Uncertainty in Utah Hydrologic Data – Part 2 – Streamflow Data
A three part series that examines some of the systematic bias in Snow Course, SNOTEL, Streamflow data and Hydrologic Models
Randall P. Julander Snow Survey, NRCS, USDA
Hydrologic data collection networks and for that matter, all data collection networks were designed, installed and operated – maintained to solve someone’s problem. From the selection of sensors to the site location, all details of any network were designed to accomplish the purpose of the network. For example the SNOTEL system was designed for water supply forecasting and while it is useful for avalanche forecasting, SNOTEL site locations are in the worst locations for data avalanche forecasters want such as wind loading, wind speed/direction and snow redistribution. All data collection networks have bias, both random and systematic. Use of any data from any network for any purpose including the intended one but especially for any other purpose should include an evaluation for data bias as the first step in quality research. Research that links a specific observation or change to a relational cause could be severely compromised if the data set has unaccounted systematic bias. Many recent papers utilizing Utah Hydrologic Data have not identified or removed systematic bias from the data. The implicit assumption is of data stationarity – that all things except climate are constant through time thus observed change in any variable can be directly attributed to climate change. Watersheds can be characterized as living entities that change fluidly through time. Streamflow is the last check paid in the water balance – it is the residual after all other bills have been paid such as transpiration, evaporation, sublimation and all other losses. Water yield from any given watershed can be impacted by vegetation change, watershed management such as grazing, forestry practices, mining, diversions, dams and a host of related factors. In order to isolate and quantify changes in water yield due to climate change, these other factors must also be identified and quantified. Operational hydrologic models for the most part grossly simplify the complexities of watershed response due to the lack of data. For the most part they operate on some snow and precipitation data as water balance inputs, temperature as the sole energy input, gross estimations of watershed losses mostly represented by a generic rule curve and streamflow as an output to achieve a mass balance. Temperature is not the main energy driver in snowmelt, short wave solar energy is. Hydrologic models using temperature as the sole energy input can overestimate the impacts of warming.
Water yield from the watershed is a residual. It is what is left over after all other processes have claimed their share of the annual input of precipitation. As these processes change, water yield is impacted. In order to quantify the impacts climate change may have on water yield, it is essential to identify, quantify and remove the impacts other processes may have had. As a first step, accuracy of the data set needs to be defined. To what level of accuracy can we actually measure streamflow. The USGS uses a rating system to rank each gaged point. Streamflow data can be: excellent, good, fair or poor. Each streamflow point is not really a point such as 60 cfs but should be thought of as a range depending on its rating.
In this chart, the point values are represented by the red line – these are the observed values. If this site were excellent, there is a high probability (90-95%) that the actual value is somewhere between the light green lines. Unfortunately, there are no sites in Utah that fit the excellent criteria. About 50% of the USGS sites in Utah are in the good category and the actual value of any given point will be between the light blue lines. About 40% of the sites in Utah are classified as fair and fit between the magenta lines. The other 10% of sites are rated poor and values may regularly be outside of the magenta lines. The point of data accuracy is that if one can only measure to the nearest 10% or 15% then the limit of our ability to quantify change or trends in these data also resides within that data accuracy. To say I have observed a 5% change in a variable with 15% accuracy may or may not have validity. The assumption would be that all data error associated with these measurements are equally random in all directions such that everything cancels to the observed value. These potential errors in measurements can compound as one tries to adjust streamflow records to obtain a natural flow such as the inflow to Lake Powell. One must add interbasin diversions and reservoir change in storage back to the observed flow in order to calculate what the true observed flow would have been absent water management activities.
In this graph, the USGS observed flow for the Sevier River near Marysvale Utah is adjusted for the change in storage of Piute and Otter Creek Reservoirs. Piute Reservoir is directly above the streamgage and Otter Creek is on the East Fork of the Sevier some 20 miles upstream with very little agriculture between the reservoir and the stream gage. Each line represents a full year of monthly total acre feet adjusted flow. Notice that fully 1/3 of the time from April to September, the adjusted flow goes negative, as much as 15,000 acre feet. We know that this is an impossible figure and clearly these data points are in gross error. As important is what we now don’t know about every other data point – are they any better estimates of adjusted flow than the ones that are clearly in error? What about those that are close to negative, say in the zero to 5,000 acre foot range – are they accurate? Or even those that “look normal”, can we be sure they are? The only reason we don’t suspect the Colorado River of this kind of data error is that it’s flow is large enough to mask out the errors.
