Catalog of stock ponds using machine learning

Overview

Development stage

PilotSource: b) Pilot – The use case has been deployed in a limited test or pilot capacity.

High-impact designation

Not high-impactSource: c) Not high-impact

Impact justification

Not reported

Start date

2026-04-01 (day precision)Source: 4/1/2026

Withheld from public reporting

NoSource: a) No

Mission

Topic area

Science

Operational functions

No function tag matched Derived by keyword rules; see methodology

Problem the AI is intended to solve

There are many undocumented stock ponds that retain water for use by farmers and ranchers throughout the landscape in the Dakotas. The stock pond AI model could identify unknown stock ponds to water managers for use in contaminant monitoring or water budget analysis and drought mitigation.

Expected benefits

Identifying stock ponds in North Dakota using an automated ML model rather than manually doing so increases efficiency at the state level that could free up labor for other tasks.

System outputs

The stock pond ML algorithm outputs predicted locations of stock ponds and dams in lidar images in jpg format and a csv file of locations given in latitude and longitude.

Technology

AI classification

Classical / predictive MLSource: Classical/Predictive Machine Learning: Models trained on data to make predictions or classifications based on identified patterns or relatio…

System name(s)

Not reported

Custom-developed code

No

Public source code

Not reported

Sourcing

How it was built

Developed in-houseSource: b) Developed in-house

Vendor (as reported)

Not reported

Vendors (standardized)

None on the standard list

A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.

Data and privacy

Involves PII

No

Privacy Impact Assessment

Not reported

Authorization to Operate

No

Demographic features

k) None of the above

Training and evaluation data

The model is trained on the publicly available lidar data disseminated by the North Dakota Department of Water Resources using the U-Net image segmentation architecture in PyTorch. The model is evaluated during training based on validation images through the use of RMSE. After the model development the output will be evaluated against known stock pond locations downloaded from the NDDWR and the U.S. Geological Survey National Hydrologic Data.

Federal Data Catalog

Not reported

Governance

0 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.

Potential impacts and how they were identified

Not reported