Our group's new paper #2 on differentiable modeling (DM), focusing on stream temperature. Recall again that DM allows you to use physical equations to narrow your question (represented by an NN) to an interpretable scope. Can we thus identify better priors?

New semester gift for environmental ML newbies!!! We put together a website that collects our model codes, benchmarks, docs and tutorials. mhpi.github.io Our line of code is collectively called HydroDL.

Regurgitation of underdiscovered paper from last year: global soil moisture product. gmd.copernicus.org/articles/16/15…

Review: River water quality shaped by land–river connectivity in a changing climate by Li Li 李黎 et al go.nature.com/3Tt0yk5 River water quality affects water availability. This Review assesses the impacts of climate change & extremes on water quality

Our paper is one of the most downloaded articles in GRL 2022-2023. Thank you very much for your interest in our research. Chaopeng Shen Farshid Rahmani Kathryn Lawson

Last drum!!! Abstracts welcome!! Richland WA is 3 hours from Seattle or 3.5 hours from Portland, OR. Thanks PNNL for hosting it!

⚡️Hydropower capacity factors trending down in the United States nature.com/articles/s4146…

Excited to share our latest paper on suspended sediment concentration (SSC)! 🌍 We developed a continental-scale SSC model and discovered that it performs best in the humid Eastern US, where SSC-streamflow correlation is high, and SSC is low. Read more: authors.elsevier.com/a/1jLOh52cuf9dX

A couple of weeks ago, I presented a workshop on creating a differentiable model at the Texas A&M University WATER Summer School. That presentation is now available on YouTube youtube.com/watch?v=NhCebO… Thanks, Debasish Mishra , for organizing! and Chaopeng Shen for providing some helpful slides!

Can differentiable models incorporate implicit numerical schemes to enhance performance and correct parameter distortions caused by numerical errors? Yes, they can! Explore our latest publication: doi.org/10.5194/hess-2…

A huge preprint and lots of work to get us here. High-resolution national-scale water modeling. High-quality data-informed hydrologic simulations at MERIT basin level (median ~37 km2). Coord effort from my group and lead-authored by Yalan Song Tadd Bindas bit.ly/3NnqDNB

This work develops high-resolution, multiscale, differentiable models for national-scale water modeling. Our continental high-resolution dataset is now available online! The code will be released upon the acceptance of the paper.

It’s AGU time, and I am excited to meet you all there. I am having a talk on Tuesday, about multi-objective differentiable modeling and a poster about large scale recharge prediction on Monday.

Check out our latest paper!

Hydrograd: A differentiable hydrodynamics modeling framework for scientific machine learning. It solves the universal shallow water equations (USWEs). Examples: flow resistance learning and bathymetry inversion. Code: github.com/psu-efd/Hydrog… Manuscript: arxiv.org/abs/2502.12396

A massive paper and a generational leap in national-scale high-resolution seamless hydrologic simulation. We published the continental-scale simulation dataset. If you have interests, please inquire! doi.org/10.1029/2024WR… /run