Advanced Predictive Hydrology And The Identification Of Future Wetlands

Advanced Predictive Hydrology and the Identification of Future Wetlands: A Geospatial and Ecological Analysis of Paludification and Topographic Modeling for Infrastructure and Resource Management

Theoretical Framework of Predictive Hydrology in Wetland Science
The discipline of wetland science is currently navigating a profound transition from descriptive, inventory-based mapping to predictive, process-based hydrological modeling. Historically, the management of aquatic resources across the United States has relied upon the National Wetlands Inventory (NWI), a dataset largely derived from the visual interpretation of aerial photography and satellite imagery. While the NWI has served as an essential baseline for decades, its methodology is inherently retrospective, capturing only those features with visible surface water or distinct hydrophytic vegetation signatures at the time of image acquisition. This approach often fails to identify “cryptic” wetlands—ecosystems such as forested seeps, vernal pools, and areas of subsurface saturation that lack a permanent surface-water expression but perform critical ecological and hydrological functions. For large-scale land managers, specifically the Minnesota Department of Natural Resources (DNR) and utility providers such as Xcel Energy, the limitations of static inventories present significant operational and regulatory risks.
The concept of “future wetlands” emerges as a critical technical frontier in this context. Rather than merely documenting existing boundaries, predictive hydrology seeks to identify areas with a high “intrinsic potential” to support wetland conditions based on their topographic, pedologic, and climatic characteristics. This predictive capability is particularly valuable for long-term infrastructure planning. If a transmission line or substation is sited in an area with high saturation probability, the subsequent development of wetland conditions through processes like paludification can lead to structural instability, increased maintenance costs, and unforeseen regulatory entanglements under the Clean Water Act or the Minnesota Wetland Conservation Act (WCA).
The technical foundation of this predictive work rests on high-resolution topographic data and sophisticated geospatial indices. By utilizing Light Detection and Ranging (LiDAR) data, analysts can resolve micro-topographic features that dictate the movement and accumulation of water across the landscape. The primary tool in this analysis is the Compound Topographic Index (CTI), also known as the Topographic Wetness Index (TWI). The CTI provides a continuous probability map of soil saturation by integrating the upslope contributing area with the local slope gradient. When combined with an understanding of paludification—the process by which peat accumulation raises the local water table—the CTI allows for the identification of wetland areas before they are officially delineated or even fully formed.
The Ecological Mechanism of Paludification and Peatland Expansion
Paludification is a dynamic and often irreversible ecological process that facilitates the conversion of previously dry mineral soils into waterlogged peatlands. Unlike terrestrialization, which occurs when organic matter fills an existing open-water body, paludification involves the “swamping” of upland areas. This process is driven by a complex feedback loop between vegetation, hydrology, and soil chemistry. In northern latitudes, particularly in the boreal forests of Minnesota and Canada, paludification is a primary driver of landscape-scale peatland initiation.
The initiation of paludification typically begins with a change in the local moisture regime, often triggered by allogenic factors such as increased precipitation, climatic cooling, or human-induced changes to drainage. Once the soil becomes sufficiently moist, it favors the colonization of bryophytes, particularly species of Sphagnum moss. Sphagnum is an ecosystem engineer; it has a high water-holding capacity and releases organic acids that lower the soil pH and inhibit the activity of decomposing microbes. As a result, organic matter accumulates faster than it can decompose, leading to the formation of a peat layer.
Attribute
Impact of Paludification on Mineral Soil
Long-term Ecological Result
Soil Temperature
Significant decrease due to insulation from thick organic layers
Reduced microbial activity and slower nutrient cycling
Decomposition Rate
Dramatic reduction under anaerobic, acidic conditions
Net accumulation of carbon and peat
Hydraulic Conductivity
Decreased as peat fills soil pores and inhibits drainage
Rising local water table and expanded saturation
Nutrient Availability
Sequestration of nitrogen, phosphorus, and sulfur in organic matter
Shift from minerotrophic to oligotrophic vegetation

As the peat layer thickens, it acts as a physical barrier to vertical drainage, forcing the water table to rise toward the surface. This rising water table, in turn, facilitates the lateral expansion of the peatland into adjacent upland forests, a phenomenon often observed as the “forest-to-bog” transition. This expansion is particularly problematic for forest management and infrastructure. In harvested stands with low soil disturbance, the post-harvest rise in the water table—known as “watering up”—can create favorable conditions for Sphagnum colonization and permanent paludification, potentially rendering the site unsuitable for future timber production or heavy infrastructure.
