Watershed Models for Development and Implementation of Total Maximum Daily Loads

Abstract

This paper reviews 14 prevalent watershed models for their capabilities, credibility, and suitability in total maximum daily load (TMDL) development and implementation. Brief descriptions of the models, including sources, capabilities, and applicability are presented. General information such as intended watershed and simulation types, simulated outputs, uncertainty analysis capabilities, graphical user interface, and availability are also presented. Mathematical bases of the hydrologic and water quality simulations, which indicate credibility, expected performance, and accuracy, and dictate model features (e.g., structure, input data, and parameters) are presented. Routing procedures, the backbones of the models, are compared and ranked. The Gridded Surface and Subsurface Hydrologic Analysis (GSSHA) and MIKE SHE (Système Hydrologique Européen) models rank high on overland and Storm Water Management Model (SWMM) on channel/pipe flow routing as the most accurate for representation of the physical processes and also the most numerically complex. The Generalized Watershed Loading Function (GWLF) and Spreadsheet Tool for the Estimation of Pollutant Load (STEPL) rank at the bottom on both the aspects. The rest of the models are in between, although the Dynamic Watershed Simulation Model (DWSM) is computationally efficient among kinematic wave models. Notable strengths and limitations of the models for TMDL development and implementation are presented. All these provide valuable information on the models, not readily available in a concise form elsewhere, to compare and help determine relative credibility and make informed selections for TMDLs and similar studies. Future research should focus on further comparisons of the models based on other key aspects such as simulation capabilities of processes, uncertainty analysis, required resources, and performances on watersheds followed by developing better models or improving existing ones by strengthening the weaknesses found. Robust physically based algorithms, uncertainty analysis capabilities, and use of remotely sensed and high-resolution data are recommended to be part of the model improvements.