Assessment of CLIGEN precipitation and storm pattern generation in China

Authors: WANG, WENTING - Beijing Normal University; Flanagan, Dennis; YIN, SHUIQING - Beijing Normal University; YU, BOFU - Griffiths University

Submitted to: Catena
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/18/2018
Publication Date: 5/31/2018
Citation: Wang, W., Flanagan, D.C., Yin, S., Yu, B. 2018. Assessment of CLIGEN precipitation and storm pattern generation in China. Catena. 169:96-106. https://doi.org/10.1016/j.catena.2018.05.024.
DOI: https://doi.org/10.1016/j.catena.2018.05.024

Interpretive Summary: Rainfall is the most important driving force behind soil erosion by water – raindrops impact and detach soil particles, soil surfaces can seal and crust due to rainfall disintegration of soil aggregates, and water from rainfall can become surface runoff that can carry sediment and also dislodge and erode more soil in rills, channels, and gullies. Computer simulation models are often used to estimate the occurrence, amount (depth), and intensity of rainfall events, and then these predictions can be used as input to other models to predict runoff, soil erosion, and sediment losses. In this study we evaluated the ability of a climate generation model (CLIGEN) to adequately simulate rainfall characteristics for 18 weather stations in China where very detailed information (1-minute precipitation depth) was available to assess model performance. We found that CLIGEN was able to predict the occurrence and depth of rainfall well, but did not perform as well at estimating the rain storm durations and intensities for these Chinese sites. These results impact scientists, university faculty, students, and others utilizing the CLIGEN software to predict strings of daily simulated rainfall for use in soil erosion, water quality, or other modeling applications. Users should be aware of deficiencies in the climate generation software, and future efforts to further improve CLIGEN are warranted.

Technical Abstract: CLIGEN is a stochastic weather generator widely used to simulate daily precipitation and storm pattern as input to hydrological and soil erosion models. To assess its applicability in China, long-term precipitation data at 1-min interval from 18 sites in eastern and central China were used to generate 100-year precipitation data with CLIGEN. CLIGEN performance was evaluated in terms of (1) daily precipitation, storm duration and peak intensity for all events as well as four separate categories in terms of the daily precipitation depth: light (< 10mm), moderate (10-25 mm), heavy (25-50mm), intense (> 50mm); (2) climate inputs for the Revised Universal Soil Loss Equation (RUSLE), (3) peak storm intensity for durations from 5 min to 24 h and return periods from 2 to 100 years. Results showed that CLIGEN was able to reproduce accurately the mean, standard deviation, skewness coefficient, and the probability distribution of daily precipitation depth in general. The mean storm duration was underestimated for moderate, heavy and intense storms, but overestimated for light storms. CLIGEN underestimated the maximum 1-min intensity and overestimated the maximum 30-min intensity, whereas there was no obvious bias in the maximum 5-min intensity for these 18 sites. In addition, the relative error in the generated peak intensity increased as the magnitude of storm event increased. Rainfall erosivity generated with CLIGEN outputs was systematically larger than, but well correlated with, the measured erosivity values (slope=0.547, R2=0.96 for the R-factor; slope=0.576, R2=0.81 for the 10-year storm erosivity), which was consistent with previous finding for sites in the United States and Australia. Intensity for a given storm duration and return period was systematically over-predicted using output from CLIGEN. The average regression slopes between measured (X) and generated (Y) intensities (Y = b X) increased from 1.19 (5 min) to 1.43 (1 h), then decreased to 1.01 (12 h) and 0.96 (24 h). As a stochastic weather generator, CLIGEN is able to reproduce daily precipitation very well, but its capacity to simulate storm duration, and the peak intensity for a given time interval needs to be improved, based on this study for 18 sites in China.