Water quality models like the Agricultural Non-Point Source Pollution model (AGNPS) use a grid-cell representation of a catchment and use the USLE to predict erosion in the grid cells during a rainfall event. The USLE is not well suited to this task. Firstly, the USLE was not designed to predict event erosion. The USLE-M can do this better. Secondly, using the USLE to predict erosion in grid cells results in erosivity (erosive stress) not varying over a catchment when the rainfall is uniform over the catchment. However, erosivity increases as runoff accumulates down slope. Using the USLE-M overcomes this problem.
 

Thirdly, Desment and Govers (1996) developed a USLE slope length factor that  is based on the contributing area (cell area plus the area contributing to the runoff entering the cell from upslope). However, this factor does not account for varitions in runoff that result from, for example, variations in the hydraulic characteristics of the soil. When used in grid cells, the USLE-M overcomes this problem via a hydrodynamic slope length factor. The combination of spatially varying erosvity and the hydrodynamic slope length factor results in cell erosion estimates that vary with landscape position and antecedent soil moisture conditions. Cell erosion predicted by AGNPS is the same irrespective of the catchment being initially WET or DRY . A version of AGNPS where the USLE is replaced by the USLE-M is called AGNPS-UM (see KINNELL,P.I.A. (2000). AGNPS-UM: Applying the USLE-M within the Agricultural Non Point Source Pollution model. Environmental Modelling and Software 15, 331-341).

Read AGNPS-UM User Guide
Download AGNPS-UM software

Using Sediment Delivery Ratios to predict sediment delivery to channles