Submitted to: Journal of the American Water Resources Association
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/19/2008
Publication Date: 4/1/2009
Citation: Chen, J., Zhang, X.J., Liu, W.J., Li, Z. 2009. Evaluating and extending CLIGEN precipitation generation for the Loess Plateau of China. Journal of the American Water Resources Association. 45(2):378-396. Interpretive Summary: Computer-based weather generators are often used to generate daily weather data to drive hydrologic and crop models. The quality of generated daily weather directly affects the output of those response models. This work was to evaluate the ability of a widely used climate generator (CLIGEN) in generating daily, monthly, and annual precipitation amounts, and storm patterns (i.e. storm duration, peak rainfall intensity, and time to peak rainfall for a new region of the Loess Plateau of China. Historic daily precipitation data from 12 stations were used in the evaluation. Results indicate that CLIGEN satisfactorily generated daily, monthly, and annual precipitation amounts. However, CLIGEN generated storm patterns rather poorly. For better storm pattern generation, the model must be modified and improved for use in the region. Overall, this work shows that the CLIGEN model is a useful tool for generating daily precipitation amounts for the region, which can be used by scientists and extension specialists to evaluate natural resource responses to precipitation variations.
Technical Abstract: Climate generator (CLIGEN), which is used in the United States, needs to be evaluated before used to a new region of the Loess Plateau of China. The objectives were to: (1) evaluate the reproducibility of CLIGEN in generating daily precipitation depth (R), as well as storm patterns including storm duration (D), relative peak intensity (ip) and peak intensity (rp) at 10 stations dispersed across the Loess Plateau; (2) test whether an exponential distribution for generating D and a distribution-free approach for inducing desired rank correlation between R and D can improve storm pattern generations. Mean absolute relative errors (ARE) for simulating daily, monthly, annual and annual maximum daily precipitation depth across all 12 stations were 3.5, 1.7, 1.7 and 5.0% for the means and 5.0, 4.5, 13.0 and 13.6% for the standard deviations, respectively. The model reproduced the distributions of monthly and annual precipitation depths very well (P=0.3~0.95). The first–order, two–state Markov chain algorithm is adequate for generating precipitation occurrence for the region. The CLIGEN generated storm patterns poorly. It under-predicted means and standard deviations of D for storms ' 10 mm by -60.4% and -72.6%, respectively. Compared to D, ip and rp were slightly better generated. Mean AREs of means and standard deviations were 21.0% and 52.1% for ip, and 31.2% and 55.2% for rp, respectively. When an exponential distribution was used to generate D, mean AREs were reduced to 2.6% for means and 7.8% for standard deviations. However, when the exponential D was used, ip became worse due to the overestimation of rp, with mean AREs of 128.9% for the means and 241.1% for the standard deviations. Overall, CLIGEN is adequate in generating precipitation depths and occurrence for the region, but the storm pattern generation needs improvement. For better storm pattern generation for the region, new methods such as Copula may be used to generate correlative R, D, and rp.