Q5. I am puzzled by negative values for total heat flux, while system still gains heat, very counterintuitive. I’ve attached the SSTEMP files, and a QPRO v5.0 (*WB1) file with data and graphs on first sheet in the hopes that you may help in this quandary. I appreciate you willingness to look into this.
A5. None as yet
Q6. I graphed out the new flux values, and they are much more appealing as a way of portraying the actual physics of stream heating. But I am bothered by two aspects of the curve. Inspection of total flux values shows no change between the last two stations, while temperature rises at a rate close to previous two stations, also should not the last total heat flux value be close to zero since the actual temp is approaching very close to the equilibrium temp? Are both of these an artifact of using the upstream station values to portray the flux values between two stations? If so then it might be more appealing to portray the end station
A6 None as yet
Q7 None as yet
A7. 1. Flux as J/m2/s is in my opinion the best standard, as other people will have a multitude of circumstances to interpret. Given specific widths, distances, and travel times, the flux rate as presented can be used to calculate total heat gain or whatever, as one desires.
2. Actual flux rates at the start, mean, or end of a segment all have useful applications, and I could imagine arguments in favor of all. I realize that I am preferring to examine temp and flux along a series of segments downstream from the watershed divide, and that not everyone who uses your program has that goal in mind. Having said that…
Presenting End of Segment Rates Only
Pro: 1) corresponds with model output for end-of-segment predicted and equilibrium, temp. 2) allows corresponding value for my example to see flux go to zero when actual approaches equilibrium temp. 3) end of rate of previous segment could be substituted for starting rate of next segment (except for first segment!)
Con: 1) end rate alone for a single segment modeling effort, without context of starting rate.
Presenting Start of Segment Rates Only
Pro: 1) represents better than end rate, forces driving heat flux in segment. 2) rate in next downstream segment could be substituted for ending rate in an upstream segment (except for last segment!).
Con: 1) starting rate for a single segment modeling effort, without context of end rate
Given the above garble of logic, I see the only way to satisfy all situations would be to present BOTH the starting and ending heat flux rates for a segment. I realize this would complicate your output screen, perhaps beyond practicality, or your original intentions for the model. Also I am but one of your users and most may not have the same wishes for the output as I do. [Added 12/2001]
Q8. The individual who is interested in this was, as expected, more interested in the relative values of the fluxes that are heating the water "toward equilibrium" rather than "at equilibrium." Which would be more correct (recognizing that none would be exact, I suppose)? (1) Would it be better to plug in the initial flux equations, or (2) plug in the final downstream temperature, or (3) simply say that the equilibrium fluxes would be in about the right proportions as the forcing functions on the water in the river? Or may you can think of another way entirely.
Assuming that option 3 is a reasonable starting point, is it correct to assume that "ignoring" the water’s back radiation, you get a good idea from the other flux components what is happening to the water on the ground? I.E., the relative magnitudes of the remaining fluxes give a reasonable idea of how rapidly temperatures are approaching equilibrium? Really the k1 parameter I guess?
A8. My recommendation still stands. I prefer the equilibrium water temperature as the best indicator of the relative importance of the "driving forces". Yes, ignore the water’s back radiation for this purpose because it is the basic dependent heat flux; i.e., it is the major heat flux component that adjusts to balance the net heat flux to zero. And, yes, the first-order thermal exchange coefficient (K1 parameter) does indicate the rate at which the water temperature approaches equilibrium. [Added 12/2001]
Q66. Someone asked me a question, which I found intriguing: What is the role of turbidity? I have heard that turbid streams trap and retain heat to a greater degree than clear streams. Does this make a big difference or is it minor?
A66. The question about turbidity comes up now and again. Try as I might, I have found nothing definitive about the role of turbidity and heating in streams or lakes. There remains a lot of speculation, but little or nothing about actual measurements or even real theory as far as I can tell. The specific heat of water is so much greater than soil or rock that I can’t imagine turbidity affecting the heat "content" of water. And it should have no effect on effect reflectivity. However, turbidity does affect light penetration of course and therefore may affect absorption. It also may play a role in re-radiation. [Added 12/2001]
Q69. Since I am doing a TMDL, the damn lawyers want me to include a table with the work "load" on it. In past work, I have used SSTEMP, which gives output of various temperature heat fluxes in J/m2/s. I have taken the solar load from this output to make the lawyers happy. Now with SNTEMP, I can’t find the same result data in the output tables. The closest I come is the solar radiation in table VI, but I think this is the solar input prior to the model using the shade input values.
