absolute errors (MAE) between the measured value and the value obtained by the 3T model and between the measured value and that obtained by the P-M model were 0.45 and 0.42 mm d(-1), respectively. The MAE between the measured value and that obtained by the Bowen ratio model was 0.63 mm d(-1), between the measured value and that obtained by the temperature difference model, 0.69 mm d(-1), and between the measured value and that obtained by the ENWATBAL model, 0.88 mm d(-1). These results show that the MAE values of all the 5 models were < I mm d(-1), and the performance of the 3T model was as good as that of the conventional models.