Al Safa 2 Park, reimagined as a living system that reads the climate, the crowd, and the land, then designs itself toward shade, water, and belonging.
Al Safa 2 is one of Dubai's most loved green spaces, yet like every desert park it fights three quiet enemies: surface heat that empties it by noon, irrigation that drinks against scarcity, and zones people never walk. SIDRA treats those as solvable measurements. Before a single line is placed, AI maps how the site heats, how people move, how wind and shade behave across the seasons, then lets that evidence shape every path, tree, and gathering point. The result is not a prettier park. It is a smarter one.
Every move pairs a human-centred design intent with the specific AI method that shaped it, and the outcome it targets. This is what the brief asks for: data analysis turned into practical outcomes.
A heat model built from satellite land-surface-temperature tiles and on-site sensor sweeps finds the hottest square metres first. Canopy, pergola, and water are placed against that map, not spread evenly for symmetry.
Anonymised mobility and footfall patterns reveal how people really cross Al Safa 2. A path network is generated to match those flows, so the worn shortcuts become the designed routes and dead concrete disappears.
Wind, humidity, and shade are simulated across all four seasons. Mist lines, tree clusters, and water bodies are positioned to extend the comfortable outdoor hours, especially the brutal summer afternoons.
A planting model weighs each species on water demand, heat tolerance, shade yield, and habitat value, then composes a palette led by the Sidr and Ghaf and native grasses. Beauty and biodiversity from plants that belong here.
Sensors plus a scheduling model move active zones by time and season: shaded play and sport in the heat, open lawns and events at dusk. The same ground serves more people across more hours.
An agent-based crowd model walks the design as a child, an elder, a wheelchair user, a visitor with low vision. Pinch points, slopes, and sightlines are fixed in the model, so inclusion is proven on screen, not hoped for on site.
The challenge asks entrants to show exactly how artificial intelligence shaped the design. Here is the chain, end to end, with nothing hand-waved.
Satellite thermal tiles, wind and humidity records, anonymised footfall, soil and tree survey.
Models surface the heat islands, the desire lines, the dead zones, the comfort windows.
Path networks, canopy layouts, and planting palettes are generated against those constraints.
Microclimate and agent-based crowd models stress-test each option for comfort and inclusion.
A human design team curates, refines, and grounds the output in Dubai Municipality standards.
AI is the analyst and the draftsman; people remain the authors. Every machine output is reviewed against feasibility, cost, and cultural fit before it enters the plan. The intelligence is in service of a human, water-wise, unmistakably Emirati park.
Figures are modelled design targets for the concept stage, to be confirmed against full site survey data.
Plant the densest canopy on the hottest mapped zone and lay the cooled water line first, so the park gets measurably more comfortable before anything else is touched.
Rebuild circulation to the generated desire-line network and open the flexible gathering grounds and shaded active terraces.
Activate the environmental sensing and adaptive programming, and complete the accessible sensory garden validated by the inclusion model.
SIDRA is our concept entry to Dubai Municipality's world-first AI Park Design Challenge: a Dh200,000 open call to reimagine Al Safa 2 with artificial intelligence at the heart of the process. This page is the living face of that submission.