puzzled about Z factor

Hello !
     I am using InVEST 3.3.0 x86  in a small watershed and  have some puzzled about how to accurately determine the Z value , I've tried many times and the total amount of water produced in the experiment is always smaller than the natural runoff.
    And another puzzled: The PAWC is a fraction from 0 to 1. I use the soil texture data(%clay, %sand, %silt,%oc) which obtained from the World Soil Database provided by FAO to estimate my PAWC.There is no value in the world soil database for waters, buildings, bare land, etc.Can I set PAWC for these places to 1 and the Root restricting layer depth to 0?
Thank you 


  • swolnyswolny Member, NatCap Staff
    Are you using the Z value to calibrate the model against observed data? If so, Z isn't the only value that should be considered for calibration. Doing a sensitivity analysis to see which inputs (precipitation, Kc, etc) have the greatest impact on results can show you which of these inputs to take a closer look at, and get more accurate values for. For example, if you have a lot of forest in your watershed, and the model is sensitive to Kc for forest, then see if you have best available values for that coefficient, or if that value can be improved. I'll also note that if you're using HWSD for soils, that is a very coarse data layer that can be a significant source of error. It might be the best available data for your area of interest, but it's definitely worth looking for something more detailed. 

    For a discussion on sensitivity and calibration for the water yield model, check out the paper Uncertainty analysis of a spatially-explicit
    annual water-balance model: case study
    of the Cape Fear catchment, NC (2014), P. Hamel and A. J. Guswa

    It is unfortunate that the HWSD does not give values for areas with water, etc (I think that it does give values for bare land, but maybe not glaciers, rock, and similar places where there's not really soil per se.) As the User Guide indicates, for non-vegetated LULC types (water, urban, etc) the model does not use soil and rooting depth information to calculate AET, so theoretically you can set these values to 0/1, as they will be ignored. However, just because the HWSD thinks there's urban in a place, doesn't mean that your LULC thinks there's urban in that same place, so there is likely to be a mismatch (say, where your LULC says there's forest, but HWSD gives no soil data.) And the model determines which AET equation to use based on LULC classes, not the soil map. 

    So yes, this gets tricky. You can try doing the comparison between the LULC and HWSD to see if there are any mismatches, and where there are, fill in the HWSD soil data with the values surrounding the soil polygons where there is no data. You can also look at sources for AWC such as FutureWater's HiHydro (available upon request to FutureWater), which has AWC prepared (although I am not sure how they have handled the HWSD missing data.)

    ~ Stacie
Sign In or Register to comment.