Downscaling Precipitation for Local-Scale Hydrologic Modeling Applications: Comparison of Traditional and Combined Change Factor Methodologies

Abstract

Future precipitation projections—and subsequent variation in simulated runoff response—can have a large impact on the planning and design of hydraulic structures and water systems and are, therefore, an important input in hydrologic modeling. These projections are often derived from coarse-scaled climate models and may require downscaling and bias-correction techniques to be suitable for local-scaled applications. Here, one simple and commonly used downscaling approach, called change factor methodology (CFM), is modified to combine both additive and multiplicative change factors depending on the characteristics of the empirical cumulative distribution functions and limitations of precipitation data. The combined change factor methodology (CCFM) is applied as a secondary bias-correction technique to general circulation model (GCM) data for a comparison period of 1985–2014 and a future period of 2055–2084 in six locations throughout the United States that differ greatly in local climate and precipitation patterns to examine the method in a range of settings: Salt Lake City, Utah; Toledo, Ohio; Seattle, Washington; Houston, Texas; Miami, Florida; and Phoenix, Arizona. The CCFM successfully addresses several common issues inherent with traditional CFM, including negative precipitation, overestimation, and artificially inflated numbers of precipitation events. During the comparison period, the CCFM results in precipitation time series that more closely match observed precipitation patterns (average, extreme values, number of events) than traditional CFM, and are generally closer than the values produced by the uncorrected projections. This study also identifies remaining limitations to the CCFM, such as representation of potential nonstationarity in future precipitation events or differences in extreme precipitation values. The uncorrected and CCFM-scaled projections are also used as inputs for a hypothetical urban hydrologic model to demonstrate the consequences of using the CCFM in modeling applications. This modeling exercise shows that on a monthly scale, the projections with no secondary correction result in greater magnitudes of change (compared to historical conditions) in average runoff than the CCFM projections. This highlights that the use of the CCFM as a secondary bias-correction technique has potential to have a substantial impact on the resulting hydrologic analysis and subsequent planning, design, or management strategies.