Snow Melt Modeling

[NeffariasBredd]

Do any of you have tips to better model Rain on Snow Events for far northern climates? I'll be using probably modified rational or SCS Runoff. Thanks!

 

[psmart]

Depending on the nature of the snow, runoff can vary from 0% (fully absorbed) to 100% (impervious ice surface), or more (if the snow is melting). More information is needed to answer your question.


Peter Smart
HydroCAD Software

http://www.hydrocad.net

 

[NeffariasBredd]

This would be an early spring rain event in the UP so assuming frozen/saturated ground cover.

 

[psmart]

If the surface is frozen and does not absorb any rainfall, you could model this with CN=98+ for the SCS/NRCS method, or C=0.95+ for the Rational method. The Tc would presumably depend on the surface texture and "grading". Again, there is wide range of variability, making it very hard to predict. But if you're trying to model the worst case runoff, use a high CN and low Tc. Of course, we're taking about a frozen surface with no "runoff" contribution from snow melt. If you have snow melt it's no longer a strict "runoff" calculation and you're talking about an entirely different procedure.

Peter Smart
HydroCAD Software

http://www.hydrocad.net

 

[miningman]

Whats your definition of far north.?? Baffin Island, Resolute Bay , Michigan or Newcastle UK. Totally different locations with totally different snow packs and temperatures. The first two have close to 100% permafrost, the latter two havent seen perafrost in at least 4000 years.

 

[cvg]

UP = Upper Peninsula = Michigan

 

[NeffariasBredd]

@psmart I'm looking for things beyond just increasing the Curve Number. In areas with long and snowy winter the majority of the precipitation from October - March is stored in the Snowpack throughout those months and is released during snowmelt events in early spring. A rainfall event just adds to the effect. I was wondering if there was a way to model a 10-year storm that had additional runoff due to snowmelt.

 

[miningman]

ah way down south is what you mean

 

[NeffariasBredd]

@miningman Yeah, when you're talking Resolute Bay this would be WAY south. Though I'd be interested to see what it would take to build a hospital that far north.

 

[miningman]

Nothing more than LOTS of $$$$ and an appreciation of the practicalities of permafrost , 24 hour darkness betwen October and February, and slightly difficult logistics because everything has to be brought in with ice breaker assisted ships in August.... the ocean is still frozen solid in July and its snowing again big time mid Sept....or fly it in , but thats no big deal , the runway there can accommodate Hercules.

Teck Cominco built a 2000 ton per day mine about 80 miles out of Resolute in the mid 70s, so a hospital would be no big deal. From memory , right now I think they make do with a nursing station.

 

[civilman72]

Good luck trying to figure out snowmelt. I've tried to get more documentation on this exact topic for over a decade. The Urban Drainage Manual, the manual most of us use in Colorado for H&H generally ignores snowmelt.

The scholars that I have spoken with state it's nearly impossible to try to predict snowmelt flow, mostly due to having too many varying factors. Along with the difficulty of predicting the travel path, travel time, conveyance, etc. under the snowpack there are also a lot of factors that effect the rate of snowmelt:

Current air temperature
Prior air temperatures for last seven days (did it get below freezing at night?)
Ground temperature
Snow pack thickness
Moisture content and density of snowpack
Sun Exposure
Precipitation

So try dumping all of these factors into the already ambiguous calculations to predict basic rainfall and runoff and you might as well throw darts.

I've been trying to figure out how to predict which bridges and culverts may have issues with scour and overtopping during snowmelt and spring runoff by paying attention to the snowpack levels in the adjacent upper drainage basins but it's not even that simple. During particularly warmer weeks in the spring I've seen larger flows in streams when there was less snowpack in the tributary basins than previous years. So rather than concentrating on snowpack I've learned to pay attention to successive daily high and low air temperatures and then react accordingly.

Definitely far from an exact science and not something we can "put in a box" like most engineers like to do.

 

[NeffariasBredd]

@Civilman72 Thanks! That's exactly what I was looking for (sorta, I was hoping for something prepacked in a box...) How did you use the daily highs and lows to estimate the runoff? Do you have any resources that I could use to learn more about this type of thing? My work is mostly in the Southeast (with notable exceptions) but my dream is to move out West in a few years so this would be a great thing to know.

Thanks!

 

[psmart]

Predicting stormwater runoff is a statistical process to being with. When you add snowmelt you're added another set of parameters with considerable variability.

The answer to your question depends on the exact purposes of your study (i.e. meeting stormwater regulations, designing a bridge, sizing a culvert, building a detention pond, etc.) and the consequences of system overflow.

From a computational standpoint, you could do something very basic like this:

Assuming that the snowpack contains 10 inches of water content, and the peak snowmelt is 20% in 24 hours, you can estimate a 24-hour smowmelt flow rate based on the snow-covered area. (2" water x snow area / 24 hours, with appropriate units conversions) Add this to your runoff from the superimposed rainfall event to estimate the total flow.


Peter Smart
HydroCAD Software

http://www.hydrocad.net

 

[pinkpig]

SWMM and HEC-HMS have snowmelting function I remember. but as others said, it is highly unpredictable and I have no confidence with the results from either model. Even if you can calibrate your model, due to the nature of snowmelting, your model may not be any close to "being valid" for the next event. I am eager to know any successful stories about snowmelting+runoff modeling.

