Wetlands are critical components of freshwater systems, yet they remain difficult to map and integrate into operational hydrologic and ecological analyses. In the United States, the National Wetlands Inventory (NWI) remains the most widely used wetland dataset, but it is limited by high errors of omission—particularly for small, vegetated, and intermittently inundated wetlands—and by spatial misalignment resulting from its legacy, photo-interpreted origins. These limitations hinder integration with other spatial datasets, including remote sensing imagery, hydrologic models, terrain analyses, and stream network data, constraining the ability to characterize wetland functions at management-relevant scales.
We describe the Wetland Intrinsic Potential (WIP) tool, a scalable, process-informed remote sensing and machine learning framework designed to serve as a foundational spatial layer for wetland characterization. Rather than producing a static wetland classification, WIP represents wetland presence as a continuous probability surface that reflects the intrinsic capacity of landscapes to support wetlands. The model is implemented using a random forest approach that integrates multi-scale terrain metrics derived from digital elevation models, hydrologic indices, soils, land cover, and climate variables. By explicitly incorporating geomorphic and hydrologic controls on water flux, WIP captures cryptic ,ephemeral, and degraded wetlands that are systematically missed by spectral classification approaches and legacy inventories.
WIP is designed to enable integration across datasets and disciplines. As a spatially continuous, topographically aligned surface, it provides a common spatial framework for linking wetlands with stream networks, hydrologic connectivity models, time-series remote sensing products, and biogeochemical datasets. Building on this foundation, we describe modular applications that characterize wetland functions, including estimates of water storage and hydrologic connectivity, and discuss how the same framework supports extension to additional function-based analyses.
All workflows are implemented in a repeatable, cloud-based architecture that enables high-resolution lidar-based analyses at the watershed scale and scalable processing at 10–30 m resolution across state and national extents. We illustrate applications in the Pacific Northwest, at national scale, and through partnerships with Ministries of Environment in several African countries via Digital Earth Africa. These examples demonstrate how WIP operationalizes wetland science as durable, reusable infrastructure for freshwater analysis rather than a one-off mapping product.