By Matt Owens January 2, 2014
Streamflow in the Pacific Northwest has been significantly declining - 16% over the past 60 years. But why? And will the trend reverse course?
Previous tabulations have laid the blame for the diminishing water supply on increased evaporation from land surfaces and increased transpiration from plant leaves (together called evapotranspiration, ET). This presumed increase in regional ET was in turn presumed to be a consequence of warming from the human-enhanced greenhouse effect as it continues to push the energy balance of the global climate further out of kilter.
But something doesn't quite add up.
I recently spoke with Charlie Luce about this. He's a research hydrologist at the Boise Aquatic Sciences Laboratory, and he led a research team that investigated this issue.
They identified a paradox in the current understanding of the water cycle for the Pacific Northwest, and they also proposed an explanation for it, the just-published details of which involve the changing westerly winds.¹
Here's the basic math problem that Luce's team examined: Human-caused global warming has added about 1.6 W/m² in net extra downwelling radiation to the climate's energy budget, on a global average basis. Therefore, if extra evapotranspiration were being caused by that increase in radiation, then the amount of energy required to evaporate the lost streamflow - the missing water - should balance with the increased radiative forcing.
You can think about this in terms of evaporating water from a pan on top of an electric stove. Turning up the heat uses more watts of electricity. Either there are enough watts to do the job, or there aren't.
But the local imbalance could be greater than the global 1.6 W/m² average because of earlier snowmelt that exposes dark ground which in turn absorbs more of the solar energy (more watts) than the reflective snow. The change in annual “darkness/reflectivity” of the ground, or, technically speaking, “albedo,” could be contributing about 2.9 W/m² on an annualized basis per week of earlier snowmelt in this region, as stated in the research paper. But even factoring in this earlier snowmelt, there still doesn't seem to be enough extra energy in the budget to evaporate the region's missing streamflow - especially in the driest years.
Another option could be that extra heat is coming into the region from an adjacent area - but the temperature record doesn't support that.
Then there's also the tempting trend in average snowpack. As more precipitation falls as rain instead of snow, snowpack has been on the decline across the US West. Because of this, streamflows are now lower in summer (because of the smaller snowpack), and higher in winter and spring (as the snowpack gets attacked by rain and/or melts out early). So one might think this trend is somehow responsible.
But the total volume of water that flows down waterways gets recorded by stream gages. And there's just no way around a stream gage.
If the snowpack trend explained this mystery, then the gages would show unchanged annual average streamflow. But the gages on mountain streams do show a strong decline in annual average streamflow, Luce said.
So really, there's only one option remaining.
If the water isn't coming down the streams from the mountains, and if it hasn't evaporated off the mountains, and if it isn't hanging around in the mountains as extra snow, then it must never have fallen on the mountains to begin with. That means precipitation in the mountains must be declining.
That conclusion however, contradicts the accepted and “expected” trend for the Pacific Northwest's rainfall over the past 60 years, a regional trend that shows only a very slight decline.
But that regional trend, according Luce, is based on a flawed interpretation of the data record. The flaw stems from a faulty assumption that both high and low elevations would experience similar rainfall trends over time and with warming.
To understand this reasoning we need to step back a moment and see that the historical rainfall trend has been assembled from individual weather station records. Collectively, these are known as the Historical Climate Network, or the HCN. And, before the days of automated stations, record-keeping was done manually - someone had to be there every day of the year. So, it should be no surprise that the Pacific Northwest's remote mountains have had many fewer stations on a per-area basis in comparison to the lower elevation areas. And because records must be weighted on a per-area basis to obtain a regional trend, faulty assumptions about differential trends between elevations could lead to faulty conclusions about total regional rainfall.
The low sampling rate across higher elevations has been recognized, but what was not recognized until now is that rainfall might change substantially at higher elevations while staying unchanged at lower elevations.
So, Luce's group looked outside of the HCN to synthesize a new regional rainfall trend with better representation of mountainous areas. They used various other data sets, including streamflow and snowpack data. They concluded that there has indeed been a decline in mountain rainfall while lower-elevation rainfall has remained mostly steady.
