SNTEMP (In)Frequently Asked Questions: Geometry Issues

Q9. Should diversion return flows ever be treated as point sources? (i.e.: situations where water is diverted, used for agricultural purposes, and returned to the stream with reduced volume and higher temperature than that of the receiving stream).

A9. Sure. You must define significant of course in terms of the heat added to the stream. You can model them as return flows (R-nodes) or point sources (P-nodes) with the former offering more options on simulating temperature. Typically, the R node for overland irrigation returns would be modeled as an equilibrium temperature, and if in drain tiles as ground water temperature. [Added 12/2001]

Q19. We have a large number of potential validation nodes. I am thinking of limiting it to 5 or 6. But even so, managing the hydrology files seems daunting. For every year, that will be 52 lines at every H, T, J, B, V, and E node. So 20 nodes, 52 weeks, 2 years = 2000 lines of data. Or is there some way to streamline this?

A19. I’m confused. You said validation, but mentioned hydrology nodes. You will need flow at all hydrology nodes and temperatures at H and V only. But, yes, it’s a lot of data no matter how you cut it. Usually, an outside program or spreadsheet is really helpful. Do you need all 52 weeks?

Q94. If I change my tributary inputs to point sources and insert P-nodes where they join the mainstem - do I need to put B and J nodes, before and after them, respectively?

A94. No, you do not. [Added 12/2001]

Q116a. I am confused about what to plot against discharge to determine the width's A-term. The SSTEMP V.1.0.0 manual talks about "wetted width" in the first sentence, but uses the term "width" in the rest of the paragraph and the graph.

A116a. Heat flux takes place at the air-water interface, so what SSTEMP "wants" is the width of that interface. To me, this is the same as "wetted width" and would be the equivalent of a measure from the left water's edge to the right, subtracting any islands. [Added 12/2001]

Q116b. If it is wetted width, does wetted width equal wetted perimeter?

A116b. No, wetted perimeter is the complete length that water is in contact with the bed, such that wetted perimeter is always greater than wetted width in a full U-shaped channel, but approximately equal in a wide shallow stream. [Added 12/2001]

Q116c. How is the width value in the Intermediate Values box calculated? Is this mean wetted width (wetted perimeter) or the mean width from LEW to REW?

A116c. The value that shows up in the Intermediate Values box is calculated from the equation W = a Q ^ b just as you might expect. (Note that Q as used in this equation is the average of the segment inflow and outflow, i.e., approximately half way through the segment.) Width is as I defined it earlier. [Added 12/2001]

Q116d. In many situations, we do not have more than one discharge measurement per stream segment to be modeled. I wanted to verify that it is possible (but probably not preferred) to solve for the A-term if you have a width value at one particular flow and use a B-value from Leopold et al. (1964).

A116d. Yes. I believe I suggested a value of 0.2 as a B-term if nothing else were available. If you can compare your stream to one in Leopold (and I cannot remember how many or how well he describes them) you may be able to do better. Alternately, if you are willing to assume a constant width, just set B to zero and go from there. Any shortcut, however, will likely reduce the accuracy of the model for handling a range of flows since width is often such a sensitive variable, but then it is only a model. [Added 6/2002]

Q117. I have been asked to look into SNTEMP for a proposed hydro on a fairly large river in B.C. Base flow is around 800 cfs. The tricky part is that the proposal is to build a weir at the diversion point. So, pre- and post-project differ in terms of flow regime, and in terms of hydraulics. Pre-project is easy enough. But what are the caveats of post-project? As long as the area behind the weir acts more like a river than a reservoir, can SNTEMP still be used? I can put in good estimates of the change in width, but what else might I need?

A117. If you can assume that the river doesn't stratify (and at that flow I'm certain it doesn't) pretty much all you have to do is adjust the elevations and widths. The width is really the important factor, though elevation can occasionally be important. Be careful, though, not to try to simulate the actual drop at the weir as the model will boil the water since the friction flux doesn't have an upper limit and it goes berserk. Just increase the elevation at the weir site and let the river flow down to the next unchanged geometry node. The operable assumption then becomes that you just have a flatter, wider river with no change in temperature attributable to storage, just normal heat flux. [Added 6/2002]

Q118. Has anyone tried to apply SSTEMP to small flow-thru ponds, i.e., assume it's a widening of the stream? Is there another software (simple) that you know of that would be more appropriate. These are less than 0.5 ha with only approximately 1 cfs flow through.

