Suitability of CLIGEN precipitation estimates based on an updated database and their impacts on urban runoff

Authors: CHEN, JINGQIU - Purdue University; GITAU, MARGARET - Purdue University; ENGEL, BERNARD - Purdue University; Flanagan, Dennis

Submitted to: Hydrological Sciences Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/20/2018
Publication Date: 9/24/2018
Citation: Chen, J., Gitau, M.W., Engel, B.A., Flanagan, D.C. 2018. Suitability of CLIGEN precipitation estimates based on an updated database and their impacts on urban runoff. Hydrological Sciences Journal. 60(10):1502-1518. https://doi.org/10.1080/02626667.2018.1513655.
DOI: https://doi.org/10.1080/02626667.2018.1513655

Interpretive Summary: Daily rainfall is one of the most important weather parameters that affects water supplies, crop growth, flooding, drought, runoff, and soil erosion by water. Various natural resource processes such as runoff and soil loss are commonly predicted using computer simulation models, where one of the driving input parameters is the amount of daily rainfall and its characteristics (duration, intensity). Climate generators are other computer simulation models whose purpose is to generate long sequences of daily weather values, that reproduce the observed statistics from a weather station. In this study, we used an updated set of climate generator parameters files containing weather station rainfall statistics created with information from 1974-2013 at over 2700 locations in the United States. In particular we examined the effects of changes in the climate generator predictions for 5 Great Lakes states weather stations, compared to the existing station database, on rainfall predictions, and also the effect on use of the updated data as input to a hydrologic model. We found that climate generation with the updated database produced rainfall predictions that matched the observed station data well. The use of the updated database also affected runoff predictions with the hydrologic model, with more average annual urban runoff being simulated. These results impact research scientists, university faculty, students, environmental engineers, soil conservation agency personnel and others involved in natural resource modeling with generated climate inputs. Use of the updated climate database is recommended as it contains the most recent station observations, that are consistent over the same 40-year period of time.

Technical Abstract: The quality of synthesized weather inputs to natural resource models has direct influence on the responses of model applications. Weather generators simulate synthetic weather inputs time series that have similar statistical properties with observed data for specific sites. Up-to-date historical records and their statistical parameterization are important for the usefulness of weather generators. The database for the weather generator CLIGEN has recently been updated to comprise up-to-date historical measurements from 1974 to 2013 (updated CLIGEN database, UCD). In this study, we statistically evaluated the performance of CLIGEN using the updated database regarding precipitation estimates. The UCD-based daily precipitation were then input to a hydrological model - the Long-Term Hydrologic Impact Assessment—for urban runoff simulation in five states in the U.S. Great Lakes Region (Wisconsin, Illinois, Indiana, Michigan, and Ohio) based on the National Land Cover Database 2011 edition (2001, 2006, and 2011). Results were further compared with those obtained using precipitation estimates based on the current CLIGEN database (CCD). Statistical analysis revealed that UCD-based precipitation can adequately replicate observed daily precipitation up to the 99.5th percentile. However, maximum precipitation for different time series, including daily, monthly, and annual values tended to be largely underestimated. L-THIA model results differed between precipitation estimates, with about 0.57 billion cubic meters more average annual urban runoff being simulated using UCD-based precipitation compared to those based on CCD. The total number of counties in the five states with medium, high, and very high urban runoff computed using UCD-based precipitation increased by 31%, 67%, and 14%, respectively compared with those based on CCD. Moreover, urbanization intensify precipitation impacts on urban runoff and offset the effects of decreased precipitation. This study showed that the updated database provides suitable inputs for CLIGEN to simulate precipitation. The assessment of urban runoff based on UCD-based precipitation provide useful information assisting sustainable development.