Big Data in Hydrology (CUAHSI 2016): part 2

The speaker was Praveen Kumar from University of Illinois at Urbana-Champaign with a lecture entitled “Intensively Managed Landscapes: Anthropocene in Action”.

My notes from this talk are somewhat fragmented: “check out CZO Illinois [Critical Zone Observatory]”; “fibre-optic cable within a well to monitor the temperature profile in depth”; “rethink models: going away from rainfall-runoff, include high resolution spatial data”; “the future arrived too quick”; “on upscaling: going from small scale to large scale, add other processes in each step”… Praveen Kumar gave a great overview on how humans have altered and continue to alter the landscape in the US midwest.

Here I feel obliged to give a link related to the “Anthropocene” debate. It’s been going on for a while, and, it may be the largest clash between Geoscience fields in our lifetime. It’s a bit like a boxing-match between Stratigraphy (and Geology) and Environmental Sciences (and maybe Hydrology), as far as I see it. It was interesting experience to witness people passionately defend/oppose Anthropocene few years back.  Nowadays, it seems that the debate has shifted in how exactly to define the beginning of this new geological epoch. Here’s a nice Nature news feature by Richard Monastersky with links to other materials.

Praveen Kumar’s lecture is also in YouTube, so enjoy: click! Here’s also collection of the tweets about this talk (or at least the ones I found post-factum):

• The Wolman lecturer is Praveen Kumar of UIUC, talking about intensively managed landscapes of midwest – anthropocene in action #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Simple looking landscapes of corn & drainage ditch are really complex on closer inspection. Starting with glacial legacy. #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Kumar has given us a tour de force of the hydro, biogeochem, geomorph, ecology connections at work in corn fields/drainage ditches. #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Waveform lidar allows mapping 3D structure of individual trees! Can ID to species from plane based lidar. #CUAHSI — Anne Jefferson (@highlyanne)July 25, 2016
• This hyperspectral lidar can get allow mapping of sediment grain size in floodplain deposits! Whoa. #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• We’re going to need new models that make use of these new datasets. We’re going to need new scientific questions! #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Kumar: significant human inputs in intensively managed landscapes overcome rate limits & states for enhancing ag productivity #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Kumar: landscape is being re-sculpted – heterogeneity and connectivity are non-stationary #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Kumar: explore opportunities for new ways to approach modeling in a data rich world for novel insights – phenomenological prediction #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• UIUC’s Praveen Kumar emphasizes the need to exploit high res data for predictive modeling #bigdata #bigchallenges #datarichworld #CUAHSI — Natalie Nelson (@natnels) July 25, 2016

This was one of the two parallel morning sessions. Session chair was Scott Jasechko from University of Calgary. There were three talks in this session:

Carol Kendall from USGS with “Lessons learned from 30 years of developing basin to national-scale hydro-biogeochemical studies, generating huge datasets, and interpreting the data”. Carol Kendall gave a nice overview of her career (project involvement) at the USGS. She really likes the word “piggybacking”, which according to her, describes well what she does, e.g. using other people’s projects and/or data to answer different questions (different from the original data collection purposes). There was a slide about kayaking vacation, when she also collected samples for future study (that’s just how cool her experience was :)). I’ve written a note to myself to check the book “Isoscapes: Understanding movement, pattern, and process on Earth through isotope mapping”. Carol Kendall is the first author of Chapter 5: Applications of Stable Isotopes for Regional to National-Scale Water Quality and Environmental Monitoring Programs (link). Here are the tweets about her talk:

• Carol Kendall is showcasing @nitrogenfixer atmospheric isotope work at #CUAHSI — Krissy Hopkins (@kghopkin) July 25, 2016
• Caroll Kendall on lessons learned from 30 years of isotopic big data studies. [Her work makes me swoon] #CUAHSI — Anne Jefferson (@highlyanne)July 25, 2016
• Kendall’s lesson #1: Good idea to piggyback your sampling onto monitoring programs for other things. Lets you test big hypotheses. #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Kendall lesson 2: The more tracers the better. Multi-isotope multi-tracer approach to enviro studies allows more insight. #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Kendall lesson #3: Archive samples for future interest. (She archives 3 L w/in 24 hours. Need lots of freezers tho) #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Kendall lesson 4: Conducting free pilot studies especially longitudinal transects might ID unsuspected hot spots or hot moments #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Kendall lesson 4[5]: don’t let huge datasets stop you from starting with simplistic analyses, like 2 endmember mixing models. #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Kendall lesson 6: 1/2 of what anyone (including you) thinks is true is NOT and you don’t know which half! Treat all as hypotheses. #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016

Naresh Devineni from The City University of New York with “America’s Water in the 20th Century: Measures to address climate induced risk”. His talk was about the new indicator of drought-induced water stress, see more details in his paper “America’s water risk: Current demand and climate variability” published in Geophysical Research Letters (April 2015). Tweets:

