#21: More discharge data to address more hydroclimate questions

[Dan Isaak]

Lots more to know regarding the flow

Hi Everyone,
So wrapping up the hydrology module with a few miscellaneous thoughts
this time out….

As you might recall from previous blog posts in this module since last
September, we actually know quite a bit regarding how stream
discharges are evolving across the western US in response to climate
change. Warmer temperatures are causing snowpacks to shrink in many
places (blog #16), streams are running off sooner, flood risks are
changing (blog #17), summer flows are trending lower for a variety of
reasons, of which climate is likely one (blog #18), groundwater inputs
may or may not buffer some streams against these changes (blog #19),
and GIS tools exist for mapping historical trends and future discharge
scenarios in streams across the region (blog #20). Fundamental to most
of what we can infer regarding long-term trends are the empirical
measurements of stream discharge that are routinely taken at thousands
of stream gages (~10,000) long maintained by the USGS and a few other
agencies across the country (graphic 1).

Thousands of monitoring sites sounds like a lot but these sites are
not uniformly distributed in space, nor were they originally set up in
association with any sort of probabilistic statistical design that
would be representative of the river and stream networks across the
U.S. so there are important limits regarding the information this
network can ultimately yield. Moreover, there are literally 100,000’s,
perhaps more than a million, stream kilometers across the U.S. that
the existing gage network has to provide information about. Just as a
small sub-regional example of the disparity between the magnitude of
stream networks and number of monitoring sites, graphic 2 summarizes
what the full stream network across Montana and northern Idaho looks
like when the composite stream segments are plotted by their
contributing areas and elevations. That universe consists of more than
300,000 stream kilometers, yet it’s presently monitored by only about
400 active gages, which translates to 1 gage for every 750 kilometers
of stream.

No doubt, monitoring even at low densities, especially when maintained
for as many years as discharge gages often are, provides invaluable
information. But a low sampling density also means that the existing
network is best at describing only the most general things about
regional patterns and the places being monitored. Discharge measured
at the outlet of a basin integrates across many places doing different
things at different times, and one of the primary challenges
hydrologists face is how to determine what is going on in the places
that are not measured. Models can help us estimate different flow
patterns in different parts of the basin (blog #20) but validation at
finer resolution is also necessary to test and refine the models. As
was the case with stream temperature monitoring (blog #’s 3, 4, 8, 9),
new sensor technologies now make it possible to collect accurate
discharge data more cost-effectively than was historically the case
(graphic 3). These sensors might be used to densify the existing
discharge monitoring network in key ways that enable important
questions to be addressed with relatively small investments of time
and money if done with proper forethought.

So that’s it for the hydrology module, but did want to give a random
shout-out to a great paper by Olden & Poff (2003), which is a meta-
analysis that examined and synthesized all the different ways that
stream discharge has been quantified to describe hydraulic regimes.
They threw 171 different discharge metrics into a multivariate
analyses to describe their common variability and most were proven to
be highly redundant (graphic 4). This suggests it’s possible to
describe the core attributes associated with a regime through a much
smaller set of relevant descriptors. Things aren’t quite there yet
with regards to stream temperatures because we have so little full
year data that “regime” thinking isn’t quite as far along (see Poole
et al. 2003 for an important exception), but we’ve got a good start on
a confusing & redundant mix of summer temperature metrics that all
carry pretty much the same information. Hopefully before lots of full
year stream temperature data start coming online from the massive
regional monitoring networks now in place (blog #4), someone will put
good thought into developing a concise but fully descriptive set of
temperature metrics that describe thermal regimes so we don’t get to
171. Having a common lexicon with which we all speak “temperature”
would be a very useful thing since we’ll be doing a lot of it this
century.

Until next time, best regards.
Dan