The effect of transpiration uncertainty on root zone soil water by Bayesian analysis

It is very important to identify the uncertainties of transpiration and root zone soil water for improving water, fertilizer and agricultural chemical management. The canopy transpiration, dominating over evapotranspiration in a close planting orchard, is the main input data of soil water movement simulation for the model based on the Richard equation. However, transpiration estimations are always uncertain because of uncertainties of input data and model structure, which bring the uncertainties of root zone soil water simulation, reduce the simulation accuracy and cause model instability. So, quantitative study of uncertainties of transpiration and root zone soil water simulations are very important to raise the reliability of the simulation. In this study, a Bayesian approach was used to fit the transpiration model to half-hourly cherry transpiration rates, probabilistically estimate its parameters and predict the uncertainties. The probabilistic transpiration model was extended by adding a normally distributed error term, and the Markov chain Monte Carlo simulation method was used to determine the posterior parameter distributions, and the estimated transpirations of average, 95% upper and lower confidence limits were obtained as the input data of the Richard equation, and the simulations of soil water and root water uptake were then obtained. The results showed there were a large number of uncertainties for the transpiration, soil water and root water uptake, and soil water greatly changed in the 30 and 50 cm soil layer for the intensive root system, but there was only minor change in 10 and 110 cm soil layers for small roots. The mean relative error (MRE) was 4.3%, 8.64%, 14.53% between simulated and measured soil water for average, 95% upper and lower confidence limit estimated transpiration as input data, and it was 11.68%, 19.55%, 25.35% for root water uptake, respectively. Moreover, there were also very large uncertainties for cumulative root water uptake, and the maximum difference can reach about 100 mm after 100 simulation days. So, the effect of transpiration uncertainties on soil water and root water uptake should be paid attention to for improving water management and contamination risk assessment.