Am I missing something? Is there a way to get the solar radiation load as affected by the shade from the SNTEMP output files?
A69. Sorry to disappoint you, but there is no current way to get the same info out of SNTEMP. The Fortran code could be modified to print this out of course, but it would not be trivial and I’m not sure it’s worth it. Probably the most straightforward way to get that would be to take the parameters from your calibrated SNTEMP model and plug them back into SSTEMP. With some care, you will get the same answers. But you have to carefully get all the right stuff out of SNTEMP, e.g., elevation corrected air temp, etc. Sort of full circle, right?
Q120. I am using SSTEMP to model the discharge from a sedimentation pond into a small creek. I am considering a range of discharge rates, and am noticing that the temperature will often increase along the stream. It seems that the friction and vegetative radiation terms are rather high. I was wondering if this is common, or is there a glaring error. I have attached one of the runs as an illustration.
A120. You are correct to question the vegetative and friction heat flux components. I believe the friction component is far too high for your high gradient (25 ft/100 ft). I have noticed this problem once before in trying to simulate a high gradient headwater. What I ended up doing was widening the stream. This seemed fair to me at the time because what the model really "wants" is the width of the air-water interface. In my situation, the stream was really a series of rivulets and cascades that, effectively, widened the stream and mitigated the friction component of heat flux. In your case, assuming you have the gradient correctly described, I feel that the friction component is just too high. I don't think Theurer ever envisioned simulating such a high gradient stream and the model component for friction is likely outside its "theoretical" domain. I would simply lower the elevation of the upstream boundary. Now the question becomes "How much?" I really cannot answer that, but you should feel free to experiment and reach your own level of comfort. Then document what you have done so that others will understand.
Now, vegetation. Hum. You have 95 percent shade. This is definitely on the high end of what people typically simulate, but I suppose not out of the question. If you really have quite an effective canopy, the water's back radiation will indeed be trapped under that canopy and the vegetation will re-radiate the trapped heat quite effectively. I wouldn't say that this component is beyond question, but I would say that I have no reason to question the model in this area. Rather, I would just ask you to verify that the shade is indeed 95 percent using a light meter (in and out of shade) or the gray card technique mentioned in the help file. 95 percent shade would be really dark.
On the other hand, I don't think any changes to your model will be very sensitive for your short reach. What is happening in your stream is dominated by the release temperature of your sediment pond. Little else has an effect as you can tell if you run SSTEMP's sensitivity analysis. SSTEMP cannot, of course, simulate what is going on in that pond due to stratification, etc., so I just hope you have that end pretty well pinned down. [Added 6/2002]
Q203. In my continuing attempt to understand thermal modeling at a deeper level, I was reviewing the heat flux equations in SNTEMP and CE-QUAL-W2 and have come upon an issue about evaporative heat flux that disturbed me. Could you take a look at the little white paper I wrote and the attached spreadsheet and let me know if I'm way off here.
Basically, I do not understand the Stream Evaporation equations given in Theurer, et al. and find they give strange results. If you agree with my analysis (and the new formula) then a larger problem is what effect this may have on Theurer's solution to the overall Heat Flux equation. But one step at a time...
Followup1 - Thanks for your very thoughtful response to my question. I really appreciate having the literature you cited...I will look into these other modeling treatments. I have to admit that I do remain confused about the "theoretically correct" representation of evaporation.
The reading I have done from "basic physics" texts indicates that evaporation cools the water body (by releasing the latent heat of vaporization into the gas phase) and that condensation also heats the air (the latent heat of condensation is released into the air as the molecules slow down to liquid rates--this heat release is so important in weather that they talk about how it drives thunderstorms). It seems that the only possible heat exchange upon condensation would be if the droplets coming out of the air were at a lower temperature than the water (which is possible since this usually occurs at night when air temperatures are low). I have no idea if this is a realistic mechanism or not, if it is significant, and whether the equations I sent (from Theurer) are supposed to capture that type of mechanism. Also, I was under the impression that net rates of condensation only occurred when relative humidity was at 100%, not just high (like the 75% in the spreadsheet I sent to you).
It will be interesting to hear what Theurer has to say and to read some of the literature you referenced. In talking to Tom Cole he indicated that evaporative cooling is a major heat exchange mechanism. So again, I am confused about how important or critical this is to the overall modeling of thermal processes in lakes and streams. Obviously, SNTEMP has proven itself over the years with (I'm guessing 1,000's of applications) and any "error" associated with the representation of evaporation is minor (in comparison with other aspect of the model; e.g., the steady flow assumption).