So, how is the official record for Lake Powell inflow adjusted? This, again is a case where a data set has been generated to serve a specific function – to allow us to make a reasonable inflow forecast for Lake Powell that has some meaning to the Water Management Community. It does not reflect the true natural inflow to Lake Powell.
We adjust the observed USGS streamflow at Cisco by 17 major diversions and 17 reservoirs represented here by this schematic. What might the real inflow adjustments look like? On the Colorado side, there are 11,000 reservoirs, ponds and impoundments, 33,000 diverted water rights, ditches, canals, etc, and over 7,000 wells. On the Utah side, there are 2,220 reservoirs, ponds and impoundments, 485 diverted water rights, ditches and canals as well as an unquantified number of wells and center pivots. Wyoming certainly has some number of these kinds of water infrastructure as well but I don’t have those numbers yet. As one can clearly see, this becomes an accounting nightmare. Not just in trying to measure each of these diversions or reservoirs but how much evaporation is coming off of each reservoir and canal. In regards to surface area, canals and ditches may well have far greater evaporation than reservoirs. Reservoirs also have bank storage issues that alter hydrograph characteristics by storing on the fill side and slowly releasing (minus consumptive vegetative use) on the draw down.
It is likely that for most diversions – the error associated with those data is positive. Why make such a declaration – visit any water rights office and ask a simple question “has anyone ever come in and complained that they got too much water and would like to give some back?”. The complaint uniformly is ‘I am not getting my full allocation and I need/demand more’. Check any water gate and you will find that the gate wheel has been turned to its maximum extent against the chain lock either by the water master or by the farmer checking to see that it is. When water is life and the means of providing, each will try to maximize the amount taken. Many of these diversions are simple structures easily altered. The assumption that all of these water managements are consistent in time is not likely true. From this context, it is clear that any reconstructed inflow to Lake Powell will have the potential for serious deficiencies, especially as water use in the upper basins increases. Each 0.01 percent here and there be it a well or pond evaporation or diversion slowly adds to the incremental error in the data set.
Changes to the Watersheds
Since the settlement of the west, there have been extensive changes in watershed characteristics. These changes can have a substantial impact on water yield and consequently have direct bearing on current trends. Let’s start with grazing – there is a substantial body of literature documenting the impacts that grazing can have on water yield. Overgrazing leads to less vegetation, soil compaction and greater water yield and soil erosion. In the 1870’s, there were approximately 4,100,000 cows and 4,800,000 sheep in the 17 western states. By 1900, there were 19,600,000 cows and 25,100,000 sheep. This was the get rich quick scheme of the day – eastern and mid west speculators could buy up herds, ship west, graze for a couple of years then ship back east for slaughter – no land purchase necessary, no regulations – simply fight for a spot to graze. Western watersheds were denuded and devastated. The Taylor Grazing Act, passed in the 1930’s was implemented in part because of a change in hydrology – people in the west and Utah in particular were the victims of annual floods, mud and debris flows brought on by snowmelt and precipitation events on damaged watersheds. This change in hydrology – increased flooding and flow led to action to curtail grazing and heal damaged watersheds.
North Twin Lakes – 1920. Notice the erosional features, the lack of vegetation including trees.
Photos courtesy of the repeat photography project:
North Twin Lakes 1945. Notice that the erosional features are slowly filling in, sage and grasses are more abundant, trees are growing, the watershed is healing. Bottom line, less runoff, more consumptive use by vegetation. Hydrologically, this watershed has changed dramatically.
North Twin Lakes – 2005. Notice the erosional features are pretty much gone, excellent stands of all kinds of vegetation. Now, what is the difference in water yield from the watershed today compared to 1920? Water yield has decreased and consumptive use increased.
