Proposal Draft: Regional Urban Heat Monitoring Service

Reference: CALL-2026-EO-047 | Submitted by: [Your Organisation]

Section 1 — Understanding of Need (Technical)

Urban heat islands (UHIs) represent one of the most acute climate adaptation challenges
facing European cities. Studies confirm that UHI intensity exceeds 8°C in dense urban
cores during heat events, directly correlating with excess mortality (Eurostat, 2025).
The public EO programme has established the data infrastructure to monitor UHI at
continental scale; what is lacking is an operational, policy-actionable service layer
that translates satellite measurements into decision-ready indicators.

This proposal directly addresses [REQ-SOW-1]: "Development of an operational
urban heat monitoring service using public multispectral and thermal satellite constellations data at ≤10 m resolution."

Summary Box — Cardinal Requirements Addressed
| CR ID | Requirement | Section | Status |
|---|---|---|---|
| REQ-SOW-1 | Urban heat monitoring, ≤10 m resolution | 1.2 | ✅ Addressed |
| REQ-SOW-3 | Near-real-time alert capability (≤2h latency) | 2.3 | ✅ Addressed |
| REQ-SOW-5 | Open API for city integration | 3.1 | ✅ Addressed |

Section 2 — Technical Methodology

Baseline: Existing UHI products (e.g., LST-Reference-EU) operate at 1 km resolution with
24-hour latency — insufficient for emergency response decisions. Current state of the
art produces weekly composites unsuitable for heat event monitoring.

Our innovation: Multi-source thermal fusion combining:
- thermal EO (SLSTR-class) (LST, 1 km, 30 min revisit) for temporal resolution
- multispectral EO (MSI-class) (10 m, 5-day revisit) for spatial detail
- AI downscaling model (CNN-based, trained on 12 European cities 2017–2025)
- Physics-informed constraints to prevent spectral confusion artifacts

Result: 10 m LST maps with ≤90 min latency — a 16× spatial and 19× temporal
improvement over baseline.

Section 3 — Work Plan

⚠️ Human approval required before submission. Review compliance matrix and validate budget.