A118. Yes, there have been many applications similar to what you describe, but I'd have to be honest and say that I don't know how well they have worked. The problem for a situation like you describe is stratification. If you have a small flow coming into a "big" "deep" pond, you are likely to have stratification such that the outflow is simply at or near equilibrium temperature. Both SNTEMP and SSTEMP believe that water in a stream channel is fully mixed. When this is not true, the model is inconsistent with reality. SNTEMP has a feature to simulate equilibrium release from a reservoir, and this might work for you. I'm not sure. I am unaware of any other software for your problem, other than regression models. [Added 6/2002]

Q119. I've prepared a SNTEMP data set that inserts an M node and also has a Q node. Unless I have done something wrong, SNTEMP appears to calculate a single lateral temperature that corresponds only to the first upper reach. It is 18.36° that is the adiabatic correction temperature Mean Annual Air Temperature (MAAT) for the 1st reach. The lower reaches use that temperature even though the data is there for calculating new ones (warmer since it is lower elevation.) Also, note that the M-Node location (at distance 3.0) has a MAAT of 19.37°. This seems to be the only recalculated MAAT.

A119. We looked into this issue and concluded that indeed SNTEMP fails to calculate representative lateral flow temperatures in cases where there is an insufficient density of hydrology nodes. It appears that lateral flows and temperatures are calculated only between pairs of hydrology nodes and then used for almost all intervening nodes of another type (e.g., C and O). [Added 6/2002]

Even with this scheme, there also appear to be two other design issues. Although the program does recompute values for inserted M-nodes, it doesn't treat that M-node as a true hydrology node and therefore it doesn't carry that lateral temperature downstream. Further, the E-node does not appear to have full hydrology node status, because the program does not seem to recalculate the Q-E lateral temperature correctly and carry that downstream.

Although I do consider most of these issues flaws at a minimum and bugs at a maximum, they do not appear to warrant the investment in changing them at this time. I say this because I so rarely see networks in SNTEMP that do not have a much higher density of hydrology nodes, which basically nullifies the problem. C-nodes, while common, are usually broken up by hydrology nodes. O-nodes are relatively uncommon. M-nodes are even less common. I think the best for us at this time is to note these issues and put an item in the FAQ about them.

Q196. In the IF 312 class p. 93 (explaining stream geometry file) it refers to the input of "ground temperature", which defaults to mean annual air temperature, and then on p.107 (hydrology data file) it refers to the input of "lateral inflow temperature", which also defaults to mean annual air temperature. Which one represents groundwater temperatures?

For whatever reason I assumed that this "lateral inflow temperature" in the hydrology file represented groundwater temperature, and therefore input my groundwater temperatures there, leaving ground temperature blank in the geometry file. Was this a mistake? Should it have gone into the ground temperature input in the stream geometry file? I am not sure how I missed this important definition, but better now then later...I guess.

A196. Your assumption was correct. The lateral flow temperature is that of the accretions and is usually far more sensitive than the ground temperature itself. The ground temperature does affect the conduction into or out of the streambed, and is usually not a very sensitive item.

Q197. You told me awhile back that SNTEMP performs gradual accretions of discharges between nodes, that is, if a downstream node has a higher discharge then the node above it, the model will gradually increase discharge as it moves downstream.

Great. Now, is this the same for channel width and shade? If I increase/decrease shade/width at a given node, will the model gradually change the value at the above node as it moves downstream as it does with discharges?

The reason I ask is I have made some changes in width and shade at a given node in my system, however the predicted water temperatures don't start to show a decrease until after this node (see attached Excel graph for example). I would expect the model to start showing slight decreases in water temperature as it gets closer to the node where the new shade/width values have been input. I have attached an example, note that in this example I decreased the width and increased shade starting at river mile 10.2 (a junction node) and at river mile 1.4 (C and V node). As you can see, the model does not show a change in water temperatures until after river mile 10.2. Do you think this may have something to do with the fact that the changes are coming at a junction node?

A197. No and yes are the right answers. No, changes to shade at one node do not take place until that node is reached. Changes to shade are discontinuous and apply to the reach downstream from that node to the next geometry or C node. This is also true if you are using a constant width; it too is discontinuous just like shade.

However, the situation is slightly more complicated if you have width as a function of discharge. As you mentioned, the lateral accretions apply to gaining reaches. The model then can, and does, calculate the "average" discharge at the midpoint of the reach and therefore the "average" width from that average discharge. So in a sense the width changes, but it is not a function of the width at the next downstream node.

So the general rule is that all hydrology and geometry values are discontinuous and apply from the given node downstream to the next change in hydrology or geometry, or both. This is even true for the lateral accretions in the sense that the accretion rate (cms/km) is uniform just like the shade or (constant) width. There is one minor exception (sort of) and that is the reach elevation. The reach elevation in the heat flux model is considered to be the average of the upstream and downstream elevations.

Now, what does all this imply? Each reach should be considered homogeneous with respect to hydrology and geometry (and meteorology too for that matter). It is the old lumper/splitter dilemma. How finely do you divide the study area? There is no final answer to that question, though it seems that most applications end up with about 30 reaches no matter how large the study area is. I believe, but cannot prove, that the model tends to do better with a finer resolution, but that may simply be because those who split usually do a somewhat more conscientious job of measurement. There are plenty of exceptions either way.

