Annual Water yield model calibration

hi , Iam using Annual water yield model for my basin (TungaBhadra basin), india.
My results were 200% more than the actual observed value, z value 10 and Kc as 1 for vegetation (major landuse)
Iam using WorldClim data for precip and PET, PWAC is calculated using SPAW software, Root Restricting layers global data.

Hi @vijay01996 -

Uncalibrated model results generally don’t match real-world observations, especially with a very simplified model like this. So we always recommend calibrating against the observed value to have more confidence in the absolute values that the model produces. Otherwise, it is only recommended to use low/high relative values when reporting the results. Since you have observed data, you might be able to do this calibration, and the User Guide gives some guidance for this.

~ Stacie

calibration is nothing but adjusting the values of Z and Kc until the water yield matches with the observed value, is it true??
firstly z value is 13(calculated using number of rainy days /5) and Z for cropland as 1, with this values Iam getting the yield as 21X10^9 but observed value is 5.6X 10^9.
does we can change the value of Kc and Z to any number or is there any limit for this values.
if you need my data i will sent it.

Yes, we usually adjust Z and Kc until the model result matches the observed value. For more information, I recommend reading the following paper, written by one of our hydrologists, that talks about doing calibration and uncertainty analysis with the annual water yield model:

Hamel, P. and Guswa, A. J.: Uncertainty analysis of a spatially explicit annual water-balance model: case study of the Cape Fear basin, North Carolina, Hydrol. Earth Syst. Sci., 19, 839–853, https://doi.org/10.5194/hess-19-839-2015, 2015.

~ Stacie

can u send me the links of root restricting layer and PWAC of world data, i will check with that.
as iam using precipitation and ET from worldcilm only.
if possible can u be in contact with me in mail vbejagam@student.nitw.ac.in or whatsapp
+91 7780115904

Hi @vijay01996 -

There are several global soil datasets that provide soil depth. The one we probably use most often is ISRIC Soil Grids. The InVEST User Guide for this model has an Appendix that lists some global sources of data for each input.

As far as I know, there is no already-prepared global dataset for PAWC, this is something that we need to create manually using other soil data, like you would get from ISRIC or the Harmonized World Soil Database. Years ago, when we were still using HWSD, I made a simple document outlining how to process the HWSD data into soil depth and PAWC - it might still be useful as a guide for preparing data from other soil databases.

~ Stacie

thank you so much for your information, one small question
how much coefficient of Determination(R^2) or bias should be allowed with simulated and observed?

Hi @vijay01996
there is not really a clear threshold for what R2 the model needs to reach before you can consider the results sufficient. Do you have many observations of water yield in the area you’re working in? In general I would say an R2 of at least 0.8 is considered a good match but ideally the R2 should be above 0.9. For more info see, e.g., here specifically table 2.
Rafael