This forum is shutting down! Please post new discussions at community.naturalcapitalproject.org

Questions about the results of the seasonal water yield model

CharChar Member
edited November 2018 in General
Hi~
      After running the seasonal water yield model successfully, B.tif from the model's results were analyzed . Different from the results from your Myanmar research,vegetation covered areas of my study area tended to have lower baseflow. The Highest mean value was found in built-up area and bare lands and the lowest  mean value was in water bodies.The grassland and forest had the medium baseflow and the mean value of grassland was twice as high as that of forest.I find that these results of mine are  similar with Mesfin Sahle et al.(2019) in certain degrees.Though the model results may differ in some aspects from different studies,are there some general rules of this model's results?And if possible, could you help me with finding some reasons for these results so that I may explain or improve them? 

PS:Following the method to assess flood risk reduction in your Myanmar research, the flood risk reduction index was also calculated from the monthly results of quick flow and precipitation. The result of this index seemed more reseanable. Similar with the Myanmar research,the mean value of flood risk reduction index in vegetation covered areas was higher.

Looking forward to your early reply,thank you!


(the Mesfin Sahle 's article is available at https://www.sciencedirect.com/science/article/pii/S0048969718328420

Post edited by Char on

Comments

  • I noticed that the vegetation covered areas in my study area were mainly distributed in mountainous areas with high DEM values,while most of the built-up areas and cropland were on the plain next to these mountains.Since this model emphasizes the topographic position of a pixel,could this be a main reason for my baseflow results?
  • swolnyswolny Member, NatCap Staff
    Hi @Char -

    Topography certainly does affect the baseflow values, and may be a main reason for your results. Also, how do the Kc values differ between vegetated areas and built-up areas? It's likely that the vegetated areas have higher Kc, so they are using more of the water before it makes it to the stream. Built-up areas are likely to have higher curve number values, so usually have higher quickflow - do you see that in your results?

    To understand the results better, you can use the equations in the User Guide to look at some of the other outputs of the model, such as L_avail.tif and aet.tif and see if they make sense and lead to the results that you're getting.

    ~ Stacie
  • CharChar Member
    edited November 2018
    Hi,Stacie

         The Kc values and CN values used in my analysis were set on the basis of previous studies using water yield model and SCS-CN model, among which the Kc values of vegetated area were much higher than built-up areas.The quickflow map looked much more understandable with higher quickflow in built-areas and cropland.The values of aet.tif also seemed to be okay,with higher values in vegetated areas and waterbody.But the values of L_avail.tif varied slightly among different land use types and across the study area ,the mean value of which was 7.1mm.If this indicator could be interpreted as water retention service from the calculation equations,it was very low in terms of the natural conditions of the study area.The mean values of precipitation and potential evapotranspiration are about 500 mm and 1000 mm respectively. The mean values of actual  evapotranspiration is about 400mm using the Kc method.So I'm still confused about the baseflow from the model,hoping that you may give some guidance on further improvements.
    Post edited by Char on
  • PS:I did carefully prepare the meterological data as you and Perrine suggested before.The hydrological soil groups data was from an article in Scientific Data at 250m resolutions.The alpha、beta and gamma were all set to their default values in lack of empirical data.The threshold flow accumulation was detemined by comparing a river map from HydroSHEDS with the stream.tif.
  • swolnyswolny Member, NatCap Staff
    It sounds like you're being very thorough, @Char, that's great. Since I'm not a hydrologist, I'm going to ask @Rafael to help think through the results.

    ~ Stacie
  • Hi,Stacie
         Thank you for your kindness!
         Waiting to hear from him soon.

  • Hi Stacie, 

    I'm also doing a similar model (simultaneous with the SDR model on a different thread) and I'm getting similar results as with @Char particularly on the baseflow. Other items seems fine but for the B.tif, results are high in the built up and barren area. I followed the count numbers from FAO. Had there been an explanation for this from @Rafael apart from the effect of the topography?

    Thanks a lot for all the help! 
  • swolnyswolny Member, NatCap Staff
    Apologies, I just found out that Rafael hasn't been automatically receiving the forum posts. I spoke with him today and he should be commenting soon. It may be that we need to scrutinize the underlying methods, but I'm not a hydrologist so can't really speak to the theory.

    ~ Stacie
  • RafaelRafael Member, NatCap Staff
    Hi @Char

    apologies for my late reply  - I only became aware of that thread now. It is key to check the various parts of your water budget for forested and urban areas to understand if there is a structural problem with the model, or if it is the consequence of parameterization and the biophysical settings of your case study. 

    The local recharge is calculated as RECHARGE=PRECIPITATION-QUICKFLOW-ET. As Stacie pointed out, urban areas will have high quick flow and less ET. Forested areas will have lower quick flow and more ET. Hence, it is possible, that the ET in your forest areas is so high that it actually results in lower recharge than the urban areas. 

    If you are confident that this is not the case, you might increase the CN value for urban areas. As a test I would recommend to look at the total AET values, the total quickflow, and total precipitation (you could get those by summing the rasters in the temporary folder) and analyse the various components of the water balance for various land-use types. 

    So far, slope is not considered for in the calculation of quick flow. However, there is some empirical evidence that the same land-cover on steeper slope should be modelled using a higher curve number [1], you might also try such an approach to consider the steep slopes in your urban areas. 

    I hope this helps a bit and please let me know if you had any further questions. 

    Rafael 

    [1] M. Huang, J. Gallichand, Z. Wang, M. Goulet, A modification to the Soil Conservation Service curve number method for steep slopes in the Loess Plateau of China. Hydrological Processes. 20, 579–589 (2006).


Sign In or Register to comment.