Technical Implementation of the Compound Topographic Index (CTI)
The technical ability to predict these “future wetlands” relies on the mathematical modeling of terrain to locate areas where water will naturally converge and pond. The Compound Topographic Index (CTI) is the standard geospatial metric used for this purpose. It is a steady-state wetness index that serves as a proxy for the spatial distribution of soil moisture and the likelihood of saturation.
Mathematical Formulation and Geospatial Logic
The CTI is calculated using the following primary equation:
In this equation, As represents the specific contributing area (the drainage area per unit contour length), and \beta is the slope angle in radians. The logic of the index is straightforward: areas with a large upslope drainage area (As) and a very low slope (\beta) will result in high CTI values, indicating a high probability of saturation. Conversely, ridges and steep slopes will have low As and high \beta, resulting in low CTI values.
The implementation of CTI in a GIS environment, such as ArcGIS, requires several pre-processing steps to ensure the accuracy of the final probability map. The quality of the input Digital Elevation Model (DEM) is paramount. High-resolution LiDAR-derived DEMs (typically 1-meter or 3-meter resolution) are preferred to capture the fine-scale topographic variations that dictate local hydrology.
GIS Workflow Step
Tool/Function
Description of Process
Fill Sinks
Spatial Analyst > Hydrology > Fill
Removes small depressions and artifacts in the DEM to ensure continuous flow
Flow Direction
Spatial Analyst > Hydrology > Flow Direction
Determines the direction of flow from each cell to its steepest neighbor (D8 or D-infinity)
Flow Accumulation
Spatial Analyst > Hydrology > Flow Accumulation
Calculates the number of upslope cells draining into each pixel, multiplied by cell area to get As
Slope Calculation
Spatial Analyst > Surface > Slope
Generates a slope raster from the DEM, converted to radians for the tangent function
Raster Algebra
Spatial Analyst > Map Algebra > Raster Calculator
Executes the CTI formula: Ln((“FlowAcc” * CellSize) / Tan(“Slope”))

The Role of Flow Accumulation Algorithms
A critical nuance in CTI calculation is the choice of flow accumulation algorithm. The traditional D8 (Deterministic Eight-Node) algorithm directs all flow from a cell into a single neighboring cell in the direction of the steepest descent. While computationally efficient, D8 often produces artificial, straight-line flow paths and fails to represent divergent flow on convex slopes. Advanced modeling for wetland intrinsic potential often utilizes the D-infinity (D_{\infty}) algorithm, which partitions flow between two neighbors, or triangular multiple flow algorithms, which provide a more realistic representation of how water disperses across the landscape. For entities like the Minnesota DNR, which require high-precision mapping for regulatory enforcement, the choice of these algorithms can significantly influence the resulting “future wetland” probability maps.
Geospatial Infrastructure: Minnesota’s LiDAR and MnTOPO Systems
The effectiveness of any predictive hydrological model is fundamentally limited by the resolution and accuracy of the underlying elevation data. Minnesota has developed a world-class geospatial infrastructure to support this work, primarily through the Minnesota Elevation Mapping Project and the MnTOPO web application.
Evolution of Minnesota’s Elevation Data
Minnesota has transitioned through multiple generations of elevation data, moving from coarse 30-meter DEMs derived from 1960s-era topographic maps to high-resolution LiDAR. The state’s first-generation LiDAR collection (QL3), completed between 2008 and 2012, provided the foundation for the current NWI updates and hydrological modeling projects. However, as landscape changes occur due to natural events and human development, the need for more accurate data led to the Second Generation Statewide LiDAR Collection.