I hope I have not entered into one of those problems that resemble the "how many angels can dance on the head of a pin" question. Thanks again for your thoughtful and very helpful responses.
Followup2 - Thanks for looking into it for me. Theurer’s response indicates that it may be one of those issues that could use refinement, but no one has gone there yet. I suppose it could be due to the small role that evaporative cooling plays in stream environments...a place for future research perhaps.
A203. I am not an expert on the "guts" of SNTEMP, or any temperature model for that matter, but I read your note with interest. I can say that I have been occasionally bothered by what seems to happen near 100% humidities, but at first glance anyway, I am not so bothered by the changing sign on evaporative heat flux depending on the relative values of air and water temperature. In one state, the flux would represent evaporation, in the other, condensation.
In any event, I scanned some literature I had for alternate models just to see what other people have been using. Here's what I found:
As best as I can tell, DeWalle (1976), Jorgensen and Gromiec (1989), Webb and Zhang (1997), Jobson (1981), Beschta (1984), Sinokrot and Stefan (1993), and Adams and Sullivan (undated and unpublished) use a form similar (perhaps identical when all is said and done with the various units and how they were applied) to the Cole and Wells formulation you cite. The later comment that they know this linearized and algebraic formulation contains error, but the errors have only a small impact because the temperature of the water drives the calculation and cannot change too much because of that. Their logic is not completely transparent, but ... Oregon DEQ (1999) also uses this same general form, but offers a wide choice of alternative formulations for the user. Further, they note that evaporation is very important, but "for practical modeling applications the sensible heat term is negligible."
Mohseni and Stephan (1999) are way above my skill level, but seem to use a formulation that bears little resemblance to anything else.
Crittenden (1978) uses what appears to be yet a different formulation, applying the formula:
Qev = 0.008804 U(ea - 4.76 e(0.0645 Tw))
Even though most of the papers I looked at seem to use the Cole and Wells formulation (or some very similar version, Fred often solved and transformed the equations he worked with to be solved as efficiently as possible. That was a boon in the days of a hand calculator, but a pain to others in reading that code today. So, bottom line, I am cc-ing Theurer on this to see if he can dredge up the source for his evaporative heat flux equations.
Follow-up - Soon you will be the expert. Evaporation is definitely important, especially in the arid west, but seemingly far less important than the short and long wave radiation components (at least according to SSTEMP). I have not heard back from Fred on this or another query, so I'm not sure what gives, but I will bug him again when I get the chance. You could always recompile SNTEMP and experiment with different formulations. I do know that I often find, and have given advice, that I need to increase relative humidity by a multiplicative factor of say 1.25 or so (capped at a maximum of 1 obviously) to get a decent fit.
Response from Fred Theurer - I thought your response was appropriate. I suspect that an order of magnitude analysis for a relative humidity near 100% would show that the effect is negligible and within the order of predicted errors. The input errors are probably much more significant. However, if the evaporation feature can be easily improved by others, then encourage them to do so--and let me know how it was done so I can use it. It is always better to use more precise science as long as it is not too complex if for no other reason than not having to explain why less precise is still OK.
Q204. I am using thermal gradient as a calibration parameter but I am not sure what a reasonable range of values would be. Most of the reports I have seen simply use the default 1.65. The higher the value I use, the closer my model output temperatures are to my actual temperatures. What is a reasonable range of thermal gradient?
A204. As a general rule. I do not recommend using the thermal gradient as a calibration parameter. Although it is true that the thermal gradient remains largely an "unknown" or at least poorly measured parameter, it is also not a particularly sensitive one in my experience. It would be likely that calibration would drive the thermal gradient outside the bounds of values reported in the Forsythe, W.E. 1954. Smithsonian physical tables. Smithsonian Miscellaneous Collections Volume 120. Publication 4169. Smithsonian Institution, Washington, DC. (At least I think this is where you could find a range of values -- it's been a while since I looked.)
Even not knowing anything about your situation, I would be far more likely to recommend using shade, wind speed, or relative humidity as calibration parameters -- shade because it is often poorly measured and contains subjective elements, wind speed and relative humidity because they are often measured at sites distant from the stream's surface. In the case of SSTEMP, in particular, unless you have corrected relative humidity for elevation there is likely a bias. Other parameters may also be candidates for calibration. In fact, you may have a very good reason for using thermal gradient as a calibration parameter, but be sure to document why this is.