Q198. I have a question regarding the stream network (node assignment) I am developing for the hydroelectric project SNTEMP model run. A component of the system is a forebay, a lake that is not on a stream but created from waters impounded by return flows that are then diverted into another penstock to the powerhouse. How do I incorporate the forebay into the model? The skeleton network does not seem to address this type of situation. We have temperature measurements for the forebay (lake) at the two diversion outlets and the intake. Thank you for your time.

Later - Thank you for the reply and information. I have attached a schematic of the system illustrating the network and locations for temperature loggers and weather stations.

There are diversions (aqueducts) off of two different streams (one tributary to the other) that are returned in the forebay. Minimal flow is maintained in the bypassed reaches of the streams. Impounded waters in the forebay are not returned to the main channel until after passing through the powerhouse. I suppose I could put D nodes on the streams at the initial diversion points and then a P node at the project tailrace. Although I am still crunching data, I am guessing that the water temperature in the forebay (fairly small and shallow and exposed to sunlight) would be different than that in the aqueducts, which may affect water temperature where the tailrace returns flow to the main channel. Could this point be ignored and/or would identifying the forebay as a tributary be accurate? Please advise.

A198. As always, it depends. If you can consider the forebay as a tributary, then you can put it in that way, starting with an S node. If however, you mean that the "return flows" result from a previous off-steam diversion, then it gets more complicated -- unless the bypass reach will remain completely dry. Assuming the worst case for a moment, you will likely need to use a workaround we built in to SNTEMP described in the 312 class notes (I think?), but also in the attached file. This lets you take water out in one place and put it back in another BUT does not allow heat flux in the process. In other words, SNTEMP is only a dendritic model, not a true network model. Fortunately, ignoring the heat flux in small conveyance systems is usually not a problem.
Later - To some degree, what you do will depend on what you know about the existing temperatures in the system. It looks like you have a reasonably good density of monitoring stations. I am sure that temperatures do change in a forebay; you will be able to tell that. That process cannot be explicitly modeled in SNTEMP regardless of schematic layout except as a 'wide spot' in the stream. Since it is small and shallow, it may not influence temperatures too much. People have told me (I have no direct experience) that they have been able to get by with ignoring heat flux in penstocks and canals (so I assume your aqueduct too). Where do you need the most accurate temperature predictions? In the bypass reach? Only further downstream, perhaps at T1?
I'm going to assume your focus is on the bypass reach. If true, why really bother with the penstock/aqueduct route at all? The decision variable will be discharge in the bypass and you will have a tributary coming from the small lake and validation points at T5 and T3 (if that is actually above the re-entry point). If you do need fully mixed temperatures for hatchery input or something, then you probably either need to ignore heat flux in the penstock/forebay/aqueduct or (I shudder) need to run two almost parallel models, manually transcribing diversion temperatures from one model to the other and then manually mixing the results. It would be a mess, but doable.

Q200. If a change in stream azimuth is the only parameter along a sub-reach that is variable (shade, gradient, Manning's n etc are constant) is it worth putting in a change node to account for this? Or is there some other way to manage this?

A200. This depends on whether you are a lumper or a splitter, and of course whether shade really makes much difference in this model. It is not much trouble to put in an extra C node, so why not?

Q201. I started creating some of my input files for SNTEMP and came upon an issue I think may pose an error when executing the model. I have several temperature probes in the study stream that are going to act as verification points (V nodes). They were strategically located at major stream shade or discharge transition points. Is it permissible to have different nodes in the input files with the exact same stream distance? In other words, the Study File will have O nodes that are exactly where I would like to have C nodes in my Stream Geometry File, as well as where the V nodes will be in my Hydrology File. Is this going to cause a problem when executing the model?

A201. There are occasions when SNTEMP may complain about co-location. My recollection is that they are rare, but I have always suggested that folks separate their nodes by at least a small fraction (e.g., 0.1 or even 0.01) to keep themselves straight if nothing else. The data check program, TDATCHK, does check for co-location and suggests a change if it finds any instances. But really, you don't have to do this if you don't want to unless SNTEMP itself complains.

Just as a note, there is not any real good reason to have an O-node co-located with any other node since you can always get output at any node you choose. Typically you would use O-nodes when there is not any other way to get output at a particular location.

Q202. Have a question about constructing the skeleton file. Does the value for the “distance from the system endpoint” parameter for an “end node” (E) have to be set to 0.0 or can it be set to a river km distance obtained from a USGS quad map? The example in Table I.3 for the Tucannon River in “Instream Flow Information Paper: No. 16” has the distance to the E node set to 0.0 but the example hydrology node file that came with SNTEMP has the distance to the E node set to 399.0 (for the X. River confluence).

A202. No, there is nothing magic about the E node being 0.0, though I am not sure it could be negative – I’ve never tried that. I recommend a precision of 0.1 km in most cases.