Model archive component 4, Coarse Model, in: Downscaling and multi-scale modeling of stream temperature in five watersheds of the Delaware River Basin, 1979-2021

Janet Barclay Yingda Fan Lauren Koenig Runlong Yu Yiming Sun Yiqun Xie Xiaowei Jia Alison Appling 20250305 Model archive component 4, Coarse Model, in: Downscaling and multi-scale modeling of stream temperature in five watersheds of the Delaware River Basin, 1979-2021 Vector digital data Online (digital release) U.S. Geological Survey data release https://doi.org/10.5066/P1UP5DXN This model archive component contains model weights, inputs, outputs, and performance metrics for the source coarse model for which downscaling was desired. Some methods in Fan et al. (2025b) explore methods for downscaling from this source coarse model, while others explore different uses of these coarse-resolution source data in conjunction with fine-resolution data (see model archive component 2, Model Inputs, for the fine-resolution data). The parent model archive (<a href="https://www.sciencebase.gov/catalog/item/66787f3ed34efbe36238c80a">Fan et al. 2025a</a>) provides all data, code, and model outputs used in the corresponding manuscript (Fan et al. 2025b) to test machine learning (ML) methods for downscaling and multi-scale modeling of stream temperature to combine an ML model and/or input data at coarse spatial resolution with an ML model and/or input data at fine spatial resolution to predict stream temperatures at fine spatial resolution in a watershed. The data are organized into these child items: <li><a href="https://www.sciencebase.gov/catalog/item/6682f4f8d34e57e93663d655"> 1. Geospatial Information </a>- Stream reach and catchment shapefiles </li> <li><a href="https://www.sciencebase.gov/catalog/item/6682f50bd34e57e93663d65a"> 2. Model Inputs </a> - Meteorological data, river network matrices, and stream temperature observations </li> <li><a href="https://www.sciencebase.gov/catalog/item/6682f522d34e57e93663d65e"> 3. Model Code </a>- Python files and README for reproducing model training and evaluation </li> <li><a href="https://www.sciencebase.gov/catalog/item/6682f545d34e57e93663d665"> [THIS ITEM] 4. Coarse Model </a>- Trained coarse stream temperature model to be downscaled </li> <li><a href="https://www.sciencebase.gov/catalog/item/6682f556d34e57e93663d668"> 5. Model Outputs </a>- Model simulation outputs and evaluation metrics </li> The publication associated with this model archive is: Fan, Yingda, Runlong Yu, Janet R. Barclay, Alison P. Appling, Yiming Sun, Yiqun Xie, and Xiaowei Jia. 2025. "Multi-Scale Graph Learning for Anti-Sparse Downscaling." In Proceedings of the AAAI Conference on Artificial Intelligence. Vol. 39. Washington, DC, USA: AAAI Press. This data compilation was supported by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, Environmental System Science Data Management Program, as part of the ExaSheds project, under Award Number 89243021SSC000068. Work was also supported by the U.S. Geological Survey, Water Availability and Use Science Program. Water quality modeling methods development 19791001 20210930 See publication date Complete None planned -76.3879905924101 -74.380785128688 42.4544544671721 38.7894548062558 none machine learning deep learning water resources water temperature streams modeling downscaling mathematical simulation ISO 19115 Topic Category environment inlandWaters USGS Metadata Identifier USGS:6682f545d34e57e93663d665 Department of Commerce, 1995, Countries, Dependencies, Areas of Special Sovereignty, and Their Principal Administrative Divisions, Federal Information Processing Standard (FIPS) 10-4, Washington, D.C., National Institute of Standards and Technology United States US U.S. Department of Commerce, 1987, Codes for the identification of the States, the District of Columbia and the outlying areas of the United States, and associated areas (Federal Information Processing Standard 5-2): Washington, D. C., NIST Delaware DE Maryland MD New Jersey NJ New York NY Pennsylvania PA none None, but see "Distribution Liability"/"distliab" below. Users are advised to read the associated publication thoroughly to understand appropriate use and data limitations. Janet Barclay U.S. Geological Survey Research Hydrologist Mailing and Physical