As a final caution, I find that many users of SSTEMP also tend to "over calibrate" their models. By this I mean using a small data set (typically one or a few days) and then expecting the model to be robust to large changes in flow. This can be misleading at best. Please refer to Q76 in the write-up at http://www.fort.usgs.gov/products/Publications/4037/faq_calibration.asp. You might review some of this material just to see if it helps. SSTEMP is best for learning the system. SNTEMP is best for "real" applications.
Q205. I would like to ask you about ground water impact to stream water temperature modeling or on the river. Based on your statement about "Factors That Influence Water Temperature", you said that air temperature is the most influence to water temperature.
Could you more explain about it or could you lead me to some your papers or other papers (journal) which talk about factors that influence water temperature especially from ground water.
A205. Thank you for your excellent question. This is a question that arises frequently and has been the subject of much confusion. Here is a synopsis of what I believe to be the case:
The real answer to your question will always be "It depends on the situation!" What most people are referring to is the factor or factors which govern stream equilibrium temperatures, usually downstream of some perturbation (diversion, reservoir release, timber harvest, etc.). From my experience based on the models that I have used, air temperature is the variable most closely associated with downstream mean daily water temperatures. Much of the controversy, if that's the right name, stems from distinguishing the factors most closely associated with mean daily water temperatures from those associated with maximum daily water temperatures. In the case of maximum daily temperatures, solar radiation (coupled with shading effects) plays a significant role, especially in small, shallow, unshaded streams. And if the question really is, as I suspect it should be, what factors most closely determine the difference between mean daily and maximum daily water temperature, then insolation is a prime candidate. This should be typically true given that the two main components of heat flux in a free flowing stream are short and long wave radiation.
Now, let's talk for a moment about when air temperature and solar radiation are not the most important variables. One way to characterize these situations is when there are significant discontinuities in "normal" stream temperatures. For example, immediately below a hypolimnetic release reservoir, stream temperatures are, of course, governed exclusively by the release temperature. Near and within so-called thermal refugia associated with springs, seeps, stratified pools, tributary inflows, and the like, stream temperatures are controlled or largely influenced by those advective fluxes. This is where ground water inputs can be very important in determining in-channel water temperatures. Obviously, the importance would be directly related to the volume of groundwater accretions relative to the in-channel flows.
In addition, the whole area of hyporheic flows is one that needs more investigation, and one that is not well supported by existing models. If a large percentage of stream flow is actually percolating through gravels below the visual surface of the stream, the effects can clearly serve to dampen the deviation of maximum and minimum in-channel temperatures from the mean daily value.
A sample sensitivity analysis for a sample stream may be found at http://www.krisweb.com/krisweb_kt/biblio/general/usfws/bartholo.pdf
Other resources -- see Q70 at http://www.fort.usgs.gov/products/Publications/4037/faq_other.asp
You could also download the SSTEMP program from our website and get some idea of the relative heat flux values from each component of a model you could construct from your own data.
Q206. I've been asked a couple questions about the model by my adviser's that I can't really answer dealing with the evaporation and convection equations.
For evaporation, two equations are given in Paper 16, the lake type and flow type model. I believe that in the course material you say that if the evaporation factors in the job control file are left at zero, then the defaults are 40, 15, and 1.0640 which would mean that the model is using the flow-type model. Is that correct? Now, I left those factors as the defaults in my model and did not change them in my calibration, this is what my adviser is questioning me about. Should I pay more attention to these factors? I will add these to my sensitivity analysis but wanted to hear what your opinion was.
For convection, again there is a lake type and flow type equation listed in Paper 16, but there is also equation II(72), which just applies the Bowen ratio to the evaporative heat flux to get the convective heat flux. Which of these methods is actually used in SNTEMP? I'm now concerned because in my job control file, I left the Bowen Ratio as zero - have I totally screwed up by effectively making the convective heat flux zero?
A206. The model uses the flow type models, II(70) and II(74). The defaults you get if the values are zero for the first four values on record 7 of the job file are:
EFA = 40
EFB = 15
EFC = 0 (don't ask me why)
BowenR = 6.19e-4
I have never recommended that anyone change these values, but then they are certainly fair game. I have never found it worthwhile to change these values.
I should mention that SNTEMP's evaporation flux has recently been called into question. I will forward an e-mail exchange on this issue.
Q207. Have you seen the work of Sherri Johnson out of Corvallis? Attached are her papers that assert that air temperature is a minor influence on stream temperature and that it is all about shade. This is contrary to your findings of 1989. One thing that troubles me is that none of this research on temperature takes place during winter, when we may be getting anchor ice in streams at higher elevation.
