Bangalore Climate Signals: Satellite-Based Temperature Analysis (2001-2020)
Domains
Tech Stack
Project Summary
Abstract
Bangalore's rapid urbanization created a useful case study for understanding how land-use change, vegetation loss, and atmospheric shifts show up in long-term climate signals.
This project combined MODIS, ERA5, and NEX datasets with exploratory analysis and quarterly forecasting models using Linear Regression and Support Vector Regression to study temperature behavior from 2001 onward.
The work quantified clear warming signals and surfaced an unusual inverse relationship between temperature and evapotranspiration, likely shaped by Bangalore's altitude, dry climate, wind patterns, and limited proximity to large water bodies.
What I Built
- The forecasting pipeline reached an R2 of 0.95 for mean-temperature prediction using SVR and Linear Regression.
- The long-term dataset showed a 30.03% rise in dust and smoke, an 8.13% drop in green cover, and a 29.52% decline in evapotranspiration.
- Bangalore displayed an unusual inverse correlation between temperature and evapotranspiration compared with typical climate expectations.
Impact
- Quantified urban-warming signals in a way that connected environmental degradation to visible changes in Bangalore's development pattern.
- Turned remote-sensing data into an interpretable climate narrative with practical forecasting value.
Page Info
Remote Sensing Dataset
Merged MODIS, ERA5, and NEX datasets from 2001-2020 to track aerosol optical depth, vegetation cover, evapotranspiration, precipitation, and mean surface temperature.
Climate Modeling
Built quarterly forecasting models with Linear Regression and Support Vector Regression to estimate temperature trends and future climate shifts.
Bangalore's Anomaly
Identified a rare inverse correlation between temperature and evapotranspiration, linked to Bangalore's altitude, dry climate, wind patterns, and distance from large water bodies.