And they went further, linking the trend to changes with the winds. In their paper, they write there have been “substantial historical declines in precipitation in the Cascades and Northern Rockies” and that these are linked “to observed changes in atmospheric circulation,” specifically the westerly winds.
The Westerlies are two bands of winds, one circulating in the Northern Hemisphere and one in the Southern Hemisphere. These westerly winds flow roughly midway between the poles and the equator. If you imagine the globe, they are like two doughnut-shaped bands of winds blowing from west to east around the globe in both hemispheres.
Like all winds, these westerly winds are strongly influenced by temperature differences; and so as the planet warms - and as some parts warm faster than others (e.g. the Arctic) - the winds have been changing.
In fact, the Northern Hemisphere's set of westerlies and its associated polar jet stream have both been slowing down in recent decades.
The slowing jet stream has been linked to a decline in Arctic sea ice extent in recent years too. The loss of sea ice means an extra release of heat into the polar air during winter, and that changes the temperature difference (the gradient) between air masses in a north-south direction; and thus the changed gradient leads to a change in the pattern of wind flow.²
Slower westerly wind flow means slower movement of air across the mountains, Luce said, and thus less total water-laden air flowing over the mountains. As a consequence, there has been less mountain precipitation.
Water-laden air, as it rises over the mountains, tends to lose some of its water in a very important process called “orographic enhancement.”
In this process, moist warm air is forced higher into the atmosphere as it flows over mountains, and as it rises, it also cools and expands. That cooling and expansion reduces the water-holding capacity of the air, and so some of the water precipitates out as rain or snow.
This is why the western sides of the Cascades and Northern Rockies are generally green and lush compared to the dry and arid eastern sides. The precipitation gets “squeezed out” as the air rises, but then as the air sinks on the other side of the mountains, it contracts and warms again, thus lowering its relative humidity and inhibiting precipitation. The same pattern is seen around the world in many other mountain chains too.
The researchers explain this uplift effect in their paper, writing that “[o]rographic enhancement of precipitation is a primary driver of the geographic distribution of precipitation across the [Pacific Northwest] and other mountainous regions. A majority of the precipitation received in the [Pacific Northwest] occurs during October to April, when a progressive storm track and embedded cyclones advect moist and stably stratified air masses toward mountain barriers, thereby providing conditions conducive to upslope precipitation enhancement.”
Part of the reason the newly identified differential trends came as a surprise is that global climate models have a blind spot for orographic enhancement.
Fine-scale topographical details are very important for a physics-based simulation of orographic uplift, but until recently, models have not had enough computing power to do much of any fine-scale work.
To put some numbers on this, a typical general circulation model has a pixel size of about 120 x 120 km, far too large to produce complex orographic enhancement effects. If you imagine a mountain chain, large peaks within it are just a few kilometers in diameter, and there are lots of topographical intricacies in the peaks themselves and the foothills too.
Luce emphasized that there are many other important factors potentially at play here too, including: (1) Decreasing temperature contrast between land and ocean during winter leads to reduced relative humidity over the land and thus less potential for rainfall. (2) Increased aerosol content of the air leads to decreased droplet size and thus less efficiency (e.g. more of the rain evaporates before hitting the ground). (3) And, increased rainfall due to higher absolute holding capacity (warmer air can hold more water) could lead to an opposite effect, causing more rainfall. Although, this last factor would tend to lead to more intense rainfall events and thus a higher fraction of the precipitation would runoff quickly; the net impact could be more total rainfall but lower average streamflow and lower average soil moisture, accompanied by increased erosion (in other words, more drought and flooding in the same area).
The changes in rainfall also impact forests and other vegetation, which in turn changes wildfire patterns, albedo, snowpack, and so on. Each of these can amplify or counteract the trend in rainfall to a certain extent, but obviously the trend in streamflow has been all downhill so far.
Luce said one of the more dire pieces of news from this investigation is that “the effect seems to have been most strong in the driest years.” Extremes are getting more extreme he said; and “if you look at the wind speed, you see that the slowest years are getting slower than the average - and the same is true of precipitation - the driest years are getting more extreme faster than the mean.” A similar pattern of extremes leading the mean has also been observed in extremely high-rainfall events across the United States, with high confidence in the trend.³
And on the ground, Luce and his colleagues are seeing a pattern of declining streamflows in the mountains, and many cascading ecological consequences. For example, rainbow trout are maturing very quickly and staying small, exhibiting a “live-fast, die-young” life-cycle pattern. That probably stems from factors that include a mismatch in the timing of food availability (out of synch seasons) and consequences from the warmer water itself.