This new generation of data meets or exceeds USGS Quality Level 1 (QL1) standards, offering significantly higher point density and improved vertical accuracy. The resulting seamless 0.5-meter LiDAR DEM represents a transformative leap in predictive hydrology. For a utility company like Xcel Energy, this resolution allows for the identification of micro-depressions and subtle flow-path obstructions (such as field disturbances or small culverts) that could lead to localized paludification and future wetland formation around infrastructure footings.
Data Access and Application via MnTOPO
The MnTOPO system serves as the primary portal for viewing and downloading this high-resolution data. It supports three primary tiers of users: Terrain Explorers (landowners and hikers), Terrain Analysts (engineers and hydrologists), and LiDAR Point Cloud Analysts. For technical wetland modeling, the 1-meter and 3-meter raster products are the most frequently utilized.
Data Product
Format/Specification
Primary Use in Predictive Hydrology
1-Meter DEM
Raster (ESRI Geodatabase or Shapefile)
Base layer for CTI and WIP modeling; identifies fine-scale drainage patterns
Hydro Breaklines
Vector (Line features)
Used to “enforce” hydrological flow in DEMs by defining stream edges and shorelines
Raw LAS Files
Binary Point Cloud
Allows for analysis of vegetation height and canopy structure to identify “cryptic” wetlands
2-Foot Contours
Vector (Line features)
Visual aid for understanding complex drainage basins and floodplains

The MnTOPO infrastructure is funded by the Clean Water Fund of the Clean Water, Land and Legacy Amendment, highlighting the state’s commitment to using detailed topographic information to protect and preserve water quality. This commitment is operationalized through collaborative efforts between MNIT @ DNR and the Minnesota Geospatial Information Office (MnGeo).
The Wetland Intrinsic Potential (WIP) Tool: A Machine Learning Approach
Building upon basic topographic indices like the CTI, researchers have developed the Wetland Intrinsic Potential (WIP) tool. This tool represents a significant advancement in the ability to identify “future wetlands” by integrating multi-scale topographic indicators into a random forest machine learning model.
Methodology and Model Architecture
The WIP tool is implemented as an ArcGIS toolbox using a combination of R and Python scripts. Its primary objective is to mimic ground-based wetland detection by using spatially explicit input variables that represent the three core wetland indicators: hydrophytic vegetation, hydrology, and hydric soils.
The model utilizes 19 different input variables derived from LiDAR data. While standard CTI models focus on local slope and drainage, the WIP tool incorporates “multi-scale” indicators. These indicators measure topographic position and moisture potential at various scales (e.g., across 10-meter, 100-meter, and 1,000-meter windows), allowing the model to distinguish between small localized depressions and large regional floodplains.
To ensure the model is robust, an iterative training process is employed:
Preliminary Modeling: An initial random forest model is run to create a preliminary probability raster.
Stratified Sampling: 600 sample points are randomly distributed across four probability strata (0.0–0.25, 0.25–0.5, 0.5–0.75, 0.75–1.0). This ensures the model learns from high-probability wetlands, clear uplands, and the “uncertain” transitional zones.
Analyst Labeling: Each point is evaluated by experts using LiDAR hillshades, NWI maps, and aerial imagery, with a portion of the points being field-verified to ensure accuracy.
Comparative Performance and “Cryptic” Wetland Detection
The WIP tool’s ability to identify wetlands surpasses the capabilities of traditional NWI mapping. In a test case in the Hoh River watershed, the WIP tool identified over twice the wetland area compared to the NWI. More importantly, it dramatically reduced the errors of omission—wetlands that are actually present on the ground but missing from the map—from 47.5% in the NWI to just 14.1%.
Accuracy Metric
National Wetlands Inventory (NWI)
WIP Tool (0.5 Probability Threshold)
Overall Accuracy
–
91.97%
Omission Error (Wetlands Missed)
47.5.1%
Commission Error (Uplands Mapped as Wetland)
1.9.5%

The slight increase in commission errors (uplands labeled as wetlands) is often a result of the tool’s sensitivity. Many of these “errors” are in fact areas of marginal saturation or “future wetlands” that are in the early stages of paludification or are only saturated during extreme precipitation events. For Xcel Energy, these commission errors represent “zones of caution” where the landscape has the intrinsic potential to become a wetland, even if it does not yet meet the strict legal definition during a field delineation.
