UrbanGraph · Microclimate Prediction

How urban elements form a temporal heterogeneous graph to predict microclimate heatmaps.

Start from this prompt

How urban elements form a temporal heterogeneous graph to predict microclimate heatmaps.

Create a 16:9 landscape UrbanGraph conceptual methodology diagram explaining how urban elements form a temporal heterogeneous graph and predict microclimate heat maps.

Title: "UrbanGraph: A Method for Urban Microclimate Prediction | Conceptual Example".

On the left, inputs of Building, Land Cover, Tree, Dynamic Weather Conditions, and Time form a Candidate Heterogeneous Graph; Physics-informed Topology Construction prunes edges without causal relationships based on Shading, Wind, Diffusion, and Convection, retaining dynamic shadowing, wind, and static spatial proximity, feature similarity, object integrity; the main pipeline runs through RGCN Blocks, MLP fusion, and LSTM temporal propagation, outputting heat maps at t=1, t=2, t=3.

Use green for physical constraints and valid causal edges, blue for spatiotemporal features, and orange for outputs, on a white background with clean academic-illustration linework.

Use only the qualitative copy "reduces invalid edges and lowers computation".

Add ablation note

Adds a small before/after edge-count note beside the pruning stage, leaving the heat map outputs unchanged.

Try Kimi Design

Create a 16:9 landscape UrbanGraph conceptual methodology diagram explaining how urban elements form a temporal heterogeneous graph and predict microclimate heat maps.

Title: "UrbanGraph: A Method for Urban Microclimate Prediction | Conceptual Example".

On the left, inputs of Building, Land Cover, Tree, Dynamic Weather Conditions, and Time form a Candidate Heterogeneous Graph; Physics-informed Topology Construction prunes edges without causal relationships based on Shading, Wind, Diffusion, and Convection, retaining dynamic shadowing, wind, and static spatial proximity, feature similarity, object integrity; the main pipeline runs through RGCN Blocks, MLP fusion, and LSTM temporal propagation, outputting heat maps at t=1, t=2, t=3.

Use green for physical constraints and valid causal edges, blue for spatiotemporal features, and orange for outputs, on a white background with clean academic-illustration linework.

Use only the qualitative copy "reduces invalid edges and lowers computation".

Switch to deep green

Recolors physical constraints to deep green, spatiotemporal features to indigo, and outputs to amber.

Try Kimi Design

Create a 16:9 landscape UrbanGraph conceptual methodology diagram explaining how urban elements form a temporal heterogeneous graph and predict microclimate heat maps.

Title: "UrbanGraph: A Method for Urban Microclimate Prediction | Conceptual Example".

On the left, inputs of Building, Land Cover, Tree, Dynamic Weather Conditions, and Time form a Candidate Heterogeneous Graph; Physics-informed Topology Construction prunes edges without causal relationships based on Shading, Wind, Diffusion, and Convection, retaining dynamic shadowing, wind, and static spatial proximity, feature similarity, object integrity; the main pipeline runs through RGCN Blocks, MLP fusion, and LSTM temporal propagation, outputting heat maps at t=1, t=2, t=3.

Use green for physical constraints and valid causal edges, blue for spatiotemporal features, and orange for outputs, on a white background with clean academic-illustration linework.

Use only the qualitative copy "reduces invalid edges and lowers computation".

No numbers

Keeps only the qualitative line about reducing invalid edges; no numeric values appear anywhere in the figure.

Try Kimi Design

Create a 16:9 landscape UrbanGraph conceptual methodology diagram explaining how urban elements form a temporal heterogeneous graph and predict microclimate heat maps.

Title: "UrbanGraph: A Method for Urban Microclimate Prediction | Conceptual Example".

On the left, inputs of Building, Land Cover, Tree, Dynamic Weather Conditions, and Time form a Candidate Heterogeneous Graph; Physics-informed Topology Construction prunes edges without causal relationships based on Shading, Wind, Diffusion, and Convection, retaining dynamic shadowing, wind, and static spatial proximity, feature similarity, object integrity; the main pipeline runs through RGCN Blocks, MLP fusion, and LSTM temporal propagation, outputting heat maps at t=1, t=2, t=3.

Use green for physical constraints and valid causal edges, blue for spatiotemporal features, and orange for outputs, on a white background with clean academic-illustration linework.

Use only the qualitative copy "reduces invalid edges and lowers computation".