A207. Yes, I am familiar with much, but not all, of Sherri's work, and concur that we (all) have lots more to learn about which factors dominate thermal processes in which circumstances -- a safe thing for researchers to say, no? My river temperature expertise, such as it is, is largely confined to modeling, and then really only with one or two specific models developed some time ago that focused on easily available data. I have never been involved in the level of field research that Sherri has undertaken, so will yield to her on that. But I have tried to absorb from the wide variety of published literature and techniques and so I do have a few opinions on what is likely going on in many of the situations that we try to model. With these points in mind, here are a few thoughts.
At the top of the list is that one must always be clear whether you are referring to mean daily, maximum daily, or minimum daily phenomena. I believe that many of the misunderstandings we perceive can be traced back to over-generalizations and miscommunication. It seems likely that (mean daily) air temperature will remain a better predictor of mean daily water temperature than (mean daily) solar radiation in many situations, whereas solar radiation will be a better predictor of maximum daily water temperature. Of course, mean daily water temperature really doesn't mean anything in and of itself; it is (almost) solely a function of the maximum and minimum.
Sherri is certainly right that correlation does not mean causation [among other things she is right about :)] but I do remember working with a fellow in the Bureau of Reclamation several years ago who got curious about this very thing. His study area was a grazing situation in eastern Oregon (I think) with a lot of sun and little shade. He had decent on-site measurements of both mean and maximum air temperature and solar radiation and did a bunch of correlations with various water temperature metrics. His conclusions were that variation in mean daily air temperature consistently "explained" more of the daily variation in mean daily water temperature than did solar radiation. By "explained" I am referring to higher R2 and lower RMSE, if I remember correctly. [His report is presumably available, but unpublished except by Reclamation.] Now, how might this relate to situations in, say, western Oregon where air temperatures may be more consistent from day to day but solar radiation may be more inconsistent, and where stable groundwater inputs may tend to dominate the thermal regime? I could easily imagine that at least some of the correlations (if not causations) might turn out considerably different.
Taking this a bit further, let's imagine a situation where the weather in a specific geographic location is dominated by "fronts" coming off the ocean. Now, it may well be that direct solar radiation is "controlling" the ocean air temperatures that, in part, drive those fronts. [But of course the extraterrestrial solar radiation is essentially a constant, given time of year and cycle of the sun, and therefore the ground-level solar radiation is governed more by cloud cover.] But here comes the weather front. Its air temperatures may well govern the resulting overland water temperatures far more than that day's solar radiation at that location. Now, add shade as a variable. An open stream’s maximum daily water temperature is very likely to be highly sensitive to shade when the afternoon sun is full force.
A second thought that I might throw in for what it is worth is something that Matt Boyd once told me. (He and Fred Theurer had been discussion the relative merits of alternative formulations of the mathematics behind water temperature modeling, and as you may remember, the HeatSource model concentrated on predicting maximum daily temperatures where SNTEMP concentrated on predicting mean daily temperatures.) Anyway, Matt said that in his opinion a lot of the differences between the models were simply based on how air temperature and solar radiation were lumped or split in terms of how their interrelationships on each other and other heat flux components might be expressed mathematically. How or whether this may affect Sherri's "direct" calculation of heat flux, I do not know.
Third, though somewhat off your question, I don't believe we have really solved the "controversy" of whether the aquatic community responds most directly to the mean or the maximum daily water temperature, in between, or neither. I have just been reading some interesting material here in Colorado relative to reworking this state's water temperature criteria for TMDLs, much like what was done in the Pacific Northwest. Some individuals have been raising some interesting questions about the inappropriateness of MWMT as opposed to MWAT-type criteria. My opinion -- it depends and we need more definitive science!
Last, I certainly concur that minimum daily water temperatures remain largely unexplored. I think that we can easily learn more about them, but no one has systematically looked at phenomena that likely control minima. It is clear that conceptualizing daily minima as a symmetric process is, at best, only a crude rule-of-thumb. Minima seem to have interesting stationary signatures that, as Sherri has stated, may often be dominated by groundwater interactions. I'm not sure whether this holds in larger rivers though. In fact, as I implied earlier, I think we have to be very careful about generalizing between small and large streams/rivers, generalizing across large changes in the meteorological landscape, and so on. Measurement errors that are inconsequential in some situations may be quite significant in others. Further, when you think about it, maximum daily water temperature, and by implication mean daily, is not independent of the minimum. Yet another reason that "it depends" may be more correct than just a simple shrug.