Case Study: Infrastructure Risk Management for Xcel Energy
For utility providers like Xcel Energy, the integration of predictive hydrology into infrastructure lifecycle management is not merely an environmental goal but a critical component of risk mitigation. The development of high-voltage transmission lines and the maintenance of the electrical grid require a nuanced understanding of where the landscape is becoming wetter over time.
Transmission Line Routing and Regulatory Compliance
When Xcel Energy plans a new 345-kilovolt (kV) transmission line, such as the project from the Wilmarth Substation in Mankato to the North Rochester Substation, the routing process must balance engineering requirements with environmental impacts. The Minnesota DNR pays close attention to Right-of-Way (ROW) expansion, particularly at river crossings and near Minnesota Biological Survey (MBS) sites.
Predictive hydrological maps allow Xcel to anticipate wetland impacts before they are formally identified in the field. This is critical because the destruction of wetlands for road or bridge projects is only permitted if the project is the “least environmentally damaging” option and if compensatory mitigation is provided. By using CTI and WIP modeling, Xcel can:
Optimize Route Selection: Choose “greenfield” routes that cross primarily agricultural lands rather than sensitive wetland mosaics. 2. Predict Mitigation Needs: Forecast the amount of wetland “credits” required for a project years in advance, reducing the risk of project delays.
Assess Long-term Stability: Identify areas prone to paludification where the accumulation of peat and the rise of the water table could compromise the foundations of transmission towers.
Cumulative Wetland Effects Analysis (CWEA)
In the context of large-scale projects like the Keetac Mine Expansion, a Cumulative Wetland Effects Analysis (CWEA) is required to meet the requirements of the National Environmental Policy Act (NEPA). This analysis evaluates the incremental impact of a project when added to other past, present, and “reasonably foreseeable future actions”. Predictive hydrology provides the quantitative basis for these “reasonably foreseeable” impacts. For example, if a project blocks a drainage pathway, the CTI can model how the upstream area will respond, identifying where “future wetlands” will form as a result of the project itself.
Watershed-Scale Application: Coon Creek Watershed and Functional Capacity
The practical application of these predictive techniques at a watershed scale is demonstrated by the work of the Coon Creek Watershed District in Minnesota. Here, researchers such as Justin Hawley and Tim Kelly utilized raster math to assess the functional capacity of wetlands based on their Hydrogeomorphic (HGM) classification.
Raster Math and the Functional Capacity Index (FCI)
The project’s goal was to assess wetland functional capacity without the prohibitive time and cost of establishing intensive field reference sites. By using LiDAR-derived data and raster math, the team developed models for various wetland functions, including groundwater discharge modification and nutrient retention.
Functional Category
Indicator Variables
Model Output (FCI)
Hydrologic Function
Drainage area, slope, soil permeability
Capacity to store and slow stormwater
Water Quality Function
Land use, CTI value, vegetation cover
Capacity for sediment and toxicant retention
Biological Function
Habitat patch size, connectivity, diversity
Capacity to support threatened or endangered species

The Functional Capacity Index (FCI) generated by these models indicates the degree to which a wetland (or a potential wetland area) performs a specific function. This information is highly portable and modular; as more accurate data (such as the Second Generation LiDAR) is integrated, the models can be easily updated to provide greater localized accuracy.
Navigating the Anoka Sand Plain
The Coon Creek Watershed is located within the Anoka Sand Plain, a landscape characterized by nearly level lake plains and fine sandy soils. In this region, the regional water table is exceptionally shallow, often less than 15 feet below the surface. This high water table, combined with the sandy soils, makes the area highly susceptible to rapid hydrological shifts. Predictive CTI modeling in this context is essential for identifying areas where minor changes in land use—such as the construction of a new road or the installation of a drainage ditch—can cause the water table to intersect the surface, creating new wetland conditions.
