20 DoRare water-level monitoring stations for riverside flood monitoring in 4 Thai provinces

How the brief came about

The challenge in this case was to see water levels at 20 monitoring points in 4 provinces from one place. These points are in riverside communities across several districts, so traditional water monitoring was slow and incomplete. Operators wanted to identify locations at risk of flooding in good time.

Operators wanted water levels and actual images from every point on a single map. Water levels show which points are starting to reach the watch threshold, while camera images are used to check water levels by eye and are kept as evidence to support the readings.

Requirements and constraints

Every cabinet has to operate outdoors by the river, powered by solar panels and a battery and sending data via SIM card, so no power or network cables need to be run to the installation site. The main requirements in the proposal document are as follows.

• Powered by solar panels and a battery, sending data over the mobile network by SIM card
• A camera for checking water levels, with still images recorded as evidence
• 5 designated locations must have a weather station measuring temperature, humidity and rainfall
• Operators must be able to see cabinet condition remotely (temperature, humidity, air pressure, battery, fan and router) and select the cabinet's power mode

How each station measures water level

Each station measures water level in 2 ways. The first is a radar sensor that gives a numerical water level reading. The second is a camera photographing the staff gauge for AI to read. Camera images also let operators check water levels by eye and are kept as evidence to support the readings.

For power and networking, we chose cabinets powered by solar panels and LiFePO4 batteries, sending data through 4G routers, in line with the job requirements. Each station works as follows.

1. 1

   Photographing the staff gauge and measuring by radar

   The IP camera at each station photographs the staff gauge, while the radar sensor measures water level separately. Locations with a weather station also collect temperature, humidity and rainfall.

2. 2

   InSensor reads sensor values

   InSensor (X1 Board) in the cabinet reads values from the radar sensor and weather station (where fitted), then forwards them to the 4G router. The IP camera connects to the router by LAN cable and sends images directly to the central system, not through InSensor.

3. 3

   Sending data through the 4G router

   Images and measured values go to the central system via SIM card. The cabinet status page shows how much SIM data has been used and how much remains.

4. 4

   Reporting cabinet condition and selecting the power mode

   The control cabinet reports its internal condition (temperature, humidity, air pressure, battery, fan, router) to the Monitor System page. Operators can choose from 4 power modes: Normal, Save, Full Power and Plan Mode, which schedules operating times in advance.

5. 5

   AI reads values and displays them on the dashboard

   The system uses AI to read water levels from the staff gauge images approximately every 15 minutes, then displays them on the station map and water level history table, alongside camera images. Operators can export the history as a CSV file for further reporting.

6. 6

   Alerts based on each station's thresholds

   System administrators set 3 threshold levels separately for each station. When the water level reaches the critical threshold, the system sends alerts by LINE and/or email as configured. The system can also send equipment notifications, such as a camera going offline or a low battery.

What we built and installed

We supplied and assembled 20 DoRare Station control cabinets, developed the web dashboard, carried out on-site installation and prepared a user manual. The main items are as follows.

• 20 control cabinets, codes CAB-001 to CAB-020, each with an IP camera, radar water level sensor, InSensor (X1 Board), solar panel, LiFePO4 battery and 4G router
• 5 weather stations, installed in designated cabinets
• DoRare web dashboard: station map, camera images, water level history exportable as CSV files, Monitor System page and 4 power modes
• 3-level alerts (normal/watch/critical) with thresholds set separately for each station, sent by LINE and/or email
• Viewing through a web browser, with separate permissions for system administrators (Admin) and general users (Viewer)
• Scale calibration of each station's staff gauge, by dragging the upper and lower points to match the marks on the gauge and entering values in centimetres
• System user manual, and a 24-month hardware and software warranty

Confirmed results

We installed DoRare Station cabinets at the 20 listed locations in 4 provinces in 2026, and have site photographs for all 20 locations, organised by location name. Operators can view images and water levels for every station from a web browser, with images and water levels updated approximately every 15 minutes.

We prepared a system user manual and issued a 24-month hardware and software warranty. This case does not yet have figures measuring usage results, so we report only what is confirmed by documentation.

What we learned

This work offers 3 principles for unattended field measurement stations.

• Values read from images should have the original image stored alongside them. In this work, staff gauge images serve both as data for AI to read and as evidence for people to check by eye, and each station also has a radar sensor measuring separately.
• Field equipment must also report its own condition, not just measured values. In this work, operators can see the battery, fan, router and SIM data of every cabinet, and receive alerts when a camera goes offline or a battery is low.
• Settings must be tied to the station, not a single value for the whole system. Each location has its own staff gauge and camera position, so scale calibration and 3-level alert thresholds are set separately for each station.

How this approach can apply to your site

If your agency looks after several water level monitoring points without electricity or network cabling, you can use the same approach as this case: solar-powered cabinets, data sent via SIM, water level measured from images and radar, and all points combined on one dashboard. What needs adjusting is the camera position, calibration values and thresholds for each location.

• Prepare the location name, address and latitude/longitude coordinates of every station from the outset, so cabinet codes can be matched correctly to positions on the map
• Estimate monthly SIM data usage for each cabinet before choosing a package, because the system regularly sends images to the central system
• Plan long-term maintenance of field equipment, especially batteries, solar panels and power mode selection during periods of low sunlight