SWY Calibration: Changing beta_i or gamma does not change quickflow outputs

Running SWY 3.9.0 and 3.9.1_post464, I tried to calibrate modeled to observed (seperated) quickflow for a sub-watershed in my area of interest but calculated mean monthly QF [m³/s] did not change at all when trying values from 0 to 1 for beta_i (also tried TWI of 6.86, AOI average) and gamma, but resulted in slightly lower qf-values for trying advanced modus (monthly alpha table and climate zones) - btw mod_qf and obs_qf are fairly similar being in the same order of magnitude yet not close to observed-seperated quickflow values.

  1. One issue I encountered, that while in the handbook it states that beta_i can be used to apply calculated Topographic Wetness Index (TWI) which for me resulted in 6.86 but I can only enter values [0,1] (same for gamma [0,1] allowed only)!?
  2. Any idea why no changes in quickflow values occured when changing beta_i or/and gamma values?

I would be glad for assistance/advice, I tried 14 different parameter combinations and dont know further…:confused:

Hey Felix -

I don’t know about your first question, but for the second one, check out the User Guide equations. It looks like beta_i and gamma are only incorporated after quickflow is calculated, they are used for local recharge and baseflow, not quickflow.

~ Stacie



For your second question, as @swolny said, beta_i and gamma are not used in calculating quickflow, so it is expected that changing them has no effect on quickflow.

For your first question, I am not sure. From the Appendix 2 it seems intended to be a true 0-1 index, hence you cannot enter 6.86. But that contradicts the ln(A/tan(b)) definition, which can be greater than 1. I’ll check with the science team about this.

1 Like

@swolny and @esoth, thanks for the hints on the intended use of beta and gamma in the swy modeling from both of you … than it seems a little unclear (unspecific) in the SWY user guide - may this could be clarified after checking with the team :)! Meantime, I will try to succeed calibration via Kc, CN etc adaptation.
Happy to hear from the science team soon :)!

Update: I tried playing with the Kc values (catchment calibrated → Liu et al. 2017 doi:10.5194/hess-21-311-2017; 0.01 or 1.5 for all LULC types and months/seasonal) and beta_i and yamma once both 0 and once both 1 but nothing (also trying ) changed despite changing modus to advanced inputs and simple default. However, what varies monthly quickflow outputs are changes in CN-Values. Any idea, why changing the other parameters (beta,gamma,Kc) wouldn’t effect monthly quickflow outcome? InVEST-Seasonal-Water-Yield-log-2021-10-21–11_06_56.txt (52.6 KB)

I only found that quickflow could be modified by adjustments on Rain Events (or theoretically Precipitation) or curve number (CN) values. As some months got better and other ones worser modelled I believe adjusting curve numbers first and fine tuning rain events should be fine (to not change the real data, e.g., precipitation raster, too much). Any objections, why not do that? Happy for opinions, thanks! :wink:
Kindly, Felix

I tried calibration for 4 adjacent watersheds, why is it that quickflow value changes easily for one month reaching 100% match with observed values and others do not at all stucking at 10% match despite individually adapting rain events? (considerung the above mentioned, input + adjusting CN (-10%) and iteratively rainevents per month) Shouldn’t the formula 79 apply equaly to all pixels and months…?