Trevor Amestoy

Trevor Amestoy

PhD Candidate at Cornell University studying water resource systems under deep uncertainty

WaterSciCon24: Themes, Highlights, and Links

12 minute read

Published:

Last week I had the chance to travel to St. Paul, Minnesota, for the WaterSciCon conference hosted by AGU and CUAHSI. In this post, I give some of my own reflections on the core themes from the conference and share links and resources that were highlighted during the sessions.

Many of the things included here could constitute a whole blog post by themselves, but I have neither the time nor experience with these topics to do that. I will only give a cursory overview of each, but think that it is worthwhile for researchers to be aware of what’s out there and what direction things are going.

I break up my conference highlights into three common themes:

  • Open datasets, models, and tools
  • Reproducibility of computational experiments
  • Strengthening community computational potential

I should note that a decent portion of the conference attendees specialized in water quality or biogeochemical research, but I largely ignore those topics here given my background and the fact that I can only be in one conference room at a time. If I did not include mention of your work, I apologize - nothing personal!

Open datasets, models, and tools (oh my)

If it isn’t open source did it really happen?

OpenET

I heard lots of conversation around OpenET, which provides remote sensing based estimates of evapotranspiration at different temporal and spatial scales. OpenET positions themselves as “Filling the biggest data gap in water management”. This data can be downloaded using their website or accessed using an API key.

The strengths of OpenET seem to be that they (1) assimilate weather station data as well as multiple gridded weather datasets (e.g., DAYMET, NLDAS, PRISM and others) into their methodology and (2) use an ensemble of models to generate the ET estimates.

NASA ARSET

On the topic of remote sensing, I heard a few presentations from NASA about there numerous satellites and data products. While a lot of people are eagerly waiting for data to be released from the Surface Water and Ocean Topography (SWOT) mission, there are many datasets that are already being leveraged in cool and insightful ways. One major strength of NASA’s remote sensing data, compared to data from USGS for example, is that these datasets tend to have near global coverage and can serve as a valuable resource for researchers working in data-scarce regions around the world.

Dr. Amita Mehta from NASA’s Goddard Space Flight Center gave a presentation which highlighted the value of these global datasets and raised awareness of the Applied Remote Sensing Training (ARSET) program. These programs are run online over a few days and are open for free registration.

Perhaps the most relevant training for this audience is the Drought Monitoring, Prediction, and Projection using NASA Earth System Data training program running from July 23 through August 1, 2024.

USGS water resource models and tools

There was a large of USGS researchers at the conference who showed that they are continuously updating and evolving their computational tools to support hydrologic research and decision making (Thank you, USGS!). A commonality across these USGS tools is that many of them are improving the accessibility, modularity, and modernization of existing codebases. Given the number of models and tools being presented, I went ahead and gave them their own section.

pywatershed

Jacob LaFontaine (USGS), who has continued to support the development and advancement of the Pricipitation-Runoff Modeling System (PRMS), the National Hydrologic Model (NHM), and CONUS-scale hydrologic modeling more generally, talked about pywatershed. This Python based model is intended to “modernize important, legacy hydrologic models at the USGS, particularly the Precipitation-Runoff Modeling System (PRMS, Markstrom et al., 2015) and its role in GSFLOW (Markstrom et al., 2008).”

This model is still under active development, but it is promising to see this shift toward ease-of-use and modularity for these hydrologic modeling frameworks which can be intimidating for beginners.

dataretrieval package for Python

In my own work, and on this blog, I have previously advocated for pygeohydro from the HyRiver suite for USGS data retrieval. I am a big fan of these tools, but now there is a USGS-made Python package that serves as an alternative.

The dataretrieval package is used to pull USGS and EPA data directly from their National Water Information System (NWIS) webservices.

This package was developed from the R version which has a similar, camel-case variant of the name: dataRetrieval.

hyswap (HYdrologic Surface Water Analysis Package)

During the poster session I met Lee Stanish who, along with other folks from the USGS, recently released a new Python package for processing and visualizing hydrologic timeseries data called hyswap.

In their own words, hyswap is:

“*… a Python package which provides a set of functions for manipulating and visualizing USGS water data. Specifically, a number of functions for calculating statistics (e.g., exceedance probabilities, daily historic percentiles) and generating related plots (e.g., flow duration curves, streamflow duration hydrographs) are available.”

Personally, I spend a decent bit of time writing and re-writing the same workflows to process and visualize basic hydrologic timeseries statistics. I’m looking forward to checking this out, and hopefully spending less time with this boiler-plate style coding.

mosartwmpy

First these was the Model for Scale Adaptive River Transport (MOSART; H.-Y. Li et al., 2013). Then there was MOSART-WM (Water Management; Voisin et al., 2013). Then there was MOSART-WM-heat (H.-Y. Li et al., 2015). THEN there was MOSART-WM-hydropower (Zhou et al. 2018). Each of these being extensions on the Fortran version of MOSART.

Now, there is mosartwmpy, a Python based version of the MOSART-WM model which combines river routing processing and reservoir management in a single model.

