Consultants tell Pensacola council manual digitizing, refined AI could correct stormwater fee inequities

Consultants from Raftelis told the Pensacola City Council at a workshop that the city's stormwater utility could be made more accurate and equitable by replacing assessor-based impervious-area data with measured or refined AI-derived mapping, but doing so would require upfront investment and an ordinance change.

Christina Conchilla, a manager with Raftelis'stormwater consulting division, told the council that Pensacola's current equivalent stormwater unit (ESU) is 2,998 square feet of impervious area and that the city's current rate is $120.49 per ESU, billed annually as a non-ad valorem assessment. "Your ESU is 2,998 square foot of impervious area," Conchilla said, "and your current rate is $120.49 per ESU." She showed examples where assessor tabular data omitted driveways and walkways; in one example a parcel listed as 4,135 sq ft by the property appraiser was measured by consultants at 8,392 sq ft, which would move it into a higher billing tier and nearly double the annual charge in that case.

The consultants described three broad paths: manual digitizing (human GIS technicians drawing impervious-area polygons from aerial imagery), proprietary AI-derived products from imagery providers, and in-house AI/deep-learning workflows (for example using Esri tools). "Manually develop data for all of your properties would be somewhere in the ballpark of $190,000," Katie Cromwell, a vice president with Raftelis, told the council, while noting ongoing maintenance costs could range from $10,000 to $20,000 per year depending on development pace and staffing.

Cromwell said AI products can be faster and sometimes less expensive up front but that quality varies and local tree canopy and residential character can reduce accuracy: "AI-generated impervious area is improving, but many models need refinement to be suitable for billing, especially in tree-covered residential areas." She estimated a city-owned, refined AI approach would likely cost $50,000'$90,000 depending on imagery, ownership rights and the level of manual validation required.

Consultants recommended a public-facing web viewer so residents can see parcel-level impervious-area measurements, ESUs and related charges; Cromwell estimated setup in the $8,000'$9,000 range plus IT/GIS support. The consultants also noted that adopting new impervious-area data would require a simple ordinance amendment because the current ordinance refers specifically to assessor data; they warned Trim notice and non ad valorem assessment deadlines in July'August mean a manual citywide digitizing campaign would likely take several months and probably would not be ready for the next billing cycle.

A member of the administration said staff had no technical preference for a method but raised capacity concerns: the administration's representative said initial digitizing would likely need to be outsourced because in-house GIS capacity may be insufficient and that the $10,000'$20,000 maintenance estimate presumes adequate staff. In public comment, resident James Gulley praised the work so far and urged any change to be "income neutral," suggesting the city could adjust ESU rates so new data did not increase the overall revenue burden on residents.

Council members asked practical questions about sample sizes, payback timelines and treatment of pervious pavers and pools. Consultants suggested conducting a targeted sample (they noted a random sample of 400 parcels is a standard approach for ESU determinations and that smaller pilot samples can provide quick insight) to estimate how much additional impervious area would be captured and whether that would offset the data development cost. The consultants said many jurisdictions that updated measured impervious-area data see meaningful additional ESUs that can allow a lower unit rate while preserving revenue.

Next steps the consultants recommended included: (1) deciding whether to own the data or purchase a vendor product; (2) conducting a sample or pilot to quantify the problem and payback; (3) if moving forward, soliciting vendors or outsourcing initial manual digitizing; and (4) updating the ordinance and implementing a public web viewer and customer appeal process. The consultants indicated they could support the city if the council chose to pursue one of the options.

The workshop closed with council thanks to staff and the Raftelis team. No formal policy decision or vote was recorded during the workshop; council members appeared aligned in wanting more accurate data but expressed differing preferences on ownership and the balance between upfront costs and long-term fairness.