Along with restricted grazing, watershed restoration programs were implemented to improve conditions such as seeding programs to restore vegetation, bank stabilization and other watershed improvements. One of these programs was designed to mechanically reduce streamflows via increased infiltration and water storage on the watershed. Contour trenches were installed on watersheds throughout the west to reduce streamflow, floods and debris flows.
In This photo above Bountiful, Utah notice the extent and depth of the contour trenches installed in the 1930’s by the Civilian Conservation Corps by hand and by horse drawn bucket scoops. These trenches are even today several feet deep and able to store significant amounts of snowmelt for infiltration.
Mining was an activity that impacted western watersheds in way not typically thought of from todays perspective. After all, mines and the associated infrastructure and even the tailings comprise a tiny fraction of any watersheds geographic area. However, from the 1850’s to basically the 1930’s or even later, ore had to be refined on site. There was no infrastructure or capability to bring ore from the mine to central smelters nor was there ability to bring coal to the mine. Roads were steep and rugged, rail lines expensive if they could be built at all and transportation was by wagon. Thus smelting was most often done at the mine via charcoal. The large mines would have 20,000 bushels of charcoal on site. Large charcoal kilns could take 45 cords of wood per week which equates to 36 million board feet of timber per decade. The famed Comstock Mine basically denuded the entire east side of Lake Tahoe. The cottage industry of the day was making charcoal for the mines. Many farms and ranches had smaller kilns to generate an additional cash flow.
The Annie Laurie in the Tushar Mountains of Utah.
The Annie Laurie today. Notice in this recent photo how vegetation, especially trees have grown, matured and how many more conifers there are today than in the past. In addition to charcoal, timber was necessary for the mine, for the buildings, for heating and cooking.
Locations of the estimated 20,000 abandoned mines in Utah. This represents a substantial amount of timber removed from Utah watersheds over a nearly 80 year period of time. Most assuredly enough to impact species composition and water yield across many watersheds. Fewer trees equals greater water yield.
Logging on western watersheds provided necessary timber for infrastructure such as homes, businesses, barns and other buildings. Timber was most often cut and milled on site with the rough cut timbers hauled from the watershed via horse and wagon.
The Equitable Sawmill, early century.
Where the Equitable Sawmill once was. Notice the dramatic change in forest cover – more trees equals less water yield and in this case, potentially much less.
Tie Hacking was a business that provided railroad ties to the industry, basically the same as logging but with a bigger product. As the railroad came through, tie hacks would cut trees and provide the necessary ties to keep the tracks moving forward. Ties at the time were not treated as they are now and needed to be replaced on a regular basis as the soft pine wood could rot quickly.
This is Blacks Fork Commissary – the Tie Hack central provisions location on the North Slope of the Uintahs in northern Utah.
Tie Hacks high grading all the Douglas Fir off the North Slope, leaving the Lodgepole Pine. The majority of the North Slope today is comprised by dense stands of Lodgepole Pine. The rail lines required 3000 ties per mile and 600 miles between western Colorado and the Sieras – at about 14 million board feet per decade.
The policy to fight western fires has done more to change the landscape of western watersheds than possibly any other factor. At the turn of the century, fires burned 10 to 30 million acres of forest every year. With the advent of Smokey Bear, between 2 and 5 million acres burn annually. This huge reduction of burned area has change the species composition, density and age of forests across the west. Watersheds that used to have 10 trees per acre now have 200 and more. Fewer trees produce more water yield.
Danish Meadows, 1900 – with frequent fires.
Danish Meadows, 2000. No fires for nearly 100 years. More trees, less water.
The Forest Service has done much research with paired watersheds and timber harvest. The Fools Creek experiment in Colorado is a classic – two watersheds of similar characteristics measured together for more than a decade. Then one watershed was kept pristine while the other was cut by 40%. The end result was an increase in water yield of 40% for 20 years as well as a substantial 25% increase for the period of 30 to 50 years.
Note that the timing of annual snowmelt was also accelerated due to the fact that the forest cover was opened up to short wave solar radiation, the primary energy input to snowmelt.
