Naïve Water Level Monitoring using Near-Infrared Band

About the Sustainable Development Goals:
On the 27th of September, 2015, the United Nations General Assembly adopted the 2030 Agenda for Sustainable Development which was built upon the achievements of the Millennium Development Goals and seeks to address any unfinished business in the past 15 years. This universal Agenda comprise 17 Sustainable Development Goals (SDG) built from 169 ambitious and transformative targets. It details its vision to end poverty and hunger in all their forms, foster peaceful and inclusive societies free from fear or violence, and protect the planet so that it can provide for the future generations. These goals balance three specific dimensions, namely, the economic, social, and environmental.¹

Access to safe water and sanitation and sound management of freshwater ecosystems are essential to human health and to environmental sustainability and economic prosperity.

More than 2 billion people globally are living in countries with excess water stress, defined as the ratio of total freshwater withdrawn to total renewable freshwater resources above a threshold of 25 per cent. Northern Africa and Western Asia experience water stress levels above 60 per cent, which indicates the strong probability of future water scarcity.²

This project aims to create a water-level monitoring system for dams, rivers, and/or watersheds for all kinds of management purposes.

Satellite Data from Planet ©

Planet is an Earth imaging company based in San Francisco, CA. The company designs and manufactures Triple-CubeSat miniature satellites called Doves that are then delivered into orbit as passengers on other rocket launch missions. Each Dove Earth observation satellite continuously scans Earth, sending data once it passes over a ground station. Together, Doves form a satellite constellation that provides a complete image of Earth at 3–5m optical resolution.³

The Doves constellation has a revisit rate of 24 hours, so monitoring of water bodies can be done with the same resolution, given there are no obstructions (e.g. clouds) on the optical image. A Dove image has 4 bands: Red (R), Green (G), Blue(B), and Near-Infrared (NIR).

A sample R-G-B image from Ambuklao Dam, in Benguet, Philippines taken on February 24, 2014 is shown below:

RGB Image of Ambuklao Dam, Benguet, Philippines, February 24, 2017

What is Near-Infrared?

Near-infrared (NIR) light includes wavelengths between 700 and 1,100 nanometers. Water absorbs NIR, so these wavelengths are useful for discerning land-water boundaries that are not obvious in visible light. Plants, on the other hand, reflect near infrared light strongly, and healthy plants reflect more than stressed plants. Finally, near infrared light can penetrate haze, so including this band can help discern the details in a smoky or hazy scene.

The corresponding Near-Infrared image from the Ambuklao dam image above, with values plotted on a viridis color palette:

Near-Infrared Image of Ambuklao Dam, Benguet, Philippines, February 24, 2017

By plotting the histogram of each radiance value on the image, we can find the threshold for values that correspond to water. This is done by finding the knee of the plot after the first peak; the peak correspond the body of water of interest. From the threshold, we can create a mask from the NIR image. To avoid getting small patches of water area, the mask creation process is forced to only select the largest contiguous region.

[L] Histogram of Near-Infrared Radiance Values. [R] A water mask of the NIR band image from a threshold of 2400.

By monitoring the Ambuklao for several months, we can track the changes on the level of water. On the image below, we looked at 5 months of satellite data, from before the onset of summer up to the rainy season (February 2017 to September 2017).

Visualization of Ambuklao Dam from the onset of summer to the rainy season in the Philippines

The water level is tracked by counting the pixel over the generated water mask. The same can be done for other dams on different regions. Monitoring can be done on a daily, weekly, monthly basis depending on the availability of satellite images. Different frequency of tracking is shown on the images below.

Water Level Monitoring of other Local Dams

Angat Dam (Bulacan, Philippines)

A concrete water reservoir embankment hydroelectric dam that supplies the Manila metropolitan area water.

Magat Dam (Isabela, Philippines)

One of the largest dams in the Philippines; used for irrigation of agricultural lands, flood control, and . power generation through the Magat Hydroelectric Plant.

Binga Dam (Benguet, Philippines)

Sister facility of Ambuklao Dam; connected to a hydroelectric power plant.


¹ UN General Assembly, Transforming our world : the 203 0 Agenda for Sustainable Development, 21 October 2015, A/RES/70/1, available at: [accessed 19 November 2017]



One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.