An Early Warning System for Flood Detection Using Critical Slowing Down

Area of Science:

• Environmental science
• Hydrology
• Climate research

Background:

• Catastrophic events are often preceded by critical slowing down (CSD) phenomena.
• CSD indicators in time series data can serve as generic early warning signals.
• Previous applications of CSD have shown success in various scientific fields.

Purpose of the Study:

• To apply CSD theory for developing an early warning signal for floods using water level data.
• To investigate the effectiveness of quantile estimation for identifying CSD indicator trends.
• To establish a reliable flood early warning system (FLEWS).

Main Methods:

• Applied critical slowing down (CSD) theory to water level time series data.
• Utilized quantile estimation to analyze trends in CSD indicators, replacing Kendall's tau.
• Developed thresholds using quantile estimation to detect significant increases in CSD indicators.

Main Results:

• CSD theory demonstrated an increasing pattern in CSD indicators preceding flood events in Kelantan River water level data.
• Quantile estimation successfully established an early warning signal for ten out of twelve analyzed flood events.
• The remaining two flood events were detected on the first day of their occurrence.

Conclusions:

• Critical slowing down (CSD) theory is a reliable technique for developing flood early warning systems (FLEWS).
• Quantile estimation provides an effective method for setting thresholds and extracting CSD indicator trends for early flood detection.
• The developed FLEWS shows promise in providing timely warnings for riverine flood events.