Finally, I know that many people have latched on to my 1989 statements about the importance of air temperature in predicting mean daily water temperature. I tried to make it clear at the time that the example I presented was for one hypothetical situation using one model. I hope that what I have said here makes it clear that "it depends" is likely the most correct answer. I do not think that I made it clear enough at the time that I was segregating the parametric effect of ground-level solar radiation, the effect of shade on radiation actually entering the water, and cloud cover. Even if these parameters were combined, however, SNTEMP still would rarely predict that solar radiation was more sensitive than air temperature in controlling water temperature, either mean or maximum daily. But, when flows decline, width-to-depth ratio increases, and air temperatures and wind speeds are relatively low, the sensitivity of solar radiation becomes ever more dominant. In contrast, air temperatures would be expected to be relatively insensitive when flows are relatively high, widths narrow, wind speeds low, and ground water accretions are stable and cold.
In some ways, I actually don't care about some of these distinctions. Except for the obvious implications related to global warming, we have little control over dominant meteorology. What we do have some control over is flow in the river, shading, stream width and length, point and non-point source inputs, hyporheic interactions, and channel roughness. These are where we really need to focus our attention in terms of thermal management. And let me be clear that shading -- especially for small streams -- remains a critical management priority, not only for direct water temperature control, but also for a host of other ecological reasons. As streams get larger, however, the relative influence of shade on water temperatures declines and we must focus on other factors under our control.
I know very little about anchor ice. Sorry. Please see the references in Information Paper 13.
Q208. Have you read or heard of any cases where the temperature of ground water influx into a stream is actually warmer than the stream temperature? I'm thinking of cases where an entire watershed may be clear cut, exposed ground, just buffer strips along the stream. It seems that most people assume that ground water influx would always tend to have a cooling influence on a stream.
A208. There are some classic references on this subject, but not exactly with the slant you may be seeking.
Kaya, C.M., L.R. Kaeding, and D.E. Burkhalter. 1971. Use of a cold-water refuge by rainbow and brown trout in a geothermally heated stream. Prog. Fish-Cult. 39(1):37 39.
is perhaps the most cited that I can remember, but the situation there has to do with thermal springs in Yellowstone.
You might have a look at the references in my brief paper:
Also available on the Internet at http://smig.usgs.gov/SMIG/uo/features_0902/clearcut.html to see if any of these may get at what you seek. I don't recall that they will hit the nail on the head, but then again some might.
Q209. For a water temperature model that I created, I am using the exchange of heat between the soil and the water applying the equation:
GwH = (Ts-Tw)*C*WettedArea/thick
But I didn't find any literature that show any value for the "thick" between the soil and the water, I chose 0.5 meter because I thought that the thick should be a short distance but I have not any paper to support this assumption. Can you give any idea?
A209. SNTEMP assumes 1 meter. Most of my information is old, but the following references may help, some of which are newer:
Q210. You've helped me with some temperature issues in the past and I was hoping to again get your insight. I've been educating myself on the heat flux components affecting stream temperature. In that process, I have come across a number of documents that basically state that air temperature has a relatively minor effect on stream temperatures as compared to solar radiation. However, I noticed in your SSTEMP user's manual it is stated that air temperature is usually the single most important factor in determining mean daily water temperature. I would greatly appreciate your input on these seemingly contradictory conclusions.
A210. Yes, this question keeps coming up. [See elsewhere in this FAQ.] But let me just state as succinctly as I can that there is no doubt in my mind, that for the mean daily models SNTEMP and SSTEMP, that air temperature is almost always the most sensitive input parameter in determining mean daily water temperature. Note that this is a much different statement than one that would say, for example, that the largest contributor to maximum daily temperature is the heat flux associated with direct solar radiation, i.e. solar heat flux dominates all other heat fluxes in determining maximum daily water temperature.
Both statements have exceptions and I cannot 100% reconcile the two, but that's what I have been able to gather. For example, on very cloudy days, or in heavily shaded streams, or both, maximum daily water temperature is not dominated by solar heat flux. Similarly, if you are just interested in mean daily water temperatures just downstream of a dam, the only factor that is really important is the release temperature. Beware categorical statements.
[Updated 5/2007]
U.S. Department of the Interior | U.S. Geological Survey
URL: http://www.fort.usgs.gov/products/Publications/4037/faq_heat.asp
Page Contact Information: AskFORT@usgs.gov
Page Last Modified: 12:14:23 PM
