watershed in water yield model

I would like to know as only watershed boundary is getting used in the water yield model input, does it incorporate natural drainage feature in it or for getting the water yield model output at a particular spatial unit, any administrative boundary shape file can also be used?

I can understand watershed itself denotes and incorporates all the hydro-logical features in it but I have this doubt in mind as we are using only the shape file of watershed.

Any help is much appreciated.



  • PerrinePerrine Moderator, NatCap Staff
    Dear anuradha,

    In theory any shapefile can be used so yes, you can use a boundary representing an administrative boundary. 
    However, one needs to be extra careful when interpreting results: the 'water yield' output of the model, which is the sum of all the contributions from pixels inside the shapefile, will not represent a volume of water that is found anywhere along a river (since pixels will in reality contribute to 2 or more watershed yields).

    Also remember the model was tested on watersheds only; using it with administrative boundaries is similar to using the pixel values, which we don't recommend (but it still makes sense if you are grouping a large number of pixels).


  • So ideally the model is considering drainage pattern of the area, but going by this hypothesis, there should be higher water yield in the area with a river. But we can't see any positive relationship of the result with that of physical presence of the river. 
  • PerrinePerrine Moderator, NatCap Staff
    Hi anuradha, 

    The annual water yield model does not consider flow routing (or 'drainage patterns'). The pixel values are proxies for the contribution of each pixel to the watershed yield. Their average (if considering values normalized in mm; or their sum, if considering values in m^3) represent the total water yield. 

    Therefore, there is no particular reason why the presence of a river should correlate with high pixel values: a river is defined by the topography, and receives all the contribution from pixels in the same watershed. 
    However, the flow rate of a given river should be correlated with subwatershed water yield values. 

    I hope this helps.

    Again, aggregation by areas that are not defined hydrologically should be done with caution.

  • Now, after calculating annual water yield across different watersheds, we need to calibrate it to the administrative unit level on which we are working on. For this process, I see two possibilities. 

    One, after calculating which how much areal percentage of different watersheds are covering a particular administrative unit and summing up the water yield value as per that percentage. Thus, can get total water yield of an unit. But as we are not considering drainage pattern, how much realistic the result will be out of this calculation as we can't say when for eg 20% of an watershed is there in an administrative unit, the water will flow in which direction, if it will be to that unit of our concern at all.

    two, areal coverage percentage of the majorly covered watershed is put directly to that unit without taking into account other watersheds of smaller areal coverage share.
  • PerrinePerrine Moderator, NatCap Staff
    Hi anuradha,

    I'm not sure I understand what you are trying to do: as mentioned above, the aggregated values by administrative units have limited physical meaning (since they do not correspond to an amount of water flowing in the river).

    Given this, what calibration data do you want to use?
  • so basically what are the major influencing factors for the temporal changes in the model result? As per my understanding there are two- precipitation and land use land cover (LULC) and precipitation changes should have a positive relationship with the water yield changes and if urbanisation increases as evident from lulc, water yield should decrease. But for our study we cant really find any proper relationship to explain the result. How to explain this as we need to justify the result?
  • PerrinePerrine Moderator, NatCap Staff
    Hi anuradha, 

    Two points that may help:
    - the model uses long-term average climate values. So the only temporal changes that one may study is between two long-term periods (for example, 10-year dry period, and a 10-yr wet period, with the corresponding precipitation and reference evapotranspiration rasters).

    - if a watershed undergoes urbanization, the water yield, all things being equal, should increase. This is because the vegetation is replaced by impervious areas that are more "efficient" at producing runoff (instead of using water for evapotranspiration). This may explain your results, and calibration will not change this process.

    Let us know if you have other questions!

  • So, what we are deciphering from your comments, especially the second one, this water yield component of the model is not much suitable to assess water available in a landscape for agricultural activity. Can you shed some light in this regard? 
  • PerrinePerrine Moderator, NatCap Staff
    Hmm, I wouldn't put it that way. In fact, the model has been used quite a lot in agricultural landscapes. 

    I was simply comparing an urban watershed to a watershed with natural vegetation, and saying that we typically see a lower water yield in a vegetated watershed. The water yield of an agricultural watershed (with rain-fed agriculture) would typically be between the urban and the natural vegetation watershed. 
Sign In or Register to comment.