Data_refinement_Q_A_0001.pdf

140G0323Q0164amd0001 Page 2 of 3

Data refinement Q&A

1. the "Combined_Synopsis_and_Solicitation_dataset_refinement.pdf" document starts from

Page 4. Were page 1-3 redacted or could they be provided?

Page 1-3 is the description on SAM.gov.

2. Information provided in Combined_Synopsis_and_Solicitation_dataset_refinement.pdf is very sparse. To assist us respond could you provide information on 1) estimate of large the full dataset to be processed might be, 2) government preference on what type of software may be used database refinement (open-source vs COTS), 3) estimates of level-of-effort in FTE is available?

The full dataset will consist of 10-15 raster datasets for the continental United States at a

30-meter resolution. Each raster map is about 200 Mb. All process steps must be reproducible by the USGS and all software used must be open source (e.g., python scripts used in Google Earth Engine or other similar cloud based computing system). Level of effort depends on expertise, methods, and data used to construct required products. This information cannot be provided.

3. It says that the Contractor may invoice quarterly after completion of the deliverable(s). I understand that to mean that a maximum of 4 invoices can be submitted (to match the 4 deliverables). But if multiple deliverables are completed ahead of the deadline, does that mean multiple invoices cannot be submitted within the same quarter and the contractor must wait before submitting?

The funds for this work are provided on an annual basis. Thus, the work must be completed and invoiced according to the planned timeline for the project. Multiple invoices cannot be submitted during the same quarter.

4. What is the timeline for receiving payment after an invoice is approved? How many days?

According to Prompt Pay Act, the contractor should receive payment within 30 business days.

5. Does the government supply the datasets and appropriate licenses? If so, what datasets are you providing and can you provide links or supporting information such as file types, aggregate file size, and examples of data?

No, data required for this work must be determined and compiled by the contractor, and the data must be publicly available.

6. For ancillary data sets, are those identified by the government or are they at the discretion of the vendor? If identified by the government, when will they be identified and what datasets are you providing and can you provide links or supporting information such as file types, aggregate file size, and examples of data?

Ancillary data sets will be determined by the vendor and agreed upon by the government.

140G0323Q0164amd0001 Page 3 of 3

7. Are there any restrictions on using Cloud providers (AWS, Microsoft Azure, or Google Cloud

Platform) for this analysis?

There are no requirements; however, the cloud computing environment must be agreed upon by the government to establish that the government can reproduce datasets developed by the contractor.

8. Are costs for Cloud computing covered by the government or the supplier?

Cloud computing costs to develop datasets required by the contract must be paid by the contractor.

9. Is hosting the dataset at the end of the engagement a requirement?

No, the government will host the datasets.

10. Is creation of user interfaces or visualization tools a requirement?

No, work must be done using automated programming approaches and these scripts must be provided to the government to establish reproducibility and for creating future datasets.

11. Has this work (or similar) been performed for earlier versions of the NLCD? If so, will you provide the previous Solicitation IDs and Contract IDs?

No

12. Is any travel required for meetings regarding this project? If so, will that be reimbursed or covered by the government?

No

13. Machine Learning has an inherent level of inaccuracy. What are the expected quality standards for the project or accepted level of accuracy and how are they assessed?

Datasets produced must be validated using existing data sets established by the government. Accuracy standards must be consistent with other previously published land use mapping datasets and be publishable in a peer reviewed journal. Standard machine learning validation methods and statistics must be used to assess accuracy, including k-fold cross validation or other similar validation methods and correlation statistics.

14. For task 2, is there an expectation of delivery for any sample of the output of task 3?

Yes, completion of task 2 requires that datasets, methods, and results must be validated for pilot regions established and agreed upon by the government and contractor.