Advanced DEM Processing: The D2P Algorithm and Hydro-Conditioning
A significant technical hurdle in predictive hydrology is the management of “sinks” or depressions in the DEM. Traditional hydrological modeling often involves “filling” all sinks to ensure that water flows continuously across the digital landscape to the watershed outlet. However, this process often removes real, physically meaningful depressions—such as wetlands, ponds, and karst features—from the model.
Depression-Preserved DEM Processing (D2P)
To address this, researchers have developed the Depression-Preserved DEM Processing (D2P) algorithm. This algorithm is designed to distinguish between artificial artifacts (caused by LiDAR sensor errors or processing) and real surface depressions. In a case study in the Goodwin Creek Experimental Watershed, the D2P algorithm successfully resolved 86% of actual ponds at a 10-meter resolution, while reducing the number of modified cells by 51% compared to conventional algorithms.
For the Minnesota DNR, the preservation of these depressions is critical for the accurate mapping of “future wetlands.” If a depression is filled in the model, the CTI will not correctly identify it as a point of convergence, and the “intrinsic potential” of that area will be underestimated. The D2P workflow involves:
Sink Extraction: Identifying all topographic depressions by subtracting the original DEM from a filled, depression-less DEM.
Artifact Removal: Applying filters to remove systematic and random errors associated with data acquisition.
Hydro-Conditioning: Manually or automatically breaching artificial barriers (such as road embankments) while preserving natural depressions.
Levels of Hydro-Conditioning
The complexity and cost of hydro-conditioning vary depending on the required accuracy of the final model. The Minnesota 3D Geomatics Committee has identified three primary levels of hydro-modified DEMs (hDEMs) :
Hydro-Conditioning Level
Methodology
Primary Use Case
Level 1: Automated
Uses existing NHD (National Hydrography Dataset) lines to “burn” flow paths
Large-scale, low-cost planning and simple terrain analysis
Level 2: Semi-Automated
Incorporates lake routing and basic culvert locations
Sub-watershed planning and BMP (Best Management Practice) analysis
Level 3: Precision
Intensive manual interpretation and field-verified culvert data
Field-scale implementation and credible regulatory enforcement

For Xcel Energy, Level 3 hydro-conditioning is often necessary when planning transmission routes through complex terrain, as it ensures that the modeled hydrological boundaries accurately reflect the field-scale reality.
Validation and Accuracy Assessment in the Minnesota Context
The ultimate utility of predictive hydrology for the DNR and Xcel Energy depends on the validated accuracy of the models. Minnesota has established rigorous protocols for assessing the quality of its wetland mapping and monitoring programs.
The Minnesota Wetland Status and Trends Monitoring Program (WSTMP)
The WSTMP provides a baseline for wetland quantity and quality in the state. Its results are based on a combination of photo-interpretation and intensive field verification. Based on feature counts, the WSTMP has achieved an overall accuracy of 94% for distinguishing between wetland and non-wetland areas.
WSTMP Metric
Accuracy Rate
Context/Meaning
Overall Accuracy
94%
General reliability of the status and trends data
Kappa Coefficient (\kappa)
89%
Accuracy beyond what would be expected by chance alone
Omission Error (Wetlands)
3%
Percentage of actual wetlands missed by interpreters
Commission Error (Wetlands)
7%
Percentage of uplands incorrectly mapped as wetlands

These high accuracy rates are a testament to the quality of the LiDAR data and the expertise of the DNR Resource Assessment Office. However, these figures represent the accuracy of existing wetland identification. When moving toward predictive modeling of future wetlands, the accuracy levels often decrease due to the inherent uncertainty of environmental change.
Deep Learning and WOTUS-ML Validation
The next generation of validation involves deep learning models. The WOTUS-ML model, which predicts Clean Water Act jurisdiction, was trained on 150,000 jurisdictional determinations from the Army Corps of Engineers. It achieved a test set accuracy of 79%, representing a 14-percentage-point improvement over naive baseline models. In the St. Paul District, which covers Minnesota, the model was found to be “extremely accurate,” particularly in areas with low regulation rates. This suggests that machine learning can effectively replicate the complex decision-making processes of federal regulators, providing Xcel Energy with a powerful tool for anticipating the legal status of “future wetlands.”
