A personal notebook of AI-assisted, solo research in remote sensing.
These notes document personal research and experiments. Each post and project has its own stage of progress and scope of validation. All experiments and writing are carried out with the help of AI.
INCHEON — SEOUL
Reconstructing a day of temperature changes over Seoul by combining Landsat–Sentinel-2 spatial detail, VIIRS-based thermal response and GK2A temporal information. An exploration of feasibility and validation limits through 25 hourly simulated temperature fields and observational comparisons.
Combining Landsat thermal data with Sentinel-2 optical indices to map surface temperature on a 10 m grid, then comparing spatial patterns with HOTSAT. A note on finer thermal detail, its potential uses, and the limits of validation.
01 / COLLABORATIVE PROJECT
Exploring thematic maps of tidal flats worldwide.
A tool in development with global coverage as the goal. Working with specialists from different fields, we are experimenting with creating and exploring a range of tidal-flat thematic maps.
Open explorer
02 / COLLABORATIVE PROJECT
An experiment in monitoring night-time activity with light and thermal observations.
A collaborative experiment in building a night-time activity monitoring tool that brings together night-time lights, temperature, and thermal-anomaly detection. Development focuses on exploring how these observations can be interpreted together.
Open demo
03 / SUPER-RESOLUTION RESEARCH
Experiments with finer thermal and night-light grids.
I explore which super-resolution methods suit thermal infrared and night-time light data. By experimenting with and reproducing different approaches, I document their potential uses and limitations.
THERMAL IR
Landsat temperature on a 30 m grid (native thermal observations at 100 m), compared with optically guided estimates on a 10 m grid. Both use the same 20–55 °C colour scale.
VIIRS DNB
VIIRS DNB reference on a 500 m grid and Landsat-guided spatial allocation on a 30 m grid. An existing research demonstration pending further processing. Both use the same radiance scale; grey denotes missing data. DNB observes night-time lights, not thermal infrared.
Research-stage estimates. A finer output grid does not represent new high-resolution observations.
