Evaluation and improvement of residence time distribution analysis methods

Although residence time distributions (RTDs) are frequently used in wetland studies, there is no consensus about the analysis method. Generally, two types of methods can be used to analyze RTD data: the method of moments and the RTD modeling method. RTD curves are obtained by tracer tests. The real RTD curves are usually imperfect due to the presence of random measurement errors, the unreasonable sampling intervals, and the premature termination of the tracer test. This study aims to find a better RTD analysis method providing robust results under these unfavorable factors, which includes: (i) evaluating the existing RTD analysis methods; (ii) improving the existing RTD analysis method using the Rational model. The influence of measurement error, sampling interval, and ending time in tracer tests on RTD analysis methods was first simulated. The results showed that effective volume ratio was a preferable indicator compared with the variance of RTD. The stability of the moment method decreased with the increase of the measurement error and sampling interval, and the accuracy of the moment method decreased with the decrease of the test time. Although the RTD modeling method showed better stability and accuracy, the implicit premise is that the RTD model adequately describes the observed data. Further, 36 observed RTD data sets were used to test the Tanks in Series (TIS) model and the empirical Rational model. The results showed that the widely used TIS model was only effective in partial cases, while the newly proposed Rational model showed broader applicability. Both the moment method and the TIS modeling method were proven to be inadequate. Finally, the moment method was improved by using Rational models to pre-process RTD data. Although the improved method is still affected by random measurement errors, it can significantly increase the accuracy of the moment method when sampling points are small or the test ends prematurely.