• Neat talk on America’s Water data integration and modeling effort out of Columbia Water Center & CUNY. https://t.co/1fugOiq6bM #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Pointed Q from @FroudeNum about America’s Water model. Where are the forests? A: Not there, just crops. [But trees use lots of H20] #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016

James Kirchner from ETH Zurich (and few other places) with “Isotopic indicators of water ages, from minutes to millennia, in groundwater and streamflow”. He presented ongoing research on continuous isotope sampling and hydrograph partitioning and some of his most recent publications: “Substantial proportion of global streamflow less than three months old” (first author Scott Jasechko) published in Nature Geoscience (2016), “Aggregation in environmental systems – Part 1: Seasonal tracer cycles quantify young water fractions, but not mean transit times, in spatially heterogeneous catchments” and “Aggregation in environmental systems – Part 2: Catchment mean transit times and young water fractions under hydrologic nonstationarity” both published in Hydrology and Earth Science Systems (2016). Full texts available on his Research Gate profile.

• Last talk for the morning is by Jim Kirchner – this year’s @theAGU Langbein Lecturer. MOAR isotopes + water ages. #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Dr Kirchner on groundwater ages revealed by isotopes: below 50m ~ >50 yrs, 200m ~ pre-Holocene (fossil h2o!) #CUAHSIpic.twitter.com/XQP5w1Hqt5 — Natalie Nelson (@natnels) July 25, 2016

This was one of the two parallel afternoon sessions. Session chair was Kamini Singha from Colorado School of Mines. There were three talks in this session:

Markus Weiler from University of Freiburg with “Powerful hydrological datasets with novel observations under small budget”. Talk on the potential of low-cost sensor network to understand spatial and temporal dynamics of mountain snow cover. He covered the different sensors and techniques that he and his group are using… from time-laps cameras to study interception to quantifying discharge by using smartphone app (e.g. photrack)

• Great talk from Markus Weiler on low cost sensor networks – time lapse cameras for ephemeral streams, snow sensors & more. #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016
• Markus Weiler- novel observations under a small budget: 1. Modify existing sensors 2. Develop sensors 3. Innovate #CUAHSI — Laura Sugano (@Laura_Lynn007) July 25, 2016

Scott Tyler from University of Nevada at Reno with “Finding your place in the Big Sky: The development and application of small Unmanned Aircraft Systems (sUAS) for hydrologic science and engineering”. He talked about structure from motion and drones. That’s where I heard about the changes in drone regulations for scientific use, which happened this August.

• Now Scott Tyler talking about the applications of UAVs (drones) for hydrologic science. #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016

Ben Crosby from Idaho State University with “High resolution topography for measuring the response of a warming arctic landscape: A case study from Alaska and the promise of arctic-wide coverage”. He also spoke about photogrametry and stereo imaging and applications in geomorphology.

• Ben Crosby told us about the high resolution topography revolution happening in the Arctic. 2-6 m data from ArcticDEM #ThanksObama #CUAHSI — Anne Jefferson (@highlyanne) July 25, 2016

My only tweet for this day was actually a sum up of this afternoon session:

• Day1: #photogrametry & timelaps image analyses (& other remote sensing) in hydrology: the exciting new way to answer new&old Qs #CUAHSI — DenitzaV (@DenitzaV) July 25, 2016

This was one of the two parallel morning sessions. Session chair was David Hyndman from Michigan State University. There were three talks in this session:

Jim Buttler from Kansas Geological Survey with “Subsurface Characterization over a Range of Temporal and Spatial Scales“. Jim Butler showed how to analyse groundwater level changes due to pumping for irrigation and to infer aquifer characteristics for this. Also he walked us through the finding reported in his paper “A new approach for assessing the future of aquifers supporting irrigated agriculture” published in Geophysical Research Letters (2016). The good news are that 22% reduction in pumping for irrigation will stabilize the groundwater levels of High Planes Aquifer in Kansas.

• Direct push technology (e.g., Geoprobes) w/ tools for doing quick K profiles can give lots of data about unconsolidated sites #CUAHSI — Anne Jefferson (@highlyanne) July 26, 2016
• Sobering map of depletion of High Plains aquifer, but incredible datasets that underly these maps. #CUAHSI — Anne Jefferson (@highlyanne) July 26, 2016
• With data Kansas has can create water budget & water table response to quantify reduction in pumping needed to stabilize water level #CUAHSI — Anne Jefferson (@highlyanne) July 26, 2016
• Theme of this #CUAHSI biennial is big data. Hydrogeologists (gw) have simultaneously dealt w/ big data & data scarcity for a long time. — Anne Jefferson (@highlyanne) July 26, 2016
• Jim Butler (Kansas Geo. Survey): 22% decrease in #irrigation pumping – would keep W-Kansas gw levels stable: big data utilisation #CUAHSI — DenitzaV (@DenitzaV) July 26, 2016

