Field　measurements　with　38　weather　stations　in　Shanghai　revealed　a　strong　daytime　UHI　effect　that　Tmax　was　0.25°C–5.66°C　higher　than　suburban　areas　during　extreme　summer　afternoons.　Through　regression　analysis,　nonlinear　effects　of　urban　morphologies　on　pedestrian-level　air　temperature　were　observed.　The　TabPFN　model　outperformed　the　linear　regression,　random　forest,　AdaBoost,　and　XGBoost,　with　RMSE　of　0.71°C　for　Tavg　and　1.01°C　for　Tmax,　respectively.　Compared　with　the　linear　regression,　the　TabPFN　reduced　prediction　error　of　Tavg　and　Tmax　by　35　%　and　33　%,　respectively.　Among　the　investigated　urban　morphologies,　the　sky　view　factor,　ground　albedo,　and　impervious　ground　surface　fraction　were　the　key　parameters　affecting　afternoon　temperature.　The　sky　view　factor　was　the　most　influential　parameter,　accounting　for　about　0.22　and　0.27　of　the　variances　in　Tavg　and　Tmax,　respectively.　As　SVF　increased　from　0.06　to　0.71,　the　average　Tavg　and　Tmax　changed　by　4-5°C.　The　high　temperature　risk　assessment　indicated　that　a　50　%　reduction　in　SVF　led　to　a　significant　decrease　in　the　probability　of　high　temperature　events,　with　the　probability　of　Tmax　＞　35°C　decreasing　from　59.5　%　to　34.1　%,　the　probability　of　Tmax　＞　37°C　decreasing　from　44.3　%　to　9.7　%,　and　the　probability　of　Tmax　＞　40°C　decreasing　from　17.3　%　to　2.2　%.　Radiative　cooling　material　was　effective　in　mitigating　extreme　heat　events　(Tmax　＞　40°C).　When　ground　albedo　increased　by　20　%,　the　probability　of　Tmax　＞　40°C　decreased　significantly　from　17.3　%　to　6.5　%,　while　the　probability　of　Tmax　＞　35°C　increased　from　59.5　%　to　81.1　%.　These　findings　highlight　that　lightweight　shading　(e.g.,　trees,　retractable　covers)　effectively　reduces　solar　radiation,　while　minimizing　impervious　surfaces　(e.g.,　grass　bricks)　and　using　high-reflectivity　ground　materials　can　enhance　microclimate　conditions.　©　2025　Elsevier　Ltd