K-factor adjustment for known gullies?

Hi all,

I'm using InVEST which uses the RUSLE to estimate soil erosion. The RUSLE considers only sheet and rill erosion, and therefore massively under-predicts gully erosion if that is present in the landscape. In my case, permanent gullies are a frequent part of the landscape, and contribute greatly to erosion as they expand.

In my LULC map, they are classified as "bare soil," which means they get assigned high C-factor values. Now, I wonder if I should also account for gullies in my K-factor map, since gully locations (polygons) are known. 

I wonder if others have come across this issue and how they've proceeded...

-Should higher K-factor values be assigned at gully sites in order to "boost" the soil loss calculation at these places? (Or simply not since the RUSLE was not developed for this purpose?)

-And if you would adjust K-factor values at gullies, what (literature) values could be used? (I am aware of Kc, channel erodibility, but I understand from the excerpt below that Kc and Kusle cannot be reliably related.)

from Knapen et al. (2007) ("Suitability of SWAT Model for
Sediment Yields Modelling
in the Eastern Africa")
The concentrated flow erodibility factor on the
other hand does not include infiltration effects, which
are generally accounted for in the hydrology component
of the models. Therefore, using KUSLE values as a
predictor for Kc values (e.g. in the CREAMS model, the
Kc value could be adjusted by the user according to soil
texture, by multiplying 0.39 times KUSLE, Foster et al.,
1980, see Table 7) does not have merit, especially not
for sandy soils where infiltration plays an important
role. Using the KUSLE factor and accounting for
infiltration in the hydrology component of the model
would be a double accounting for the effect.


  • PerrinePerrine Moderator, NatCap Staff
    Hi jelliso,

    I don't have experience with modeling gullies but I agree that simple adjustments can be made to the InVEST model.

    Modifying the K values as you suggest is one option, but not always practical if you don't find empirical values in the literature. 
    If you have estimates of the soil loss from gullies, it may be easier to use them in post-processing. In other words, run the model without modifications, and then play with the sediment export map to modify values under the gully polygons.

    One thing to remember is that gullies are likely more connected to streams, so you may want to modify the sediment delivery ratio (SDR) too. (This is done by default if you modify the sediment export map (sed_export) since the values already account for the sediment delivery factor.)


  • Thanks Perrine,

    I think you are right to just deal with gullies in post-processing. In my case, I'm sure gully head-cut retreat and bank collapse/erosion comprise a great deal of the total soil loss for the catchment. 

Sign In or Register to comment.