RFI_USGS_PWS_PINGMapper_1.pdf

U.S. Geological Survey California Water Science Center Performance Work Statement

1.0 INTRODUCTION

Consumer-grade, low-cost, side-scan sonar enables efficient, repeat imaging of river substrate conditions. However, manual post-processing of sonograms and classification of sonar mosaics are extremely labor-intensive due to the large volume of data. PINGMapper (Bodine et al., 2022;

https://cameronbodine.github.io/PINGMapper/) is an open-source Python interface that automates the post-processing of sonar recordings and segmentation and classification of sonar mosaics using deep learning (DL) models (Bodine et al., 2024). Final products are georeferenced sonar mosaics and substrate maps with seven classes (fines ripples, fines flat, cobble boulder, hard bottom, wood, other, and shadow. This scope of work (SOW) outlines modifications to the PINGMapper source code for use in coarse-grained rivers, as well as post-processing and analysis of existing datasets to assess substrate conditions to support USGS science studies in the Klamath River, California.

2.0 BACKGROUND

Recently, the largest dam removals in U.S. history occurred on the Klamath River. In 2023, one low-head dam was removed; in 2024, three large dams were removed simultaneously. During reservoir drawdowns and dam removal, approximately 1.2 million metric tons of fine-grained (silt/clay), organic‑rich sediments were released downstream into a coarse-grained (gravel/cobble) river environment. In 2022 and 2023, before the dam removals, the USGS surveyed 120 km of the mainstem Klamath River using consumer-grade, low-cost side-scan sonar (Humminbird 710292-1 XM 14 HW MSI T SOLIX) and repeated the surveys in 2025 after the dam removals. This SOW addresses two important questions related to the dam removals. What was the fate of the fine-grained reservoir sediments released into the coarse-grained river environment? How did benthic substrates and ecological habitats change?

3.0 SCOPE

The required services funded under this contract include modifications of the PINGMapper source code, integration of auxiliary high-resolution GPS data for sonogram georeferencing, and change-detection analysis aligned with USGS research objectives to address the fate of reservoir sediments and effects on benthic substrates and ecological habitats in the downstream river environment.

The contractor shall provide monthly progress reports by email. During the first 3 months of the award period, the contractor shall meet with USGS staff for bimonthly virtual discussions. After the first 3 months, the contractor and USGS staff will schedule virtual meetings, as needed, to ensure product delivery meets QA/QC requirements and project timeline (described under 7.0 Technical Requirements)

3.1 AREA OF INTEREST

The area of interest (AOI) includes 120 river kilometers along the mainstem of the Klamath River in California.

4.0 SUPPORTING INFORMATION

https://cameronbodine.github.io/PINGMapper/

• Bodine, C.S., Buscombe, D., Hocking, T.D. (2024). Automated river substrate mapping from sonar imagery. Journal of Geophysical Research – Machine Learning and Computation, 1, e2024JH000135. https://doi.org/10.1029/2024JH000135

• Bodine, C.S., Buscombe, D., Best, R.J., Redner, J.A., & Kaeser, A.J. (2022). PING‑Mapper:

Open‑source software for automated benthic imaging and mapping using recreation‑grade sonar. Earth and Space Science, 9, e2022EA002469. https://doi.org/10.1029/2022EA002469

5.0 PLACE OF PERFORMANCE

• All work will take place at the contractor’s facility.

6.0 TECHNICAL REQUIREMENTS

This project requires expert proficiency level with:

• Python programming and

• PINGMapper source code.

The contractor shall modify the PINGMapper source code for use in coarse-grained rivers. Modifications will include: 1) incorporating confidence thresholds to support selection of appropriate class thresholds and optimization of classified substrate maps, and 2) incorporating auxiliary high-resolution GNSS data for georeferencing sonar mosaics. The contractor will use the modified PINGMapper code to 3) post-process existing sonar recordings and produce optimized substrate maps for the AOI, 4) conduct change detection analysis for the AOI, and 5) deliver USGS-ready data release and supporting information for a manuscript.

6.1 TASK 1 Confidence Thresholding

The contractor shall modify the PINGMapper source code to 1) compute per-patch confidence scores directly from model outputs for substrate classes, 2) summarize the per-patch multi-class statistics (mean, minimum, maximum, standard deviation), and 3) set low-confidence patches to ‘Unknown’ to minimize false positives and determine optimized class-specific thresholds for coarse-grained rivers.

• Deliverable: Exportable patch-level summary statistics of substrate class likelihoods (e.g.

confidence scores) in *csv format.

• Method: Upload supporting data and information and modified source code to a GitHub repository.

• Date: 3 months from the date of the award

6.2 TASK 2 Auxiliary GPS Integration

The contractor shall modify the PINGMapper source code to 1) accept auxiliary sources of high-resolution GNSS data (*csv format) for georeferencing PINGMapper products, 2) match auxiliary GPS points to sonar recordings via timestamps and account for clock offsets, when necessary, 3) update PINGMapper workflows to utilize the new GPS data source in subsequent geoprocessing workflows and data outputs.

• Deliverable: PINGMapper functionality to ingest auxiliary GPS sources and propagate positioning into metadata in the form a python script.

• Method: Upload supporting data and information and modified source code to a GitHub repository.

• Date: 5 months from the date of the award

6.3 TASK 3 Sonar Scan Processing

https://doi.org/10.1029/2024JH000135 https://doi.org/10.1029/2022EA002469

The contractor shall 1) batch-process existing sonar recordings collected by the USGS, 2) apply optimized Klamath-specific confidence thresholds, and 3) export optimized substrate maps for 75 sites (approximately 0.5-1.0 km reaches) collected across three years (2022, 2023, 2025), and 4) load optimized substrate maps into a file geodatabase (*fgb) and compress into a zip file.

• Deliverable: Georeferenced sonar mosaics and fine-tuned substrate maps for 75 sites across three survey years; auxiliary GPS integrated for 2025 scans in the form of georeferenced rasters and polygon features.

• Method: All data and supporting information shall be delivered by electronic transfer to be published on a USGS repository.

• Date: 7 months from the date of the award

6.4 TASK 4 Reach-Level Change Analysis

The contractor shall 1) determine categories suitable for change detection and communication of study findings, 2) calculate reach-level class proportions for each scan year, 3) quantify inter-annual changes, and 4) complete robust QA/QC and error analysis.

• Deliverable: Change analysis and summary tables and graphs by scan year in an EXCEL spreadsheet.

• Method: All data and supporting information shall be delivered by electronic transfer to be

• Date: 8 months from the date of the award

6.5 TASK 5 Deliverables

The contractor shall 1) curate final mosaics, maps, and metadata into a USGS-ready data release package with accuracy assessment, and 2) provide supporting information for a manuscript.

• Deliverable: USGS-ready data release package with accuracy assessment and supporting information (figures and tables in EXCEL spreadsheets, and text in a WORD document)

• Method: All data and supporting information shall be delivered by electronic transfer to be

• Date: 11 months from the date of the award

7.0 DATA RIGHTS

All raw and processed data, as well as derived maps and analyses produced under this contract, will be delivered to USGS. Source code enhancements to PINGMapper will remain open‑source under the existing license, with appropriate documentation.

8.0 GOVERNMENT FURNISHED INFORMATION

The USGS will provide the contractor with the following:

• Side scan sonar recordings collected in 2022, 2023, and 2025

• Auxiliary GPS data collected in 2025

9.0 PERIOD OF PERFORMANCE

• Date of Award through 1 year.

U.S. Geological Survey