Susan Hubbard from Lawrence Berkeley National Laboratory with “Exploring the Effects of Climate Change on Critical Zone Behavior Using Geophysical Approaches“. Susan Hubbard presented results from two projects, the first one associated with Next-Generation Ecosystem Experiments (NGEE-Arctic) and the second from a headwaters Colorado basin watershed. What I found most interesting from this talk was the part covering the paper “Identifying multiscale zonation and assessing the relative importance of polygon geomorphology on carbon fluxes in an Arctic tundra ecosystem” published in Journal of Geophysical Research (2015). They developed a method for characterizing the spatial variability of Arctic polygonal ground geomorphology (ice wedges) using unsupervised clustering method and attributes extracted from geophysical data sets and kite-based landscape-imaging data. Here one more link about that (click).

• Susan Hubbard gave us an overview of two big DOE projects – next gen ecosystem experiment: Arctic & Colorado Headwaters SFA #CUAHSI — Anne Jefferson (@highlyanne) July 26, 2016
• On North Slope Alaska, patterned ground shows covariance in above & below ground properties that then explain GHG respiration #CUAHSI — Anne Jefferson (@highlyanne) July 26, 2016

Kamini Singha from Colorado School of Mines with “Subsurface Imaging of the Critical Zone“. This was my favorite talk from the CUAHSI biennial! Kamini Singha told the story of a RoboTree – the most instrumented tree in the forest. She presented time-laps ERT from the tree trunk (circle profile). The shocking part was the heterogeneity in conductivity you can observe during their experiment.  There was also a question for further study on the representativity of sap flux measurements (single point) in the presence of such heterogeneity within the cross-section.

• Kamini Singha – How to image processes where we have limited data? Esp. Vegetation-geology feedbacks. Again, ET is a grand challenge #CUAHSI — Anne Jefferson (@highlyanne) July 26, 2016
• Kamini Singha on #ER (electric conductivity [resistivity]) inversion of a tree cross-section #CUAHSI pic.twitter.com/Gy6hzrktok — DenitzaV (@DenitzaV) July 26, 2016
• Kamini Singha – a single tree ‘RoboTree’ told an intriguing hydraulic story within its landscape #cuahsibiennial pic.twitter.com/KVqJf4Wpds — CriticalZoneOrg (@criticalzoneorg) July 26, 2016

Pathfinder is a fellowship for graduate students, so they can broaden their research, go visit another group, learn new method, collaborate with specialist from other science fields, or get another experimental site for comparative study. Here is the site with more information. In this session two (or maybe three) students presented results from their Pathfinder Fellowship.

• Scott Allen, a Pathfinder Fellow, learned that flooded wetland ET rates may be energy limited, not veg limited as prev. thought. #CUAHSI — Anne Jefferson (@highlyanne) July 26, 2016
• Allen postulates a relationship between shrub growths on hummocks & freshwater availability in coastal saline wetlands #CUAHSI — Anne Jefferson (@highlyanne) July 26, 2016
• Christian Guzman used his Pathfinder fellowship to go to Honduras & work with Zamorano U students on ag influence on hydro #CUAHSI — Anne Jefferson (@highlyanne) July 26, 2016

In the afternoon I joined the field trip to Harpers Ferry Area. The field trip leader was Steve Kite from West Virginia University. Steve Kite provided a lot of information on the floods in the Shenandoah Valley and West Virginia, both during the walk but also to read at home. He prepared a table on the deadliest floods in West Virginia, where we could also see how many people were displaced or became homeless.

The result of #floods in the confluence of #shenandoah And #Potomac rivers. That was a bridge. #nofilter #CUAHSI

A photo posted by Denitza Voutchkova (@denitzav) on Jul 27, 2016 at 8:26am PDT

#shenandoah valley from yesterday’s field trip #nofilter, the church of Harpers Ferry #CUAHSI field trip

A photo posted by Denitza Voutchkova (@denitzav) on Jul 27, 2016 at 8:14am PDT

Lucy Nowell from U.S. Department of Energy was supposed to give a Keynote lecture, but there was change in program. So, instead, there were two speakers.

• Schedule change for Day 3: David Hyndman and Mike Gooseff will be speaking as this morning keynote. #CUAHSIbiennialpic.twitter.com/H7MtB8q5tS — CUAHSI (@CUAHSI) July 27, 2016

David Hyndman from Michigan State University with “Quantifying the impact of human activities on water sustainability”.

