B12_Questions_and_Answers_CHIS_Monitoring_of_Seabirds_0002.pdf

Channel Islands National Park

Monitoring of Seabirds

140P8425Q0076

Questions and Answers

1. Can you please confirm the estimated level of effort for this project?

ANSWER: For this project, we are seeking assistance with the design of the inventory (approximately 80-120 hours of statistical/seabird expertise), sensor deployment (advise on where to located the sensors as well as optimized settings), and the analysis of 500 ARU detector months (approximately 36 hours of data per month per detector) collected in 2025-2029.

2. The RFP currently applies Service Contract Act (SCA) requirements; however, there are no identified labor categories within the RFP. The work outlined in the statement of work appears to align more closely with exempt, professional labor categories typically outside the scope of SCA coverage under 29 CFR § 541. Additionally, the Wage Determination provided does not include equivalent mappings for several of the proposed labor categories. Would the Government consider removing the SCA requirements for this procurement to better reflect the nature of the professional services being solicited and to ensure alignment with applicable labor regulations?

ANSWER: An exemption does not apply for this requirement. Please refer to FAR 22.1003-5.

3. What is the estimated Ceiling budget for this project?

ANSWER: The Government does not release its project estimates.

Machine Learning & Data Analysis

1. Could you share if there are any required accuracy levels or performance goals for the machine learning models? We do not have predefined accuracy requirements. The contractor should have rigorous methods for assessing model performance, and transparent strategies for they will handle variability in class performance to meet our objectives.

2. How will the government review the quality of the model development, especially if there aren't predefined performance targets? Quality assessment will be through manual vetting performed by the contractor’s seabird expert(s). We expect the contractor to be transparent about model performance and options to improve performance.

3. If the models don't perform well for rare species (like LESP, GUMU, RHAU), what other detection methods could we consider as backups? Manual vetting.

4. Would the government be able to provide past ARU datasets to help with model training or improvements? Yes. If so: a. What format would the data be in (such as WAV, MP3, or another type)? The data are recorded as W4V files can be opened and/or converted to uncompressed

WAV files using the free version of Wildlife Acoustic’s Kaleidoscope Analysis software.

b. Is the data already labeled or just raw recordings? The data will be provided as raw recordings on SD cards.

c. Could the data be stored locally or in the cloud? Either location is possible during the workflow, but all data will be returned to NPS.

5. Should the machine learning analysis run on local hardware (on premises), data rest at an educational institution for further analysis, or in a designated cloud setup? Ultimately, we would like the machine learning models to run on NPS local hardware, and possibly at a university.

6. What format would work best for the detection results—such as labeled audio clips, time stamps, or something else? Results should be in the form of labeled sounds files that retain the time and date stamp, as well as a summary report that provides species detected by date and location along with an accuracy assessment.

ARU Equipment, Data Format & Transfer

1. Could you describe the format and metadata details for the raw ARU recordings from the field? Please see the Wildlife Acoustics website describing the SongMeter4 for details. Metadata include the date and time, location, recording schedule, microphone type and settings, and sensor ID and firmware version.

2. Are there any guidelines for file naming, size limits, or compression that we should follow?

Not currently. Guidelines will be developed in collaboratively by NPS and the contractor.

3. Would NPS like us to calibrate or compare ARUs across islands to keep the data consistent?

NPS staff and collaborators will be monitoring the sensors and would like to be informed if any sensors fail. Calibration is not an anticipated need.

4. How often would ARU data need to be collected and shared? We anticipate that each year, ARUs will be deployed in February, and collected in June-August, depending on the length of the seabird breeding season.

5. Would the data transfer happen wirelessly, by hand collection, or a mix of both? Data will be recorded on SD cards. We anticipate that SD cards would be shipped to the contractor after being copied to an NPS cloud account.

6. What's the expected time between data collection and when it reaches the contractor?

Approximately 2-4 weeks, depending on transportation to the islands and the time to download/backup the data.

7. How would the acoustic data be shared—through secure online transfer, external drives, cloud storage (like AWS or Google Cloud), or NPS systems? We anticipate that SD cards would be shipped to the contractor after being copied to an NPS cloud account.

8. Would the contractor handle data storage and backups, or would NPS take care of that? NPS will handle long-term storage and backups, or we may work with the contractor to explore new storage/archival methods.

Field Deployment & Logistics

1. Would the contractor need to handle placing and retrieving the ARUs in the field, or just analyze data that NPS staff collects? NPS will deploy and retrieve the ARUs – the contractor is not expected to perform fieldwork. The contractor is expected to have expertise in ecological sampling design and seabird ecology.

2. If we do handle ARU deployment: a. Could NPS help arrange travel to and from the islands?

N/A

b. Who would be responsible for fixing any broken ARUs during fieldwork? NPS

3. Would NPS handle on-site checks of the audio detections (like confirming them in the field), or would that fall to us? NPS will maintain the ARUs.

4. Could NPS assign one main contact person for technical questions throughout the project?

NPS has a small team that monitors seabirds, and 2-4 people would be the contacts, depending on the topic.

5. Are there any specific rules for data security or handling, given the machine learning aspects?

Data should be stored securely and not shared with other entities. Data will remain the property of NPS.

Invoicing & Payment

1. As a small business, could you share if the final contract might include options for progress payments or milestone-based invoicing, per FAR 32 guidelines?

a. Yes, this can be added to the contract.

2. Would the contract potentially allow partial payments tied to key milestones, such as

ARU deployment/field notes, annual model summaries, or quarterly analysis reports.

a. Yes, similar to the answer above, this can be added to the final contract.