I have a simple question! I ran SWY model successfully. The total water yield should be calculated by the sum of the Qf, B, and L maps, or does the model give total water separately?
Hi @hkarimi -
The model does not calculate total water yield to streams. If you check out the Limitations and Calibration sections of the User Guide, they note that baseflow is intended to be used as a qualitative output, showing areas of high and low baseflow, not quantitative. But the math behind quickflow is fine being used as quantitative.
That said, if you do still choose to quantify total water flow in streams, you would sum quickflow and baseflow. Local recharge is what is infiltrated at each pixel (which could be thought of as potential baseflow), baseflow is what actually arrives at the stream, and is not used by vegetation along the flow path between the pixel and the stream.
Thanks a lot for your reply. Your message is so helpful for me. I did not know that the SWY model does not calculate total water yield! So, generally, if you want to talk about total water yield in an area you sum B and Qf, or only rely on baseflow result?
I know that Reservoir Hydropower Production model (Water Yield model) calculates total annual water yield. Do you think it is better model to explain total water yield in an area using this model rather than SWY?
if you want to talk about total water yield in an area you sum B and Qf, or only rely on baseflow result?
The SWY model is tricky in this way, since quickflow is more appropriately used quantitatively than baseflow is, so it’s not generally recommended to combine them.
And, it depends on what kind of results you need for your analysis. Sometimes it’s most useful to know about quickflow separately from baseflow, since quickflow can contribute to flooding, while baseflow contributes to having water during the dry season. We did do a project where we summed QF and B and used that result, but we also calibrated the total (QF+B) yield against observed data, which helps ensure that the quantitative results do reflect real conditions.
Do you think it is better model to explain total water yield in an area using this model rather than SWY?
I feel like I should have a more definitive answer for you, but again, it depends. The Annual Water Yield model is even more simple than SWY, in that it does not take into account what is happening downslope of each pixel, and its values are only annual average. AWY does not provide any separation between surface and sub-surface flow, the assumption is that the water yield will arrive by some means within a year. For cases where you do not need any seasonal information, or any differentiation between quickflow and baseflow, AWY can be useful as a very simplified assessment of which places on the landscape are providing water to streams. On the other hand, if seasonality matters, or you do need to consider flooding or dry-season water provision, then you’d need to use SWY.
Once again, I greatly appreciate your assistance and support. I calculated the total water yield by summing the values of QF and B, as you suggested. But another issue with QF and B is that my QF map contains NoDATA pixels, but these NoDATA pixels are absent in the baseflow map (have values of baseflow). This is quite perplexing! My understanding is that if any input map has NoDATA pixels, the output map will also have NoDATA pixels (please correct me if I’m mistaken) and this is why I can see NoDATA pixels in the QF. However, it’s baffling why these NoDATA pixels in the QF do not exist in the baseflow map. Consequently, when I combined B and QF, the result might be not correct!
Hi Hazhir -
Can you please post an image illustrating where you’re seeing NoData in QF and not in baseflow? Do you see any correspondence between any of your inputs, land cover classes, streams, etc?
Thank you so much for your reply. Please see the image of model’s output. In the left side you can see forest group map that I used as land use map. In this map the white spots are non forested lands and are not my study area (because I am just focusing on the forested lands of the US), so it does make sense hat I see those areas as NoData in the Qf map (right above). But why the baseflow map has value for those area? This is because Qf equation uses Kc based on land use map but L and B do not use that? thank you.
Hi Hazhir -
It is also surprising to me that there is baseflow data where quickflow has values of NoData. Can someone in the software team look at why this might be? Is it possible that NoData values in QF are being treated like a value of zero?
Hazhir, I would still not recommend using the model in this way, without data for non-forest pixels. The model will not accurately represent reality, especially when the non-forest area covers such a large portion of the study area, and undoubtedly contributes to both water yield and ET.
Thank you for your reply. Yes. it is also odd to me that why Qf does not have those value, but baseflow has.
You are right about non pixel forests but my study area is only forested lands and this is why I introduced forest map group as land cover map. Do you mean that having no value pixels from land cover map might make error for the outputs?
@swolny yes, from looking at the code it seems that QF nodata values are treated as 0 when calculating AET (and then L and baseflow). It looks like this was done intentionally and it has been this way for at least 4 years.
Thanks for your message. What can I do to tackle this problem?
Hazhir, my suggestion is to include the non-forest pixels in your modeling. After the model is run, you can focus on the results for forests, through map symbology, or by summing results within the forest LULC types.
If you do not include the non-forest land cover pixels, you are missing their effects on ET and ultimately water flow, because the landscape is connected. Baseflow in particular is dependent on not only what happens on individual pixels of forest, but what happens on pixels upstream (which may be lots of non-forest). Ultimately, the location of urban, agriculture, etc, will affect the results that you get for forest. So if you do not include non-forest pixels, you are not accurately representing what’s actually going on in the landscape, you’re basically ignoring the effects of urban, agriculture, etc, which may be significant.
Thank you so much for your valuable advice. I will do that.