Again, the goal here is to improve the convenience of developing and working with the model, with this implementation focused on being “intuitive, lightweight, and well documented, while still being highly interoperable.”.

Reproducibility of computational experiments

Lindsay Platt from CUAHSI gave a great talk on the importance of reproducibility in computation water science research. Dr. Kirstie Whitaker’s talk titled “The Turing Way: Sharing the responsibility of reproducibility” (2019). I watched this on my way back from the conference and can confirm that it gives some good food for thought to motivate and guide future research workflow organization.

Lindsay also gave a few general tips for improving your own workflow reproducibility, including:

  • Project organization
    • Numbering all scripts to indicate the sequencing of execution
    • Use of descriptive filenames
    • Maintaining a “single source of truth” for all data and code being used
  • Ensuring that all data decisions and transformations are being done with scripts
  • Maintaining a culture of reproducibility
  • Shifting the incentives
    • Hiring dedicated data or development leaders
    • Publishing data separately with alternative author lists and ordering

One package that was discussed repeatedly in the context of reproducibility was Snakemake (documentation) which can be used to construct entire experimental workflows with a single point-of-entry. Additionally, this Snakemake is able to identify which parts of the workflow have been changed, what parts are impacted by those changes, and then only re-run the necessary subset of the workflow needed to account for the changes.

I know from talking with USGS researchers that this is becoming standard in their own work, and I’m definitely keen on implementing this in my next project.

Strengthening community computational potential

As a community, we have developed countless models… but are they being used to their full potential?

Given that this conference was co-hosted by the Consortium of Universities for the Advancement of Hydrologic Science, Inc., AKA CUAHSI there was a big emphasis on strengthening our community’s potential through collaborate and sharing of resources. CUAHSI has made it their mission to advance collaborative modeling by developing a centralized resource repository called HydroShare.

From speaking with others at the conference and in workshops, one of the most common pain-points when trying to learn a new model or collaborate on a multi-model project is the lack of standardized convention and APIs.

One group trying to improve the community collaborative potential is the Community Surface Dynamics Modeling System (CSDMS) community. The CSDMS program is an NSF-funded program that promotes the development and dissemination of software used for a variety of different environmental processes, with an emphasis on modularity and integration of different models.

In my opinion, the most important work that CSDMS is doing is to develop the Basic Model Interface (BMI) which attempts to confront the model standardization problem head-on. The BMI provides formal standards for variable naming conventions, model functionality, and other aspects of the API.

I think the BMI team does a great job of illustrating the value of standards in this excerpt from their website:

“When you climb in the driver’s seat of an unfamiliar car, you are nonetheless presented with a familiar sight. Whatever the make or model may be, we take it for granted that the vehicle will provide a steering wheel, brake pedal, and speedometer, alongside the various other controls and readouts that are common to essentially all cars and trucks on the planet. … We believe that numerical models, and the sub-components that make up those models, should offer a similar kind of standardization.”

I think this is a great vision and a worthwhile goal for the community.

Concluding thoughts

I personally presented in the Integrated Hydro-Terrestrial Modeling (IHTM) session. The IHTM project brings together many domain experts from various agencies and institutions with the goal of formalizing wholistic risk assessment frameworks that account for interactions and feedbacks across the human-natural water cycle.

The work being done through IHTM definitely emphasizes the importance of all three of the themes outlined above. Through my work on the IHTM project it’s become increasingly clear to me that water resource systems research for the future will greatly benefit from (if not necessitate) prioritization of these themes:

  • Open datasets, models, and tools
  • Reproducibility of computational experiments
  • Strengthening community computational potential

I’m glad to have seen so much emphasis on the importance of collaborative work at this conference.

Other notable mentions

There were a few other notable mentions that I wanted to include in this post, but which did not so easily fit into one of my three themes above.

References

Markstrom, S. L., Niswonger, R. G., Regan, R. S., Prudic, D. E., & Barlow, P. M. (2008). GSFLOW-Coupled Ground-water and Surface-water FLOW model based on the integration of the Precipitation-Runoff Modeling System (PRMS) and the Modular Ground-Water Flow Model (MODFLOW-2005). US Geological Survey techniques and methods, 6, 240.

Markstrom, S. L., Regan, R. S., Hay, L. E., Viger, R. J., Webb, R. M., Payn, R. A., & LaFontaine, J. H. (2015). PRMS-IV, the precipitation-runoff modeling system, version 4 (No. 6-B7). US Geological Survey.

Li, H., Wigmosta, M. S., Wu, H., Huang, M., Ke, Y., Coleman, A. M., & Leung, L. R. (2013). A physically based runoff routing model for land surface and earth system models. Journal of Hydrometeorology, 14(3), 808-828.

Voisin, N., Li, H., Ward, D., Huang, M., Wigmosta, M., & Leung, L. R. (2013). On an improved sub-regional water resources management representation for integration into earth system models. Hydrology and Earth System Sciences, 17(9), 3605-3622.

You May Also Enjoy