Socioeconomic and Climate Implications of Future Wetland Prediction
The technical work of CTI calculation and paludification modeling carries significant socioeconomic weight. As the climate becomes more “oceanic”—characterized by warmer winters and cool, moist summers—the risk of widespread paludification and forest retreat in northern regions increases.
Carbon Sequestration and Climate Change Mitigation
Peatlands are the world’s most organic carbon-dense ecosystems. In Minnesota, the preservation of these systems is a critical component of climate action. Draining peatlands for development or agriculture triggers the oxidation of organic matter, releasing massive amounts of CO_{2}. Conversely, the identification and restoration of “potential wetland areas” through rewetting and paludiculture can reverse these emissions. Predictive hydrology models identify the most efficient locations for these interventions, ensuring that restoration dollars are spent in areas with the highest intrinsic potential for long-term carbon storage.
Disaster Resilience and Hazard Mitigation
Wetlands provide essential ecosystem services, including flood risk reduction and water regulation. In Goodhue and Ramsey Counties, GIS-based hazard mitigation plans integrate wetland data to identify areas where “natural infrastructure” can protect property and life. By predicting where future wetlands will form, these counties can:
Enhance Stormwater Management: Design systems that leverage natural depressions for overflow storage.
Protect Infrastructure: Identify sections of the power grid or transportation network that are at risk of flooding due to rising water tables.
Ensure Safe Development: Discourage building in areas with high CTI values, even if those areas are currently dry, to prevent future property damage and loss of life.
Technical Synthesis and Strategic Conclusion
The identification of “future wetlands” through predictive hydrology represents a paradigm shift in environmental management. For Xcel Energy and the Minnesota DNR, this work is not an academic exercise but a technical necessity. By integrating the biological reality of paludification with the mathematical precision of the Compound Topographic Index (CTI), land managers can move beyond the limitations of static, retrospective inventories.
The use of high-resolution LiDAR, supported by the MnTOPO infrastructure and advanced algorithms like D2P, allows for the creation of continuous probability maps that resolve the landscape’s intrinsic potential for saturation. These models identify the “cryptic” wetlands of today and the established peatlands of tomorrow. For a utility company, this means more resilient infrastructure and fewer regulatory surprises. For the state, it means the more effective protection of its water resources and a powerful tool for climate mitigation.
As machine learning and deep learning models like the WIP tool and WOTUS-ML continue to evolve, the distinction between “delineated” and “predicted” wetlands will likely blur. The future of hydrological management lies in the ability to anticipate change, using the fundamental laws of topography and biology to navigate an increasingly dynamic and saturated landscape. This technical expertise is the cornerstone of responsible stewardship and infrastructure resilience in the 21st century.
Works cited
1. Mapping wetland intrinsic potential through machine learning of multi-scale remote sensing – EGUsphere, https://egusphere.copernicus.org/preprints/2022/egusphere-2022-665/egusphere-2022-665-ATC1.pdf 2. Status and Trends of Wetlands in the Conterminous United States 2009 to 2019: Report to Congress, https://www.fws.gov/sites/default/files/documents/2024-03/wetlands-status-and-trends-2009-to-2019.pdf 3. The Wetland Intrinsic Potential tool: mapping wetland … – HESS, https://hess.copernicus.org/articles/27/3687/2023/ 4. Sedarski, Joe – Mankato-Mississippi River Transmission