Baseflow Modeling Review

[rburghol]

Comparison: Evidence of varying AGWRC in observed versus fixed AGWRC in model.

Observed	Modeled

[julieshortridge]

Still trying to wrap my head around these - I've mainly been trying to think about physical reasons that would explain the behavior. I have a couple questions to make sure I'm remembering the details correctly and thinking through this in a logical manner:

1. Appears to me that the discrepancy between gage versus model behavior is apparent in Strasburg and Mt. Jackson (larger watershed), but not Cootes Store (smaller). I think on Tuesday you also said that Cootes Store is a single land segment - is that right?
2. One possible reason for a varying AGWRC that we've talked about is differences in contributing areas at different flow levels. For example, as the watershed goes without rain, a greater portion of baseflow is released from a lower-k portion of the watershed that releases water more slowly. Or vice versa. So conceivably, this could result in a varying AGWRC-flow relationship in either direction (e.g., AGWRC increases OR decreases with flow), depending on soil conditions.
3. Another possible reason is small, isolated rainfall events that aren't enough to "bump" the gage up to something we'd identify as stormflow, but still leads to higher flow values. This would result in a slower rate of decline, and a higher AGWRC value.
4. Another reason would be unaccounted for ET from active groundwater. This one seems more complicated to me in terms of how it would impact AGWRC. On the one hand, you might expect a more rapid decrease in flow values (and lower AGWRC) as water is "lost" to ET. On the other, declining AGW storage results in slower declines in flow values and a higher AGWRC. I think we'd maybe need to play around with some equations to think through this possibility.

Does that logic track to you? Please correct me if I'm wrong somewhere!

[rburghol]

Hey @julieshortridge thanks for starting this dialogue. Note, in my responses below I am trying to shift to speak in terms of AGWRC varying as a function of Storage, AGWS (or $S_{AGW}$) as I've proposed here in #47 -- and I will do this below, but will note that perhaps there are differences that I am not grasping when doing so -- My responses:

1. Coote's is definitely the less complex model, maybe 80% composed of a single land seg? And YES, I agree the USGS flows suggest a non-varying AGWRC as a function of $Q_{out}$ (and thus $S_{AGW}$) in Coote's store, but NOT in MJ and Strasburg.
2. Varying contributing areas: I agree that this can be true in the real world, AND also in the model. However, the model is not showing this behavior in MJ/S despite having a varying range of contributing land segs, with a reasonably large range of $C_{AGW}$ -- so the model should show this, even if in a muted fashion, but it is not. And the real world is showing this, and the question of why the model is not showing this despite it being a theoretical possibility is vexing to me. Which makes me think that perhaps the model doesn't make this happen :), or perhaps there are other parameterizations that obscure the effect, like maybe $ET_{AGW}$
3. Rainfall can be confounding Yes, definitely agree here. Though I think it could act in both directions, that is, small sequential rain events could appear to our methods as baseflow recession, but in reality they are not. But maybe it is more likely to behave as you propose here. And I'd say that since our $C_{AGW}$ estimation method is describing the real world and model world $C_{AGW}$ as smaller than that which is in the model parameter, this could be a contributor to that.
4. Another reason would be unaccounted for ET from active groundwater. I'm going to go point by point on this one as I agree and disagree :)
   • On the one hand, you might expect a more rapid decrease in flow values (and lower AGWRC) as water is "lost" to ET - agree, and I will add that this will only be relevant when $ET_{AGW}$ is large relative to $S_{AGW}$
   • On the other, declining AGW storage results in slower declines in flow values and a higher AGWRC -- I don't agree with this one, because any extra removal of water from $S_{AGW}$ always results in tomorrows $Q_{AGW}$ being smaller than it otherwise would have been, which always results in the appearance of a smaller $C_{AGW}$ -- but when $S_{AGW}$ is large relative to $ET_{AGW}$ the effect on $S_{AGW}$ and therefore $Q_{AGW}$ can be negligible.