From the Utility Management edition of November 5, 2026
Opti-SITE: Danish and Swedish utilities train ML and radar nowcasting for stormwater and nature-based siting
DHI (with HOFOR, NSVA and Interreg ÖKS partners) · Municipal stormwater and combined sewer management · Denmark and Sweden (Copenhagen and Landskrona case areas)
- Publisher
- DHI
- Original publication
- DHI project page (accessed 5 Nov 2026; undated body)
- Source retrieved
- 2026-11-06
What happened
The Interreg Öresund-Kattegat-Skagerrak Opti-SITE project, described on DHI's site, is building decision-support tools that combine machine learning (including graph neural networks), multi-objective optimisation and real-time radar rainfall to site nature-based solutions (rain gardens, basins, ponds) and to nowcast combined sewer overflows. A prototype ML model with benchmarking and training-data workflows is reported done; final optimisation tooling and radar-hydrodynamic integration remain next steps. Testing is with utilities HOFOR and NSVA in Copenhagen and Landskrona. Results are expected to feed DHI's CityFlow Live product.
Why it matters
US municipal stormwater and CSO programs face the same siting and nowcast problems; transfer requires caution on EU funding context, climate, and the vendor-hosted roadmap, but the ML-plus-radar pattern is relevant where radar and sewer models exist.
Evidence and measured results
Vendor project page without a clear publication date. Prototype status is self-reported; no published overflow-reduction percentages on this page. Treated as vendor-claim.
Limitations and uncertainty
Vendor page; incomplete tooling; European pilots; no US measured results; publicationDate unknown.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-11-05; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Cities struggle to site nature-based stormwater controls and anticipate CSOs under heavier rain.
- Stakeholders
- Stormwater utility planner
- Wastewater operations
- Modelling consultant
- City climate adaptation lead
- Discovery questions
- Do you have radar or gauge rainfall feeds?
- Is there a calibrated sewer model?
- Which CSO outfalls are priority?
- Value hypothesis
- ML siting plus radar nowcasting may improve NBS placement and overflow warning; local proof required.
- Potential engagement
- Stormwater analytics readiness and pilot catchment design.
- Unsupported claims
- Specific overflow reduction
- US performance parity
- Finished commercial readiness
Pre-sales engineering
Role takeaway
- Fit
- Municipalities with sewer models and rainfall data willing to pilot decision support.
- Architecture and integration
- Offline/planning ML optimisation plus radar nowcast models; careful boundary before any real-time control writes.
- Prerequisites
- Rainfall data
- Sewer network model
- Catchment GIS features
- Constraints
- EU project context
- Vendor product roadmap
- No US results here
- Security
- Protect network models and operational radar pipelines; no unverified automated gate control.
- Proof of value
- Nowcast skill vs baseline
- Optimised vs manual NBS portfolio metrics
- Operator-accepted alert lead time
Delivery
Role takeaway
- Work
- Assemble data, pilot one catchment, validate with operators, decide build-vs-buy.
- Dependencies
- Radar provider
- Modelling platform
- Capital planning cycle
- Ownership
- Stormwater planning lead with operations.
- Skills and adoption
- Train planners to interpret ML siting outputs.
- Governance checkpoints
- Data-sharing agreements
- Pilot success criteria
- Procurement decision gate
- Acceptance criteria
- Proposed: published pilot skill scores
- Proposed: operator sign-off on alert workflow
- Proposed: clear separation from autonomous control
- Risks
- Overfit EU pilots
- Vendor lock-in
- Acting on unvalidated nowcasts
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
Graph ML and optimisation over catchment features plus radar-driven fast sewer models for planning and event nowcasting, eventually productised in CityFlow Live.
Governance
Who approves, reviews and stays accountable for outcomes?
Multi-utility research collaboration under Interreg; US analogues would need clear ownership of models and procurement boundaries if a vendor productises results.
Security and privacy
What data, permissions and controls need testing?
Not discussed; radar and network models are operationally sensitive if exposed.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Aimed at utilities and consultants needing decision support rather than in-house ML teams.
Procurement
What should contracts, pricing and exit terms secure?
Expect research-to-product path into a commercial platform; US buyers should separate research claims from product SLAs.
Operating model
Which teams own the service once it runs?
Supports both long-term NBS capital planning and real-time overflow response workflows.
Publication history
- 2026-11-05Utility Management · Issue 087 resources
Stable resource ID: dhi-opti-site-stormwater-ai-nature-based-solutions-2026