Skip to main content
ARS Home » Research » Publications at this Location » Publication #112219


item Harmel, Daren
item Richardson, Clarence
item King, Kevin

Submitted to: Transactions of the ASAE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/11/2000
Publication Date: 12/1/2000
Citation: Harmel, R.D., Richardson, C.W., King, K.W. 2000. Hydrologic response of a small watershed model to generated precipitation. Transactions of the ASAE. 43(6):1483-1488.

Interpretive Summary: Water quality models such as SWAT (Soil and Water Assessment Tool) simulate the impacts of land management activities on water, sediment, pesticide, and nutrient yields. Water quality models are important tools in assessing hydrologic and water quality impacts of land and water resource management alternatives. To use these water quality models, long-term rainfall data are needed as inputs. In situations which adequate rainfall data are not available, rainfall generators can be used to produce artificial rainfall data. This study provides needed information on the impacts of using various rainfall generators in water quality models. Because of the natural interactions between rainfall, soil conditions, and landuse, both properties of rainfall and resulting simulated runoff need to be examined.

Technical Abstract: In water quality models, long-term daily rainfall data are necessary for continuous simulation, but in the absence of measured data, many models use weather generators to simulate rainfall events. The objective of this study is to examine several daily precipitation generation models in terms of the hydrologic response of SWAT (Soil and Water Assessment Tool). Daily rainfall inputs included a 60-yr measured rainfall record from 1939 to 1998 for Riesel, Texas, and data generated with the precipitation components of three weather generation programs: WGEN, WXGEN, and USCLIMATE. All of these rainfall generators use a first-order Markov chain model to generate the occurrence of wet and dry days; however, each uses a different distribution to determine rainfall amount on wet days. The measured and generated rainfall were input into SWAT and run for a small watershed in Texas. Rainfall totals, extreme rainfall events, and the resulting hydrologic responses of runoff volume and peak flows were then examined. For this study scenario, WXGEN was able to more closely match observed rainfall than WGEN and USCLIMATE. In terms of resulting SWAT hydrologic response, WXGEN rainfall best reproduced runoff volumes simulated with measured rainfall, and USCLIMATE performed better in reproducing peak runoff rates. These are important results as probabilities of exceeding runoff volume or peak flow thresholds are often questions of interest in small watershed projects and environmental impact analyses.