 

[civilman72]

If the daily highs are in the high 60s or low 70s and the nightly lows remain above freezing, and this occurs for 3-4 days, we will see a large spike in stream flows. If we also have some high intensity storms mixed with these temperatures (it's not uncommon to get afternoon showers every day) the flows in the streams can get dangerous quickly. I've lived in CO almost 20 years and have only seen one year when all of these factors came together. So it really is a rare occurrence that snowmelt causes issues, but it is something we need to keep an eye on every spring. The streams I deal with are at about 8,000 feet, with tributary basins as high as 13,000 feet, so it's a fairly unique drainage scenario and probably cannot be compared to the UP.

The general consensus among design engineers is that storm drainage conveyance facilities on the Front Range (Denver, Fort Collins, Colorado Springs - elev. about 5,000 to 6,000 feet) should still be designed based on rainfall and that snowmelt should not be a design factor. But there is also a basic understanding that the higher you go in the mountains the more of a chance that snowmelt becomes the design parameter for predicting runoff, but there are simply not enough studies to reference to even start some sort of snowmelt runoff calculation.

From what I've seen over the last 20 years the storm conveyance facilities in my region that are designed with sufficient capacity to handle flows from a standard 100-year design storm (based on rainfall) have not had any major issues with peak runoff from snowmelt. But I'm also convinced that a snowmelt event that has the probability to occur once every 100 years (a 100-year snowmelt event) will generate a larger flow than the flows from a 100-year rainfall event; again this is specific to my region.

I would think that the UP is more like the Front Range than the Rocky Mountain drainage patterns that I'm observing, so I would be surprised if you really need to worry about snow melt for any conveyance design (but I find it an intriguing challenge to discuss regardless).

 

[cvg]

well, one of the larger floods on record on the Poudre was a result of rain on snow and snow melt in 1983. The most recent in 1999 was also caused by snowmelt. I would suggest that flooding as a result of snowmelt is a concern in Fort Collins

http://www.fcgov.com/utilities/what-we-do/stormwater/drainage-basins/poudre-river

http://www.wecsa.com/FCClimate/ClimHist.pdf

 

[77JQX]

If I'm thinking right, it takes quite a bit of rain to melt an appreciable amount of snow. The heat of fusion of water is 334 J/g, while the heat capacity of water is 4.18 J/g-degree K. So that if the rain is falling at 10 C (50 F) then 1 g of rain can release 41.8 J in being cooled to 0 C. If the snow is already at 0C (likely if it's raining out), then it takes 8 g of 10 C rain to melt 1 g of snow. The danger, as I see it, is that the snow can become saturated with water as it melts, and from previous day's rains, until it is saturated. Further rain may cause a tipping point when the saturated snow realeases its water or starts to flow. I don't know how one would predict that.

 

[agua84]

I agree with psmart, the level of detail of analysis depends on what you are designing or checking. In northern new Hampshire, my firm's typical practice is also to design stormwater facilities for typical rainfall events, i.e. 24 hour 10-, 25-, 50-, or 100-year rainfall, with typical conservatism and freeboard, and not explicitly account for snowmelt. That wouldn't be rigorous enough for design of a large dam, though. In my observations, the hydrograph for strictly snow melt (not including large rainstorms) would often be similar to a lower intensity storm of longer duration, but as stated there are so many factors...best to keep it simple if possible.

There are probably more specific resources, but if you look up RAINFALL FREQUENCY ATLAS OF THE MIDWEST by Floyd A. Huff and James R. Angel. (

http://www.isws.illinois.edu/pubdoc/b/iswsb-71.pdf),

they look at seasonal distributions of rainfall (not runoff), and Michigan typically receives most of its large storms and precip in the summer. Figure 11 regarding top-ranked 10-day storms by season is pretty interesting. Final thought, maybe someone at the agency regulating stormwater has a "design storm" for a rain on snow event, if that is a criteria they look at.

 

[dicksewerrat]

I would go to Marquette U to find a couple of PH.D candidates in the Engr school. Give them the problem. They could look at data from say the last 20 years. Daytime High Temp, Low temp, rainfall, recorded snow depths and depth of the various streams in that watershed. Make up graphs for each year. then do regression analysis to get a best fit for each year. Then see if next year fits. They could even put their name on the formula. Doesn't help now though.

Richard A. Cornelius, P.E.

http://WWW.amlinereast.com

 

[civilman72]

cvg - Fort Collins was not the only city with issues in 1983. There were a lot of issues throughout Colorado with snowmelt in 1983 - it was the biggest year for snowfall in recent history in Colorado.

While there may be a "concern" with snowmelt in some Front Range cities, it still has not changed the fact that no engineers on the Front Range design their conveyance systems based on runoff from snowmelt. We can conclude that it's because snowmelt does not generate as much runoff as a rainfall event in most locations along the Front Range, or it may be because it is too difficult to calculate and predict. The recent flooding that took place last fall near Boulder and Fort Collins would probably verify the first conclusion.

But back to the OP's question, I would be very surprised if conveyance systems in the UP are designed with snowmelt in mind.