Water Yield Model - factor Z

unbmaiteunbmaite Member
edited May 2016 in Freshwater Models

I am using InVEST 3.3.0 x86  in a small watershed in Brazil. I had no big troubles to find and to calculate the parameters required to run the model, no errors.  However, since I have observed data, I am trying to calibrate the factor Z to compare with my data. 

I have data from 2003 to 2015 and  when I first run the model, I used different Z values for each year, in a way I could match or get closer to my real data. 

Based on what is the user´s guide " The model is an annual average time step simulation tool", someone told me that I could use different Z values for, let´s say, the years from 2003  to 2008 and get an average of these years, and try it on the other years, but I don´t know how accurate this is. I tried to do that, and the results were not what I expected. 

I have been reading a lot about how to calibrate this parameter and how to validate it... I realize that: higher Z values, the response is often a reduction in water yield and by doing some tests Z value is very sensitive to precipitation. 

My question is if the analysis can be doing accounting a time series database (using an average or something) or if the calibration should be done year by year, different Z values for each year.  ( as the user´s guide says: "The biophysical models do not consider surface – ground water interactions or the temporal dimension of water supply."). 

And if it´s that the case, how can I use this value when I change my land use scenario? I mean, after the calibration  of the model, I would like to predict the water yield in a different land use cover situation, considering reforestation for example. 

Sorry for the long post


Post edited by unbmaite on


  • PerrinePerrine Moderator, NatCap Staff

    Great that you have data to calibrate the model. As you point out, the model should be used for long term averages. This is because the theory ignores interannual storage (and this assumptions typically breaks down if a wet year follows dry year, followed by another dry year, etc.)

    So you should *not* calibrate for each year and get a different value each time but rather calculate the average precipitation over all the years, and similarly the average reference evapotranspiration and streamflow over the years.; this is the data you can you use to find the calibrated Z value.

    For your land use change scenario analysis you then keep this Z value since it accounts for watershed characteristics that aren't captured by land use.

Sign In or Register to comment.