Methodology
A transparent model,
not a crystal ball.
TerraShield transforms real environmental observations from the JKUAT Conduit Weather Station into explainable modeled risk scores. It is designed for climate awareness and decision support—not official warnings.
From data to impact
Observe
JKUAT Conduit environmental measurements
Assess
Deterministic flood & heat risk models
Simulate
Explore hypothetical scenarios
Act
Decision-support recommendations
Flood model
risk = rainfall × 0.45 + terrain × 0.3 + drainage × 0.25
Inputs are normalized to a 0–1 range, weighted, then converted to a 0–100 score. When Conduit is active, the rainfall input comes from the actual daily rainfall total measured at the JKUAT Conduit Weather Station.
Heat model
risk = temp × .55 + built-up × .25 + vegetation × .20
When Conduit is active, the temperature input comes from the SHT sensor (temp_sht) at the weather station. Lower vegetation coverage contributes to vulnerability in the prototype scenario.
Conduit normalization
Raw Conduit sensor values are strings and are converted to numbers during normalization. The latest observation is selected by timestamp — not simply the first record.
rainfallMm = rg1tt (daily total)
Clamped to 0–150 mm to match the flood model's input range. Falls back to rg1 (gauge reading) if daily total is unavailable.
tempC = temp_sht
Clamped to 20–45°C to match the heat model's input range. Falls back to temp_bmx or temp_mcp if SHT is unavailable.
Prototype risk categories
These are illustrative prototype thresholds, not official risk classification standards.
Limitations
Modeled scores are for climate-risk decision support and are not official emergency warnings or guaranteed predictions.