The SpatioTemporal Asset Catalog (STAC) provides a standardized way to catalog and access geospatial data, making it easier for developers and researchers to find, share, and use various types of geospatial data. Increasingly, STAC is used by satellite providers to streamline access to their data, improve interoperability, and support cloud-native geospatial workflows. In a potentially controversial use case, we have been using STAC to catalog the underlying data that collectively represents environmental models, in particular hydrologic and hydraulic models. These models store geospatial, parameters, and time series data in a variety of file formats that are not cloud optimized, not conformant with open specifications, and may only be useable when opened in the model software itself. The provenance and data lineage for much of the foundational data stored in these files does come from traditional data formats (e.g. vector and raster data), but the metadata associated with source data is often lost. The model software is developed by the US Army Corps of Engineers, and while the source code is not open, it is free and is arguably the most widely used software in the world for inland flood hazards studies. While STAC is not a perfect fit, its community adoption and ecosystem of tools provide a uniquely adaptive and ready-set-go solution for cataloging models. This talk will address challenges of data management for USACE flood models—designed for the desktop—in the cloud, and how STAC provides a short term, interoperable solution for data cataloging, provisioning, and wrangling. Topics will include challenges of creating standardized—asset heavy—items, provenance linking, scaling probabilistic models, and operationalizing STAC into tools and workflows that support model development and execution. Tradeoffs, wins and losses, and some thoughts on what’s next moving forward will be included.