Early Warning Systems Rarely Fail at the Sensor. They Fail at the Last Mile.

Why detection alone does not keep communities safe?

Most early warning systems can detect danger in time. The harder part is making sure the warning reaches people and turns into action. Here is where the chain usually breaks, and what a working last mile looks like.

Topic

Disaster Management

Publish Date

Sep 30, 2026

Overview

When a flood or landslide hits a community that has an early warning system, the first questions are usually about the technology. Did the sensor work? Was the data late? Was the forecast wrong?

In many cases, the sensor did its job. The water level was recorded, the threshold was crossed, and the system knew. The warning just did not reach the people who needed to move. Or it reached them too late. Or they did not know what to do with it.

This part of the chain is often called the last mile. The UN's Early Warnings for All initiative describes an effective early warning system in four parts: risk knowledge, monitoring and forecasting, warning dissemination, and preparedness to respond. Budgets tend to focus on the second part. Many failures happen in the last two.

As Indonesia heads into the rainy season, it is worth looking at where that chain usually breaks.

Main Points

1. Detection Is the Easy Part

Sensors have become cheaper, more reliable, and easier to connect. Water level sensors, rain gauges, and CCTV can now send data every few minutes to a central dashboard.

That progress is real. It can also create a false sense of safety. A dashboard full of live readings looks like a working early warning system, even when no one outside the control room is ready to act on it.

Questions worth asking about any monitoring setup:

who is watching the data outside office hours
what happens when a sensor goes offline
how long it takes to get from a reading to a decision
whether people downstream even know the system exists

Detection tells you what is happening. A warning system also has to tell people who needs to move, and when.

2. A Threshold Only Works If Everyone Agrees on It

Every alert starts with a threshold. One water level means standby, a higher one means evacuate. On paper, this is simple.

In practice, thresholds are often set by one agency and understood differently by everyone else. A number that means "prepare" to a technician may mean nothing to a village head.

Good thresholds usually share a few traits:

they are based on local history, such as past flood heights at that exact location
each level is tied to a specific action, beyond just a colour code
they are agreed on with the people who have to respond
they are reviewed after every real event

If a threshold is crossed and no one knows what they are supposed to do, the alert is just a number on a screen.

3. One Alert Channel Is Never Enough

Messaging apps are fast and cheap. They also depend on phones, signal, battery, and someone checking their screen at 2 a.m.

Floods and landslides often happen at night, in heavy rain, sometimes during power outages. Any single channel can fail at the exact moment it is needed most.

A resilient alert setup usually layers several channels:

instant messages to responders and local officials through apps such as Telegram or WhatsApp
sirens that people can hear without a phone
announcements through mosques, community leaders, or local radio
a public map or dashboard where residents can check conditions themselves

Redundancy can look wasteful on a budget sheet. During a real event, it is what keeps the warning moving.

4. What a Working Last Mile Looks Like

When the full chain works, the result is quiet. People leave before the water arrives, and the story the next day is about damaged property, with no lives lost.

One example comes from Bekasi, West Java. On 11 February 2026, a flood early warning system sounded its siren about four hours before floodwater reached residential areas. The evacuation was completed around 30 minutes before the water came in.

The sensor was only one piece of that outcome. The rest came from:

thresholds tied to clear actions
alerts sent to officials and sounded through sirens at the same time
residents who recognised the siren and knew where to go
a team that acted on the alert right away

That is the real last mile. Technology can shorten the time between detection and action. People still have to close the gap.

Summary

Early warning systems are often judged by their sensors. A better measure is whether people moved in time.

Detection has improved a lot in recent years. The weaker links are usually thresholds nobody agreed on, alerts that travel through a single channel, and communities that were never told what a warning means.

Before the rainy season peaks, it is worth testing the whole chain, from the first sensor reading to the last household.

A warning only counts when it leads to action.