Big Data & AI for Earth systems

My work sits at the nexus of big Earth data, AI and geophysics. I build computational pipelines and physics-informed models that turn satellite, geophysical and field data into insight — to anticipate natural hazards, find water and mineral resources, and make forecasts trustworthy.

• 2Deep-learning frameworks

  Research axis 01

  Big Data in Earth Systems

  Problem

  Satellites, geophysical surveys and sensor networks produce far more data than traditional geoscience workflows can turn into insight.

  Approach

  Computational geoscience pipelines and physics-informed, explainable deep learning (GeoPriorSubsNet, XTFT) that scale from raw data to decisions.

  Explore
• R² > 0.90Subsidence model accuracy

  Research axis 02

  Natural Hazards

  Problem

  Land subsidence, contamination from failing landfills and wildfires threaten cities, water and ecosystems — often faster than monitoring can follow.

  Approach

  Physics-guided subsidence forecasting, AI detection of leachate from geophysical data, and early warning for wildfire risk.

  Explore
• >90%Drilling success rate

  Research axis 03

  Groundwater & Mineral Exploration

  Problem

  Finding water and mineral resources where data are scarce remains costly and uncertain.

  Approach

  AI-assisted EM, CSAMT and resistivity methods, supported by open-source tools like pyCSAMT and watex.

  Explore
• 100+Plot functions in k-diagram

  Research axis 04

  Uncertainty & Trustworthy Forecasting

  Problem

  Making reliable decisions requires a trustworthy understanding of forecast uncertainty.

  Approach

  Developing novel polar diagnostics (k-diagram) to analyze forecast coverage, reliability, and severity.

  Explore

1. 01

   Observe

   Electromagnetic (AMT/CSAMT), resistivity (ERT), InSAR and borehole data from the field.

   Big DataGroundwater & minerals
2. 02

   Model

   Physics-informed networks that respect hydrogeology, not only patterns in the data.

   Big Data
3. 03

   Diagnose

   Uncertainty diagnostics that show when and where a forecast can be trusted.

   Uncertainty
4. 04

   Decide

   Drilling sites that find water, and subsidence risk maps planners can act on.

   Natural hazardsGroundwater & minerals
5. 05

   Share

   Open-source packages and a desktop application so others can reuse the methods.

   Software