Submitted to: First Interagency Conference on Research in the Watersheds
Publication Type: Proceedings
Publication Acceptance Date: August 20, 2003
Publication Date: September 15, 2003
Citation: Canfield, H.E., Goodrich, D.C. 2003. Studies of scale and processes in hydrologic modeling on the lucky hills watersheds. Proceedings First Interagency Conference on Research in the Watersheds. Oct. 27-30, 2003. Benson, AZ., pp. 444-450. Interpretive Summary: Hydrologic models require data in order to develop relationships that describe runoff and erosion from rainfall. Data collected at the 4.4ha Lucky Hills 104 since 1963 has been valuable for understanding runoff and erosion processes on small watersheds. These data have been used in the KINEROS2 hydrologic model to better understand hydrologic processes as well as to understand the capabilities and limitations of hydrologic models. This paper re-caps some of the research conducted using data from Lucky Hills. Research shows that rainfall is highly variable even on small watersheds like the Lucky Hills 104, and more than one rain gauge is necessary to quantify precipitation at this scale. In addition, runoff decreases systematically as fewer of the channels observed in the field are represented in the model. However, for small runoff events, the amount of erosion is more affected by the magnitude of the event than by the number of channels in the model representation. In contrast, for large events, erosion is affected by the number of channels in the channel representation.
Technical Abstract: Hydrologic and sediment yield data collected from Lucky Hills 104 within the Walnut Gulch Experimental Watershed have been used extensively to study the effects of scale in the KINEROS model. These studies show that lumping of parameters derived at the small scale into increasingly less complex geometry has the effect of reducing runoff volume, peak and sediment yield. Model simulations mirror observations at the Lucky Hills watersheds showing a decrease in runoff per unit area as watershed size increases. Studies of rainfall variability on model response show that even at <5 ha scale, data from a single gauge cannot adequately describe rainfall input, and can lead to errors in runoff modeling. Studies of watershed representation indicate that runoff volume can be simulated using less complex geometries if equilibrium storage of runoff is maintained. Likewise sediment yield can be simulated using a simplified watershed representation by increasing the entrainment of sediment by raindrop impact on hillslopes even as entrainment by flowing water decreases as fewer channels are represented in the model.