Snow-Extrapolation - ML Algorithm Optimization for Operational Snow-Water-Equivalent Estimation

This project builds on the ongoing Cooperative Institute for Research-to-Operations in Hydrology (CIROH) National Snow Model project to provide operational, near-real-time snow-water-equivalent (SWE) estimates for updating the state of the snowpack within the National Water Model (NWM).

The Getting started file will help new users create a virtual environment and install the correct packages using the requirements.text.

Collaborators

List all participants on the project.

• Ryan C. Johnson, PhD - The University of Alabama
• Tahi Wiggins - tahiwiggins - University of Maine
• Ziqi Yin - icezyin - University of Colorado, Boulder
• Zeeshan Asghar - University of Utah
• Tony Castronova, PhD - CUAHSI
• Irene Garousi-Nejad, PhD - CUAHSI
• Savalan Naser Neisary - University of Alabama
• Akila Sampath-University of Maryland Baltimore County
• Steven Pestana, PhD - University of Washington
• Md Shahabul Alam, PhD - University of Alabama

National-Snow-Model Summary

Snow-derived water is a critical hydrological component for characterizing the quantity of water available for domestic, recreation, agriculture, and power generation in the western United States. Advancing the efficiency and optimization of these aspects of water resources management requires an enhanced characterization of the snow state variable, particularly the essential global inputs of snow-water-equivalent (SWE), peak SWE, and snowmelt onset for hydrological models. While physically-based models that characterize the feedbacks and interactions between influencing factors predict SWE values well in homogeneous settings, these models exhibit limitations for CONUS-scale deployment due to challenges attributed to spatial resolution, landscape heterogeneity, and computational intensity. Leveraging a collaborative partnership between the Alabama Water Institute (AWI) at the University of Alabama (UA) and the University of Utah (UU), we address these limitations through the NSM as a data-driven machine learning (ML) platform with a modular structure to account for the heterogeneity of climate and topographical influences on SWE across the western United States. The model consists of twenty-three regionally specific sub-models tailored to the unique topography and hydroclimate phenomena in the Western U.S., exhibiting an RMSE less than 8 cm and a coefficient of determination approaching 0.99 on predictions spanning the 2013-2017 training period. The NSM pipeline assimilates nearly 700 snow telemetry (SNOTEL) and California Data Exchange Center (CDEC) sites and combines with processed lidar-derived terrain features for the prediction of a 1 km x 1 km SWE inference in critical snowsheds.

eScience Hack Week Project Summary: Description and Goals

The goal of the Hack Week project participants is to refine NSM prediction performance in the Sierra Nevada mountains through the exploration of different ML algorithms. The current Sierra Nevada core ML algorithm is a deep neural network, and while it demonstrates satisfactory prediction skill, there still are large errors in predction and room for improvement identified by a rigorous evaluation of the model. The most notable error appear at locations below 1,500m and above 2,900, ephemeral and alpine regions, respectively, and during the snow melt period. We encourage participants to explore other ML algorithms, such as other neural networks, Long-Short-Term-Memory, tree-based, and even simple regression algorithms such as Ordinary Least Squares. Model evaluation will form a critical element of determing differences in algorithm performance, and the Standardized Snow Water Equivalent Evaluation Tool (SSWEET) will serve as a standardized method to evaluate the different algorithm architectures.

There are several long-term goal of the eScience Hack Week Snow-Extrapolation, and participants can expect the following:

• Develop of a publication stemming the the Hack Week project
• Connection with large-scale NOAA Office of Water Prediction (OWP) modeling projects
• Grow your academic network
• Contribute to a project with the potential for integration into Federal water resources operations
• Documented contribution to build your

The project deliverable will be the demonstration of a ML algorithm for the Sierra Nevada region models of the NSM that measurable improves prediction skill. Each algorithm will have a unique model folder with Jupyter Notebook file that includes clear documentation on the training, algorithm development, and model prediction. Complementing the Notebook on algorithm training, a SSWEET Jupyter Notebook will load model predictions for the approapriate testing period to determine model performance. All model predictions will be included in the respective model forlder. The ultimate goal of each folder is to function a tutorial to get new users introduced and familiar with different ML algorithms, their application, and their respective strengths and weakness.

Creating Stable CONDA Environment

• Go to home directory

cd ~

• Create a envs directory

mkdir envs

• Create .condarc file and link it to a text file

touch .condarc

ln -s .condarc condarc.txt

• Add the below lines to the condarc.txt file

# .condarc
envs_dirs:
 - ~/envs

• Restart your server

Files

• .gitignore
  Globally ignored files by git for the project.
• environment.yml
conda environment description needed to run this project.
• README.md

contributors

Each team member can create their own folder under contributors, within which they can work on their own scripts, notebooks, and other files. Having a dedicated folder for each person helps to prevent conflicts when merging with the main branch. This is a good place for team members to start off exploring data and methods for the project.

notebooks

Notebooks that are considered delivered results for the project should submit a pull request from their personal working repository. Upon review, the contributions of the participant will be integrated into this repository.

scripts and requirements.txt

All helper utilities and updated requirements.txt to replicate the working environemnt should be included in the project submissin package.

Data

Briefly describe the data that will be used here (size, format, how to access).

All data will be provided to participants with cloud access to either the CIROH publicly accessible Amazon Web Services (AWS) S3 storage or via Box. Project data includes:

• NASA Airborne Snow Observator (ASO) LiDAR-derived SWE estimates
• Copernicus 90m Digitial Elevation Model (DEM)
• Natural Resources Conservation Service (NRCS) Snow Telemetry (SNOTEL) monitoring station SWE observations
• Visible Infrared Imaging Radiometer Suite (VIIRS) fraction snow covered area (fSCA)

Participants will be provided pre-processed model training dataframes of approximately 7,000 1-km grid locations to limit the amount of time spent on data processing tasks. While a key and essential component of any ML objective, the focus of the eScience Hackweek is to optimize the ML algorithm component of the NSM.

Existing methods

The current NSM uses the same 7-layer deep neural network for each of the 23 regions. While regionally-optimized Light Gradient Boosted Models (LGBM) and Extreme Gradiented Boosted (XGBoost) algorithms were part of the original model development phase, the standard 7-layer multilayered perceptron (MLP) neural network demonstrated high prediction skill than the tree-based models, notably during prolong snow-drought. Existing model research and development explores and compares different algorithms, with the best performing algorithm transitioning to become optimized.

Proposed methods/tools

A two-pronged approach would be ideal to improve model performance: 1) Feature Engineering and selection and 2) Algorithm optimization. With the focus of the Hack Week on algorithm optimization, this will be the focus. The project is open to explore any ML algorithm, whether simple or complex. Examples include RNNs, LSTMs, physcis-informed ML, etc.

Tasks

The following tasks are expected to be completed before the Hack Week and will be made available to participants a week before the event:

• Create your own Github account if you do not have one
• Fork this repository to your personal account
• Download Git Desktop and clone the repository to your machine. Note, if you are comfortable with the command line process using git, this is an acceptable approach
• Run through the provided steps in the Getting Started page to set up your appropriate virtual environment
• Successfully run the Sierra Nevada region(s) of the NSM
• Review and develop and understanding the model training and development steps
• Review ML literature and come prepared with algorithms/approaches to enter the Hack Week with explicit algorithm and coding goals.