This forum is shutting down! Please post new discussions at community.naturalcapitalproject.org

Seasonal Water Yield - ASCE updates to CN methods

Hi freshwater modelling team, 

I was wondering if you are planning to incorporate any of the modifications to Curve Number methods that have been recommended by the ASCE-ASABE revisions to the USDA NCRS NEH Part 630 Hydrology?  
(The revised chapters, 8 to 12, were released in September 2017, and are now in a process of final review before the new NEH Part 630 is released. Available online: https://policy.nrcs.usda.gov/viewDirective.aspx?hid=41610)

For example, the recommendations replace the assumption that Ia = 0.2 S (initial abstraction vs potential storage), with Ia = 0.05 S.

This leads to a new recommended CN equation [altering Eqn 1 in the InVEST SWY userguide], and also revised tables of CN values (for each combination of LULC class and hydrologic soil group).

Page 8, point 7: 
"The initial abstraction coefficient, Ia/S, (referred to as λ, or lambda) shown in Equation [10-1] as the coefficient of 0.2 is variable, and more appropriately 0.05. The use of 0.05 value is recommended. 
With this change in λ, Equation [10-1] becomes 
Q = (P-0.05 S05)2/(P+0.95 S05) for P>0.05 S05, Q = 0 otherwise. [Equation 10-3]"

I think @Rich has made an untested build that implements this change in the equation, as described in this thread about changing lambda:
https://forums.naturalcapitalproject.org/index.php?p=/discussion/1293/seasonal-water-yield-code

I think the only other change needed to implement this recommendation, is that the CN values for each land use would need to be assigned while accounting for the changed lambda, since CN values for Lambda=0.05 are different from CN for lambda = 0.2.

The revised CN tables - and also empirical equations that enable conversion between CN for lambda = 0.05 and 0.20,  are given in the revised Ch 9:

The revised CN tables have also been extended to include a much larger range of LULC classes, which could be helpful for a lot of InVEST users (with the caution of distinguishing CN 0.05 and CN 0.20)

Anyway, I though I would ask if your team at InVEST has any thoughts on this recommendation, or plans to look into incorporating it into future versions of the SWY? 

Thanks for your time,
- Jessie

Comments

  • PerrinePerrine Moderator, NatCap Staff
    Hi Jessie, 

    Thanks for these good thoughts! We had this exact conversation internally just yesterday, following wangyao's post that you reference.

    We've referenced the evolution of the SCS-CN method in a recent paper (Guswa et al., 2018) but for now we are not planning to incorporate it in the InVEST model. As you mention, the proposed change for lambda has not been accepted by the NRCS yet, so it sounds slightly early to incorporate it into InVEST, especially given the implications for the CN values (users would need to know that the InVEST version uses the "new" set of values).

    Of course I'm willing to revise this decision if users feel this would be a significant improvement, so please share other thoughts you may have on the topic!

    Cheers


    Ref: Guswa, A. J., Hamel, P., & Meyer, K. (2018). Curve number approach to estimate monthly and annual direct runoff. Journal of Hydrologic Engineering, 23(2), 1–10. 
  • Thanks Perrine,

    I hadn't seen your 2018 paper, it's great!  

    I can understand the need for caution over adding new options, and avoiding any confusion over which set of CN values to use. 

    Still, it seems that there is sufficient evidence in favour of using Ia/S = 0.05 rather than 0.20 (i.e. motivating is use by Guswa et al 2018, and the recommendations by the ASCE's expert teams), that I would definitely like to explore the consequences of this assumption in current projects, especially if they are to inform land use planning in the coming  two years or so (so the results remain current if the revision is accepted by NRCS).

    It is great that Rich has kindly offered this change as a custom build. I'm only hesitant about it being untested.
    (Though if the only change is to lambda values in Eq 1, then I imagine this build it should be as stable and replicable as the public release on which it is based?)

    Perhaps a future version of the SWY could continue to set the default Lambda as 0.20, and gave an option to select Lambda = 0.05, perhaps with a note of caution to ensure corresponding use of CN0.05 values and links to conversion equations

    In case it's of any help to anyone else, I did some comparisons of the conversion formulae for converting between CN0.05 and CN0.20 - as there is a slight difference between the equations:
    1. ASCE 10-17 given in ASCE (2017) for CN values based on asymptotic fitting of rank-ordered data sets;
    2. Hawkins 2009 - given in ASCE (2017) as 10-18, based on direct least squares fits of CN and S to P:Q natural (not rank-ordered) data sets;
    and 
    3. Jiang 2001 - given in Guswa 2018, also based on direct least squares fits of CN and S to P:Q natural (not rank-ordered) data sets.
  • RichRich Administrator, NatCap Staff
    FYI @JessieWells, that experimental branch is tested from a software perspective, but not officially supported unless Perrine says so. So please don't avoid using it because you think there might be a software bug.
Sign In or Register to comment.