A recent Duke University study confirms that Utah forests are basically very young with the dominant age class in the 0-100 year old category. This basically confirms that post the mining/logging era from 1850’s to 1960’s a different watershed management policy has occurred on Utah watersheds. Small trees not harvested early on are now the 100 year old trees and seedlings at the time are now the 50 year old trees.
Species Composition Matters
With the virtual elimination of both fire and logging, species such as Aspen are being steadily replaced by Conifers. In paired plots to compare water consumption between Aspen and Conifers, LaMalfa and Ryel found that there was much greater SWE accumulation in the Aspen stands vs the Conifers – research already well known, but also soil moisture under the Aspen stands was much greater than it was under the Conifers. Aspens, with the first frost of the season terminate transpiration and soil moisture starts to recover. Conifers on the other hand, keep transpiring and pumping that moisture out of the ground.
LaMalfa/Ryel – 34% less SWE under the conifers than aspens.
LaMalfa/Ryel – nearly 4.5 inches less soil moisture in the conifers vs the aspens.
Overall, there was 42% less water in the Conifer Community vs the Aspen Community – a whopping 10.5 inches of less total water potentially available from the Conifers than the Aspens. This area of northern Utah, near Monte Cristo typically only gets 37 inches of annual precipitation so the Conifers could potentially produce far less runoff than the Aspens. Utah and Colorado have lost 2.5 million acres of aspens to conifer encroachment with approximately 1.5 million of that in the Colorado River Basin. That translates into about 125,000 acre feet of water lost per inch of water yield. From this single factor (Aspen replaced by Conifer), the April – July inflow to Lake Powell could be reduced by 2% to 17%.
Ground Water Withdrawals
In the long term, ground water is connected to surface water. This is an area that needs investigation as groundwater withdrawals within the basin could be substantial, certainly on the order of thousands of acre feet annually and potentially much greater. Over the period of many years, increased streamflow losses to groundwater is likely.
In the early years, agriculture was primarily flood irrigation where a big share of the water applied to any specific field would runoff back to the ditch and would eventually become return flow to the river. Much of the flood irrigation has been replaced by sprinkler irrigation with much higher evaporative losses but more efficient crop production. Nearly all of the water that hits the ground is consumptively used by crops. How this impacts streamflow is an issue for more research.
For a century in the west, we indulged in watershed practices that increased streamflows. In the 60’s and 70’s there was the beginnings of the environmental movement and with it significant changes in watershed management. Mining no longer requires vast amounts of timber, logging is but a scant fraction of its past, tie hacking is extinct, fires are extinguished, grazing is tightly managed. Watersheds now have huge amounts of vegetation and in particular vastly more trees than they have ever seen in a historical context. Species composition has changed with far less aspens and far greater confers. More conifers equals less snow, more conifers equals less soil moisture, more trees and vegetation in general equals less streamflow. For 100 years we systematically increased flows from western watersheds and for the past 50 we have done everything possible to reduce streamflows.
Portent for the Future
Forest management that has removed fire and logging from much of the equation has had a net effect of vastly increasing the number of trees per acre of land. Too many trees for a water resource has increased the competition for water to the extent that the recent drought weakened the forests and a huge pine beetle, spruce bud worm infestation has killed hundreds of thousands of acres of trees. The analogy of 10 men on the edge of a desert with water for 5 in appropriate. If one sends all 10, they all die. If one sends 5, most will likely survive. Our forests sent all 10 and the result is massive forest mortality. In Utah, of 5 million acres of forested lands, nearly 1 million acres is standing dead with the potential for greater mortality. 1 million acres of dead forest equates to the potential of 83,000 acres of additional water per inch of water yield, perhaps as much as 800,000 acre feet in total. In the short run, Utah is likely to see greater water yield, not less – all other things equal. Also, runoff will likely be earlier due to the opening of the canopy to short wave solar radiation.
There are many and complex reasons for declines in streamflows west wide of which climate change is but one. It is not a simple issue and each contributing component is certainly not easily quantified.