Warmer water means a faster metabolism for the cold-blooded fish, and so they must eat more - unless they can find a cool refuge. But their only line of retreat is to swim higher upstream, and that's being cutoff by lower streamflow.
To make matters worse, the smaller volume of water in the streams means that the water heats up faster and is more vulnerable to temperature fluctuations. Most aquatic life does not do well with strong temperature swings.
Warming can also induce a release of nutrients into the water and lead to a proliferation of unhealthy microbes. That unhealthy water quality then combines with other stressors to kill or weaken the fish. High microbe-counts can also sicken or sometimes even kill people who come in contact with the water.
Another stressor is that as water warms, it cannot hold as much oxygen. So the cold-blooded fish's metabolism rises while their oxygen supply declines. Many owners of ornamental fish ponds have seen the consequence of this first-hand during the increasingly-common heat spells of recent years.
A further consequence of declining mountain precipitation, and one of many hydrological issues Luce has studied before, is the declining mountain snowpack. Mountain stream ecology is closely connected to the snowpack via the constant supply of cooling water to the stream. On hot summer days, rising air temperature heats the stream water and may evaporate more of it than usual, but it simultaneously melts more snow, thus cooling the stream faster and adding water at a higher rate. That interlocking mechanisms helps keep the streamflow and water temperature steady, unless the snowpack shrinks too much.
Tree ring proxy data shows that compared to the past thousand year period, the recent levels of Pacific Northwest snowpack have been exceptionally low. And they have been sustained at these low-levels for an unusually long stretch of time too.⁴ Yet more research shows that the majority of this trend is caused explicitly by the human enhancement of the greenhouse effect.⁵
Regarding the future, Luce expects that 16% trend of declining streamflow will continue as greenhouse gas levels continue to rise and as the northern westerlies continue to slow.
Right now, the 2013/14 winter is showing exceptionally low streamflow and low snowpack (see the image and links at the end), well below the declining average trend Luce's group has identified. He said these conditions provide a window into what the region's future “normal” streamflow will probably be like as the average decline continues to progress.
All this raises serious concerns about crossing ecological and hydrological thresholds. Such changes can be irreversible on time scales that matter to most people.
Conceptually, ecosystems and their components respond in two basic ways to gradual disruption punctuated by extreme disruption. Either they spring back to continue their process of gradual change, or they don't recover and undergo significant rapid change. These two patterns are shown in the figure below.⁶
¹ Luce et al. 2013 - “The Missing Mountain Water: Slower Westerlies Decrease Orographic Enhancement in the Pacific Northwest USA;” doi: 10.1126/science.1242335.
² Francis et al. 2013 - Francis et al. 2012 - “Evidence linking Arctic amplification to extreme weather in mid-latitudes;” doi:10.1029/2012GL051000.
³ Kunkel et al. 2013 - “Monitoring and Understanding Trends in Extreme Storms - State of Knowledge;” doi :10.1175/BAMS-D-11-00262.1.
⁴ Pederson et al. 2011 - “The Unusual Nature of Recent Snowpack Declines in the North American Cordillera;” doi: 10.1126/science.1201570.
⁵ Barnett et al 2008 - “Human-Induced Changes in the Hydrology of the Western United States;” doi 10.1126/science.1152538. The paper states “The results show that up to 60% of the climate-related trends of river flow, winter air temperature, and snow pack between 1950 and 1999 are human-induced. These results are robust to perturbation of study variates and methods. They portend, in conjunction with previous work, a coming crisis in water supply for the western United States.”
⁶ Figure 1. reproduced from Luce et al. 2012 - “Climate Change, Forests, Fire, Water, and Fish: Building resilient landscapes, streams, and managers;” US Forest Service publication.
And, for those interested, the Historical Climate Network (HCN), click here.
And below is the streamflow conditions from the USGS for January 2nd, 2014 (click here for an updated map on the USGS page):