Predicting Flow from Snow
Updated: Apr 26, 2020
If you're wondering about the timing and magnitude of this year's spring runoff, I have a prediction for you. Back in 2013, I taught my students how to predict these two important values using historical hydrology records, and through that process, they learned how a boring series of numbers, graphs, and equations can be used to answer a relevant question of great consequence. Since we developed this snow-flow model, it's been fun to test it's prediction each year.
The San Juan River's importance to the Southern San Juans, Pagosa Springs, and the rest of the Southwest is paramount. Besides filling our irrigation ditches, reservoirs, and faucets, the river also contributes aesthetically, recreationally, and economically to the area. The amount of flow in the river is also related to the summer wildfire risk.
Because it sustains us, the streamflow of the San Juan River has been monitored continuously since 1936. The hydrograph below reveals what most Coloradans already know - that the San Juan River is a snowmelt-dominated system and the amount of summer streamflow is dependent upon how long the snow lasts. Remember the flow values shown below are the average flow for the whole month - the actual peak runoff values are always much higher.
Streamflow in the San Juan generally peaks in the late May or early June. Based on 82 years of record, the median spring runoff peak is about 2500 cfs and usually occurs on May 26th. However, on rare occasions, monsoon rains have eclipsed the spring runoff values. Most notably, the largest documented flood occurred on October 5, 1911. Sustained monsoon rainfall generated an estimated 25,000 cfs, which roared down from the headwaters, destroying most of Pagosa Springs. The timing of the highest annual peak flow is depicted in the graph below. Basically, melting snow creates the highest flow of the year in May or June 89% percent of the time, and 4 % in April. But monsoon rainfall in July-October is responsible for the remaining 7% of the annual peak flows.
So how much water will this year's snowpack generate, and when can we expect the highest spring flow to occur? The further in advance a flow prediction is made, the better, but early predictions are less accurate. We used the March 1st snow water equivalent (SWE) from the Upper San Juan SNOTEL site to predict the peak SWE, which usually occurs around April 15th. The R^2 is fairly good (0.8) considering the host of other variables involved. This year, we had 13.5 inches of SWE on March 1, and the relationship predicts that the SWE will peak at 21 inches.
Then we used the peak SWE to predict the peak spring streamflow. The R^2 is slightly lower (0.71). If we get the 21 inches of SWE as predicted, we can expect a peak flow of around 1900 cfs.
Finally, we plotted peak SWE vs the peak spring runoff day number. This is the poorest of the 3 relationships because spring temperatures and moisture can vary drastically. Basically, the median spring runoff date (May 26th) is more reliable predictor than this relationship. Just for fun (as my wife mocks my use of "fun"), I used the trendline to generate a predicted peak date of May 19th (assuming we reach 21 inches of SWE).
So after all that, assuming we have an average Spring, our Snow-Flow model predicts that the San Juan River will have a peak spring streamflow of 1900 cfs on May 19th (compared to the median of 2500 cfs on May 26th). This prediction seems overly optimistic considering our sluggish start to the snow season, but hopefully, like the last 2 years, we'll have above average late-season precipitation to prolong the melt later into the year. We'll just have to wait and see!