Skip to content
Faza.ID

Menu

Back to home
  • Published
  • Research
  • AI/ML
  • Team
  • 2025

3D flood visualization for Jakarta

LSTM rainfall forecasting feeding an ANUGA flood simulation, rendered as an interactive 3D map. Published at ICCSCI 2025.

A flooded street with closed shops and trees

How it works

  1. 01LSTM forecasts rainfall
  2. 02ANUGA simulates flood spread from the forecast
  3. 03Leafmap renders the result as a 3D map

Rainfall LSTM R²

MAPE 3.3%; the forecast feeds an ANUGA flood simulation.

Water rising over a basin

An illustration of how the 3D map shows flood spread over time. Not the paper's simulation output.

The problem

Jakarta floods every rainy season, but rainfall forecasts alone do not show which streets go under water.

My part

Co-author (2nd of 4) of the published paper.

Second of four authors.

Limitations

  • A research system for one urban area, not an operational warning service.
Actual versus predicted rainfall from the LSTM
LSTM rainfall forecast against actual rainfall.
ANUGA simulation of flood depth after 24 hours
ANUGA flood depth at t = 86,400 s (24 h).