Professor Jennifer Duan, a professor of hydraulics and sediment transport in the Department of Civil and Architectural Engineering and Mechanics at the University of Arizona, presented a study to Pima County staff and consultants that updates countywide curve‑number maps and quantifies post‑fire increases in runoff for a Catalina Mountains burn area.

The study, completed last year and funded by the Pima County Regional Flood Control District, produced three main products: (1) a 30‑meter by 30‑meter raster map of curve numbers (CN) for the county derived from LANDFIRE 2020 vegetation type and cover data; (2) a post‑fire CN map for the Bighorn Fire burn area generated using burn‑severity/vegetation‑reduction maps from an interagency burn team; and (3) a 2‑meter by 2‑meter impervious‑area overlay that assigns impervious cells a CN of 98 for urban planning and green‑infrastructure analysis.

Why it matters: CN maps are a standard input for rainfall–runoff methods and many hydraulic/hydrologic models used by engineers and floodplain managers. Duan said the higher spatial resolution and updated vegetation cover better reflect current land use and can change modeled runoff outcomes for small basins and urban design decisions, such as locating rain gardens or evaluating the runoff impact of new development.

Methods and data: Duan described two lookup sources used to assign CN values: the national TR‑55 tables and a continuous CN table used locally (referred to in the presentation as the Pima County/PC Hydro table). The study reclassified LANDFIRE's 75 vegetation classes into the CN lookup categories, combined vegetation‑cover reduction from the burn team with pre‑fire LANDFIRE cover, and built 30‑m CN rasters. For urban areas the team produced a 2‑m impervious layer and merged it with the 30‑m CN map; the full 2‑m dataset is large (about 35 GB), Duan said, and is already distributed to some firms and researchers.

Post‑fire results: For the Catalina Mountains/Bighorn Fire area, the team used burn‑severity rasters (high/moderate/low/unburned) and calculated CN changes at 30‑m cell scale. Duan reported that CN increases by cell ranged roughly from about 1 (very low severity) to a maximum observed change of 19 (high severity), with midrange values near 14.3 and a lower bound around 8.4 for many moderate cells. Using HEC‑HMS with the spatially varying CN map, the study estimated peak discharge increases of up to about 21% for one main channel (identified in the presentation as the CDO main channel), with Finger Rock showing the second‑largest increase and several other reaches showing increases near 15%.

Verification and limitations: The team performed verification runs using three CN approaches (gridded CN, basin average CN and the older PC Hydro/legacy shapefile CN) against observed stream‑gauge records for a large event (July 28–Aug. 5, 2006). Simulations used a 15‑minute timestep. Duan reported that all three CN approaches produced broadly similar results; none captured every observed peak, and the percent differences in peak discharge were generally small in many cases (she cited an example of about 0.6% for one metric). Duan cautioned that the event is from 2006 while LANDFIRE data are 2020, so land‑use change and urban expansion contribute to differences between modeled and observed hydrographs. She also noted the team did not perform full model calibration as part of this product.

Modeling considerations discussed: During Q&A Duan and attendees discussed common urban issues that can change modeled runoff: ponded/backyard water storage, antecedent soil moisture and drought effects, and the limits of routing approximations. Duan said ponded backyard storage and similar features can substantially reduce near‑term peak flows and are best represented with two‑dimensional or dynamic routing models (2‑D hydrodynamic or dynamic/ diffusion wave solvers); kinematic‑wave routing cannot represent fill‑and‑spill ponding. She also noted that base CN values should be adjusted for antecedent moisture (the standard CN adjustment for antecedent soil moisture conditions exists but was not applied in the verification runs reported).

Applications and availability: Duan emphasized the high‑resolution CN maps are intended for planners and consultants to evaluate how site‑level changes (for example, adding green infrastructure or new impervious surfaces) affect runoff. The dataset covers the county and can be clipped to local watersheds; Duan said the raster files are available from the county or directly from the research team. She identified Dr. DeMaria and Dr. Evan Canfield as the project managers from the Pima County Regional Flood Control District who advised the work.

Questions and next steps: County attendees asked whether the study includes temporal recovery of vegetation after fire; Duan said the project focused on immediate post‑fire effects and used the burn team's vegetation‑reduction map, and the team has not yet re‑run the workflow for multi‑year recovery. Attendees also asked about the geographic extent of the HMS verification (Duan confirmed the county line was used as a cut‑off because Pima County soil maps were not available for the downstream county). Duan said future research could address which CN methodology performs best under local conditions and incorporate more gauges or calibration if the county requests it.

Funding and credit: Duan said the project was funded by the Pima County Regional Flood Control District; Dr. DeMaria and Dr. Canfield served as project managers and provided advice during the study. The team published the methods in a paper referenced in the presentation and has provided the CN products to local engineers and researchers.

Data access: Duan said the 30‑m CN rasters and the 2‑m impervious overlay can be supplied by the county or the research team; the 2‑m product is large (about 35 GB) and was already shared with some local firms and researchers.