Citizen-driven monitoring of sustainable diets: linking individual behaviour to territorial indicators

• A behaviour-based Sustainable Diet Score (Care4Food) is developed and calibrated using a nationally representative survey in Portugal. • Citizen-generated dietary data are transformed into spatially explicit, municipal-level indicators through a spatial survey-dashboard pipeline. • A perception-behaviour gap in dietary sustainability is identified and can orient food related policies. • The approach enables low-cost, real-time monitoring of sustainable diets relevant for cross sectional and multilevel policy support. • Results demonstrate how geographic citizen science produces behavioural data for smart urban and place-based decision-support systems Cities, metropolitan areas and municipalities increasingly carry responsibility for promoting sustainable diets through public procurement, health promotion, and diet-related local food policies, yet they often lack low-burden indicators that capture everyday dietary behaviours and citizens’ perceptions in a way that supports routine policy monitoring. This study develops and tests Care4Food, a geographic citizen-science, behaviour-based Sustainable Diet Score paired with a self-perceived diet sustainability measure, and demonstrates how citizen-generated data can be translated into municipal-level indicators that can support broader urban and regional sustainability governance in Portugal. First, a framework-aligned scoring system was constructed and calibrated using a computer-assisted telephone interview survey of 1,813 adults in mainland Portugal. The Care4Food score integrates four sub-domains normalised to a 1–5 scale: protein balance, fruit and vegetable intake, ultra-processed food intake, and sustainability-oriented practices. The mean Care4Food score was 3.08 (SD = 0.82), with lower values primarily driven by protein imbalance and weaker sustainability practices. Self-perceived diet sustainability was higher (mean = 4.0), producing a perception–behaviour gap and a weak correlation between perceived and behaviour-based measures (r = 0.29). Miscalibration was strongest among low-scoring participants, consistent with a Dunning–Kruger-type pattern. Second, the calibrated score was implemented through a digital survey linked to an interactive dashboard to enable automated scoring, immediate personalised feedback, and aggregation of results at the municipal scale, while visualising these indicators alongside contextual food-environment layers. Forty voluntary online submissions generated descriptive exploratory municipal-level indicators and reproduced the observed tendency toward overestimation. The study demonstrates the feasibility of a low-cost, transparent, citizen-driven monitoring instrument that complements conventional urban sustainability and smart-city monitoring systems by linking individual dietary practices and perceptions to municipal decision-making.