资源环境科技发展态势分析平台

Bartlett et al. (2016) performed a re-interpretation and modification of the space-time lumped USDA NRCS (formerly SCS) Curve Number (CN) method to extend its applicability to forested watersheds. We believe that the well documented limitations of the CN method severely constrains the applicability of the modifications proposed by Bartlett et al. (2016). This forward-looking comment urges the research communities in hydrologic science and engineering to consider the CN method as a stepping stone that has outlived its usefulness in research. The CN method fills a narrow niche in certain settings as a parsimonious method having utility as an empirical equation to estimate runoff from a given amount of rainfall, which originated as a static functional form that fits rainfall-runoff data sets. Sixty five years of use and multiple reinterpretations have not resulted in improved hydrological predictability using the method. We suggest that the research community should move forward by (1) identifying appropriate dynamic hydrological model formulations for different hydro-geographic settings, (2) specifying needed model capabilities for solving different classes of problems (e.g., flooding, erosion/sedimentation, nutrient transport, water management, etc.) in different hydro-geographic settings, and (3) expanding data collection and research programs to help ameliorate the so-called "overparameterization'' problem in contemporary modeling. Many decades of advances in geo-spatial data and processing, computation, and understanding are being squandered on continued focus on the static CN regression method. It is time to truly "move beyond'' the Curve Number method.

Plain Language Summary The Curve Number method was invented to solve an immediate need in the 1950s associated with construction of soil conservation dams in small agricultural watersheds. The method is an equation form that on average fit observed storm-total rainfall and runoff data from small agricultural watersheds. In some regards, the method has been a fantastic success in that it allows a very simple calculation of how many inches of runoff will result from a certain number of inches of rainfall. The method was introduced by the USDA by fiat without peer review. Its success led to extrapolations and extensions, each one further and further from the simple original purpose of the method. At its heart, the curve number method is a static one-or two-parameter model that is a regression fit. The extrapolations of the method attempt to impart time-changing parameter behavior that is so far from the original method as to be completely without justification. These extrapolations of the method have not resulted in improved predictive ability, and the lack of theoretical underpinnings prevents improved understanding from these extensions. We assert that it's time for the hydrological science and engineering research community to leave the method behind.