Project, https://www.mmrtproject.com/documents/filing/MMRT_CombinedApplication_AppendixM-3_Apr2024.pdf 5. Geospatial Wetlands Impacts & Mitigation Forecasting Models – ROSA P, https://rosap.ntl.bts.gov/view/dot/32668/dot_32668_DS1.pdf 6. Cumulative Wetland Effect Analysis East Reserve Mining Project, https://www.lrl.mn.gov/docs/2015/other/150681/PFEISref_1/Barr 2006a.pdf 7. Which is the correct way to calculate the topographic wetness index with ArcGis Dextop?, https://www.researchgate.net/post/Which-is-the-correct-way-to-calculate-the-topographic-wetness-index-with-ArcGis-Dextop 8. (PDF) Paludification and management of forested peatlands in Canada: A literature review, https://www.researchgate.net/publication/52010181_Paludification_and_management_of_forested_peatlands_in_Canada_A_literature_review 9. Effects of Water Table Fluctuations on Peatland-A Review – SAS Publishers, https://www.saspublishers.com/media/articles/SJET23C482-487.pdf 10. Peatland establishment on mineral soils: Effects of water level …, https://aeic-iaac.gc.ca/050/documents_staticpost/59540/82534/Ecological_Engineering_Article.pdf 11. Paludification and Forest Retreat in Northern Oceanic Environments – PMC, https://pmc.ncbi.nlm.nih.gov/articles/PMC4244994/ 12. Agriculture on wet peatlands: the sustainability potential of paludiculture – UU Research Portal, https://research-portal.uu.nl/ws/files/276846336/1-s2.0-S0308521X25003014-main.pdf 13. Statewide Seamless 0.5m Lidar DEM, Minnesota, 2025, https://gisdata.mn.gov/lt/dataset/elev-lidar-statewide-gen2 14. Topographic hydro-conditioning to resolve surface depression storage and ponding in a fully distributed hydrologic model – PMC, https://pmc.ncbi.nlm.nih.gov/articles/PMC10434835/ 15. MnTOPO | Minnesota DNR, https://www.dnr.state.mn.us/maps/mntopo/index.html 16. LiDAR – Minnesota’s Digital Elevation Project | Minnesota DNR, https://www.dnr.state.mn.us/maps/lidar/index.html 17. Minutes: 3D Geomatics Committee Hydrography Workgroup – MnGeo, https://www.mngeo.state.mn.us/committee/3dgeo/hydro/3D_Hydrography_Minutes_20181211.pdf 18. MnTOPO, https://mntopo.dnr.state.mn.us/ 19. U.S. Steel Keetac Taconite Mine Expansion Project Draft … – files, https://files.dnr.state.mn.us/input/environmentalreview/keetac/draft_eis/20091124_versiondrev_with_bookmarks.pdf 20. 21st Annual Conference Sessions – Minnesota GIS/LIS Consortium, https://www.mngislis.org/page/695 21. Assessment of the Functional Capacity of Wetlands within the Coon Creek Watershed, https://www.cooncreekwd.org/wp-content/uploads/2023/11/Appendix_G_-_Functional_Capacity_of_Wetlands.pdf 22. The Development of a GIS Methodology to Identify Oxbows and Former Stream Meanders from LiDAR-Derived Digital Elevation Models – MDPI, https://www.mdpi.com/2072-4292/11/1/12 23. Status and Trends of Wetlands in Minnesota: Wetland Quantity Baseline – files, https://files.dnr.state.mn.us/eco/wetlands/wstmp_report_final_121410.pdf 24. Machine Learning Predicts Which Rivers, Streams, and Wetlands the Clean Water Act Regulates – Resources for the Future, https://www.rff.org/documents/4339/Machine_Learning_Predicts_Which_Rivers_Streams_and_Wetlands_the_Clean_Water_Ac_h6GbAc0.pdf 25. Advancements in mapping areas suitable for wetland habitats across the conterminous United States – PMC, https://pmc.ncbi.nlm.nih.gov/articles/PMC11998124/ 26. Full article: Ecosystem services delivered by small-scale wetlands – Taylor & Francis, https://www.tandfonline.com/doi/full/10.1080/02626667.2011.630317 27. GOODHUE COUNTY ALL HAZARD MITIGATION PLAN 2016, https://goodhuecountymn.gov/files/ordinances/all_hazard_mitigation_plan_2016.pdf 28. RAMSEY COUNTY, MN HAZARD MITIGATION PLAN 2025, https://assets.ramseycountymn.gov/files/2026-01/2025 Ramsey County MJHMP_Final_Combined (for publishing).pdf 29. Vikings/Ramsey County Stadium Proposal Risk Analysis – October 2011, Metropolitan Council and Metropolitan Sports Facilities Com – Minnesota Legislature, https://www.leg.mn.gov/docs/2011/other/110818.pdf