U.S. Geological Survey

Reston VA 20192 U.S.A. NA jbarclay@usgs.gov This data compilation was supported by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, Environmental System Science Data Management Program, as part of the ExaSheds project, under Award Number 89243021SSC000068. Work was also supported by the U.S. Geological Survey, Water Availability and Use Science Program. These code files were tested on high performance computing systems and laptops at the University of Pittsburgh and the United State Geological Survey. Operating systems included Windows, linux, and OSX. The open source languages R and Python were used on all systems. Fan, Yingda Runlong Yu Janet Barclay Alison Appling Yiming Sun Yiqun Xie Xiaowei Jia 2025 Multi-Scale Graph Learning for Anti-Sparse Downscaling http://doi.org/TBD/TBD No formal attribute accuracy tests were conducted. Not applicable Not applicable A formal accuracy assessment of the horizontal positional information in the dataset was not conducted. Barclay, Janet R. Simon N. Topp Lauren Koenig Margaux J. Sleckman Jeffrey M. Sadler Alison P. Appling 2023 Model Code, Outputs, and Supporting Data for Approaches to Process-Guided Deep Learning for Groundwater-Influenced Stream Temperature Predictions: U.S. Geological Survey data release https://doi.org/10.5066/P9KO49OT online 2023 publication date Barclay et al. (2023) Code files in the river-dl python package that were reused for this study with some minor modifications. In turn, the code files in this source were derived from code published by Sadler et al. (2021). Sadler, J.M. Appling, A.P. Read, J.S. Oliver, S.K. Jia, X. Zwart, J.A. Kumar, V. 2021 Multi-task Deep Learning for Water Temperature and Streamflow Prediction (ver. 1.1, June 2022): U.S. Geological Survey data release https://doi.org/10.5066/P9U0TG8L online 2021 publication date Sadler et al. (2022) Original version of the river-dl python package, which was reused and modified by Barclay et al. (2023) and this study. Oliver, S.K. Sleckman, M.J. Appling, A.P. Corson-Dosch, H.R. Zwart, J.A. Thompson, T.P. Koenig, L. White, E. Watkins, D. Platt, L.R. Padilla, J.A. Sadler, J.M. 2022 Data to support water quality modeling efforts in the Delaware River Basin https://doi.org/10.5066/P9GUHX1U online 2022 publication date Oliver et al. (2022) Compiled temperature observations and model driver data for the Delaware River Basin at the resolution of the NHM Code files from Sadler et al. (2022) and Barclay et al. (2023) and input data from this data release (NHM.zip in "2. Model data") was used to generate the coarse-resolution (NHM) model contained in this model archive component. Predictions from this coarse-resolution model were used in some models described by Fan et al. (2025b). For completeness, the inputs, trained model weights, and evaluation metrics for the coarse-resolution model are also included in this model archive component. 20250304 U.S.A. Vector String 456 1e-06 1e-06 Decimal degrees WGS84 WGS_1984 6378137.0 298.257 4_coarse_file_dictionary.csv Comma-separated values file with one row per file contained in code.zip. Describes the format, contents, and source of each file in code.zip. This study file-name File path relative to top-level "code" folder in code.zip This data release NA NA NA file-type File type, given as file extension or "dir" for directory This data release NA NA NA description Description of the file contents This data release NA NA NA file-source File source, usually some combination of this model archive (Rahmani et al. 2023a) and/or Shen (2020) This data release NA NA NA output_NHM_DRB.zip Zip archive of all model weights, inputs, outputs, and performance metrics for the source coarse model for which downscaling was desired. Users will need to unzip/decompress .zip file to view contents. Files are described in 4_coarsemodel_file_dictionary.csv and also as attributes of this zip file. This study output_NHM_DRB/pretrained_weights.pth Pytorch model weights file: Model weights after pretraining This model archive NA NA NA output_NHM_DRB/finetune_log.csv Comma-separated values: Log of model finetuning This model archive NA NA NA output_NHM_DRB/tst_preds.feather Feather file: Feather file of temperature predictions for the test partition This model archive NA NA NA output_NHM_DRB/finetuned_weights.pth Pytorch model weights file: Model weights after finetuning This model archive NA NA NA output_NHM_DRB/overall_metrics.csv Comma-separated values: Metrics of overall model performance This model archive NA NA NA output_NHM_DRB/pretrain_log.csv Comma-separated values: Log of model pretraining This model archive NA NA NA output_NHM_DRB/prepped.npz NumPy compressed file: Numpy compressed archive of model inputs and training data This model archive NA NA NA output_NHM_DRB/val_preds.feather Feather file: Feather file of temperature predictions for the validation partition This model archive NA NA NA output_NHM_DRB/overall_metrics_noised.csv Comma-separated values: Metrics of overall model performance for permutation-based feature importance This model archive NA NA NA output_NHM_DRB/trn_preds.feather Feather file: Feather file of temperature predictions for the training partition This model archive NA NA NA U.S. Geological Survey GS ScienceBase mailing address

Denver Federal Center, Building 810, Mail Stop 302

Denver CO 80255 United States 1-888-275-8747 sciencebase@usgs.gov Unless otherwise stated, all data, metadata and related materials are considered to satisfy the quality standards relative to the purpose for which the data were collected. Although these data and associated metadata have been reviewed for accuracy and completeness and approved for release by the U.S. Geological Survey (USGS), no warranty expressed or implied is made regarding the display or utility of the data on any other system or for general or scientific purposes, nor shall the act of distribution constitute any such warranty. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government. Model files are archived within a single zip file, output_NHM_DRB.zip. Contents of output_NHM_DRB.zip are described in the metadata file (4_coarse.xml) and also in the comma-separated-values file 4_coarsemodel_file_dictionary.csv. https://doi.org/10.5066/P1UP5DXN None 20250305 Alison Appling U.S. Geological Survey Ecologist Mailing and Physical

U.S. Geological Survey

Reston VA 20192 U.S.A. NA NA aappling@usgs.gov FGDC Content Standard for Digital Geospatial Metadata FGDC-STD-001-1998