Quantifying extreme precipitation events and their hydrologic response in Southeastern Arizona

Authors: Keefer, Timothy; RENARD, K.G. - Retired ARS Employee; Goodrich, David - Dave; Heilman, Philip - Phil; Unkrich, Carl

Submitted to: American Society of Civil Engineers
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 5/19/2015
Publication Date: 7/10/2015
Citation: Keefer, T.O., Renard, K., Goodrich, D.C., Heilman, P., Unkrich, C.L. 2015. Quantifying extreme precipitation events and their hydrologic response in Southeastern Arizona. American Society of Civil Engineers. 21(1):1-10. doi: 10.1061/(ASCE)HE.1943-5584.0001270, 04015054.

Interpretive Summary: Civil engineers need good estimates of extreme rainfall and runoff amounts and intensities to properly design the size of drainage and flood control structures such as highway culverts and bridges. It is common to use the storm information provided by the National Weather Service for such design problems. The National Weather Service collects and analyzes rainfall data from around the country to calculate and publish the rainfall amounts for different locations in a format that the design engineers can use. This format is referred to as “intensity-duration-frequency”. That is a certain rainfall intensity (in inches per hour), for a certain duration (in minutes from 5, 10, 15, 30, 60,…), has a certain frequency of recurring (in years from 5, 10, 25, 50 to 100). For the Southwestern US, the rain gauges are spaced much farther apart than rain gauges in more populated areas and very few of them record data at time scales less than a day. Thus, the National Weather Service must estimate the intensities of sub-daily durations. In this study we compare observations of the intensity-duration-frequency derived from a network of rain gauges on an area of about 50 square miles in southeastern Arizona. The network is called the Walnut Gulch Experimental Watershed and is operated by the US Dept. of Agriculture, Agricultural Research Service, Southwest Watershed Research Center. We find that our results and those of the National Weather Service are similar. We also show that many events that have a frequency of recurrence of 1 in 1000 years have been observed in the network over the last 53 years and that the impacts of these types of extreme rainfall can be large floods over small areas. The significance is that as the southwest US continues to grow in population because of more people moving to this area of the country, these flood events will cause more damage and potential hazards.

Technical Abstract: Design criteria such as rainfall intensities and runoff rates for small watersheds (<200km2) are needed for modeling, sizing and design of drainage and flood control structures. In the Southwest US the need for accurate information about these rates is increasingly important as development of rangeland increases due to population growth. These rainfall criteria are routinely derived from NOAA Atlas 14. For the Southwestern US during the North American Monsoon, NOAA relies on sparse rain gauge networks to measure rainfall from limited area convective storms primarily at daily time steps and estimates of sub-daily and sub-hourly event intensities are derived from temporal down-scaling from a few point locations. The US Department of Agriculture, Agricultural Research Service, Southwest Watershed Research Center (SWRC) operates the Walnut Gulch Experimental Watershed (WGEW) in the vicinity of Tombstone, Arizona. SWRC maintains a database of 60 years of sub-daily, high temporal-precision rainfall intensities and runoff rates for WGEW. Updated, temporally-extended, rainfall intensity-duration-frequency relations for WGEW are presented. The current analysis includes intensity-duration-frequency relations for July, August and September for 53 years, 1961-2013, for 2-, 5-, 10-, 15-, 30- and 60-minute durations and 2-, 5-, 10-, 25-, 50-, 100- and 1000-year return periods. Assumed spatial independence of event amounts allows the combining of four independent gauges’ 53-year time series into a longer time series of 212 years. A comparison of WGEW-generated intensity-duration-frequency curves to those of NOAA Atlas 14 shows reasonable agreement. However, across the range of durations, many observed events on WGEW are much greater than the estimated 100-year event. Several of these extreme rainfall events on and near WGEW are used to simulate runoff on two small watersheds with the KINEROS2 simulation model. Model results of runoff volume and peak discharge rate are 2 to 4 times as large as the largest observed runoff events of record. These analyses offer insights into the benefit of long-term watershed research and high temporal resolution data.