• Dave Hyndman: Quantifying Impact of Human Activities on Water Sustainability & Crop Yields Across the High Plains Aquifer #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016
• High Plains Aquifer has lost ~400 km3 of water due to use. That’s 25% in Central High Plains, 50% in Southern High Plains #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016
• Hyndman says that adding irrigation effects on atmosphere explains much of climate model misfit of regional precipitation. Wow. #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016
• Thanks to #climatechange areas where High Plains aquifer has been most over-used, the weather is going to get even drier. #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016
• Irrigation ~doubles yield for corn and wheat. Also, Yield increases through time, as f() of new corn cultivars and management. #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016
• Crop per drop: More ^ yield for corn than wheat per mm of irrigation applied + higher $ corn= lots of increase in corn irrigation. #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016
• David Hyndman: Central & S High Plains #aquifers are far from sustainable: decline in storage of 400km3 #CUAHSI — DenitzaV (@DenitzaV) July 27, 2016

Mike Gooseff from University of Colorado, Boulder with “The value of long-term data from the coldest, highest, driest, windiest continent”.

• Mike Gooseff on the value of long-term data from the coldest, highest, driest, windiest continent (McMurdo LTER, Antarctica) #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016
• The Dry Valleys in Antarctica are unglaciated & dry because katabatic winds. <5 cm of water equivalent precip per year. #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016
• Long Term Ecological Research (LTER) = 6 yrs + 6 yrs + 6 yrs… Short term experiments + network + long term data access for future #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016
• Dry Valleys Lakes (with permanent ice cover, open moats at edges) are closed basin, fed by streams draining glaciers. #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016
• In Dry Valleys soils there are nematodes and tardigrades (of course!). #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016
• Life is small in Dry Valleys ecosystem. Algal mats + phytoplankton + occasional seal that wanders in. Doesn’t end well for seal tho #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016
• Silicate weathering rates in Dry Valleys streams are among highest on Earth. Throws conventional wisdom on head. #CUAHSI — Anne Jefferson (@highlyanne) July 27, 2016

Ying Fan Reinfelder from Rutgers University and Martyn Clark from UCAR were instructors of the discussion.

It all started with the introduction of the main questions:

• Do we need a Community Hydrological Model?
• What do we want?
• What do we mean by “Community Models”?
• How do we get there?

As an outsider, what I understood from this discussion was that some people think that there is a need for a community hydrological model. Some think that this model should be flexible, so it allows different methods, processes, and scales to be incorporated. Others think that before talking about building a model or models, first, there should be a consensus on the important hydrological processes. Others were thinking about language, which language would this model be relying on. And so on… It seems to me, that this is an utopia, which both aims at unifying hydrological modeling, but at the same time, it should allow diversity, and add-on modules. The beginning of the discussion started with introduction of other communities. One of them was the R community. But in my opinion, R is a community, which is united by common programming language. Not by a scientific field.

So, I will be following up on this discussion in the future. It is interesting.

Here’s also a paper by Markus Weiler and Keith Beven published in Water Resources Research on this topic: “Do we need a Community Hydrological Model?”

• Interesting debate on community hydrology model with @cuahsi biennial conference scientists #cuahsi pic.twitter.com/kQ0K4c8Mp5 — David Guzman (@DavidGuzzy) July 27, 2016

This was one of the two parallel afternoon sessions. Session chair was Jeff Horsburgh from Utah State University. There were three talks in this session:

Touraj Farahmand from Aquatic Informatics with “Computing continuous record of discharge with quantified uncertainty using index velocity observations: A probabilistic machine learning approach”. Most important message was “include uncertainties in your analysis”. Uncertainties and rating curves, for example. Actually, he presented a new sensor (or technique) for continuous discharge measurements using index velocity observations and probabilistic machine learning approach.

• Touraj Farahmand (Aquarius informatics) on including uncertainty in hydrometry… Pass uncertainty from rating curves to Vmean #CUAHSI — DenitzaV (@DenitzaV) July 27, 2016

Matt Ables from KISTERS with “Ensuring data integrity for the National Water Model through densified monitoring”. Quite interesting talk about the data sources and how these are compiled together and used for validation of the National Water Model. At one point of his talk or in the discussion that followed, he said that there is no need to use telemetry anymore, when there is mobile network. He was talking about the new sensors, and how much smaller and smarter they are, and how they could send you an sms if there is higher water level that certain threshold that you have programmed. All is good, but there are still places without mobile network (example: the high elevation wilderness in Wyoming).

Jordan Read from USGS with “Community-sourced tools for turning Big [sensor] Data into Big Science”. He spoke a bit about the open R packages and if I’m not wrong at one point he said something like “Get your students exposed to programming” or it could have been “Get them exposed to R” or to open source. It was overall really interesting talk.

• Community-sourced tools for turning Big [sensor] Data into Big Science.getting over what’s holding us back. #cuahsi pic.twitter.com/dtMIczgaM1 — David Guzman (@DavidGuzzy) July 27, 2016