Building Insights from Satellite Imagery with Google Earth Engine
How we used Google Earth Engine to track the impact of flooding
A nearly infinite combination of tools and services can be used to address specific business needs and the overlap between different technologies is often very nuanced. Occasionally, however, a tool comes along that can revolutionize an industry.
Google Earth Engine is one of those differentiating technologies that’s changing the way we approach remote sensing and geospatial analysis, while also improving our ability to meet environmental and sustainability goals.
What is Google Earth Engine?
Remote sensing is the process of monitoring the physical characteristics of an area by measuring how objects interact with the electromagnetic spectrum. This incorporates everything from high-resolution satellite imagery to aircraft that use radar technology to identify an object’s location. It’s a wide-ranging discipline where specific topics can encompass a lifetime of research.
Google realized that the potential of unlocking this information and applying it to climate and sustainability efforts was hindered by the difficulty to organize and process this information. In 2010, Google released Google Earth Engine (GEE) to universities, and researchers and are now rolling it out to everyone.
GEE combines multi-petabytes of data with custom functionalities that simplify how the user interacts with the data, and then supplies access to processing power on their Google Cloud Platform. What used to be a cumbersome process — where every researcher had to painstakingly build their own methods from scratch — is being replaced with one integrated experience.
With GEE and the use of satellite imagery, we can travel back in time to the 1980s to learn how land features change or quantify the impact of a natural event in near real-time — an ability that is accelerating positive change. The documentation is even written in a way so that those with little remote sensing experience can work with GEE and contribute to real-world sustainability solutions.
Building a flooding impact tool
In conjunction with Google, we wanted to build a demo that would highlight the speed, simplicity, and power of GEE. We listened to the pain points of several of our clients that are involved in remote sensing and identified the need to replace manual data operations — finding, storing, organizing, and processing the data, as well as ensuring accessibility to the right stakeholders.
A civil engineering company needed to understand the impact of natural events coastline infrastructure. Because the organization was utilizing several different software packages and bespoke data in differing formats, it took several weeks to analyze each individual climate event. Our goal was to use GEE to build an automated pipeline that would address these inefficiencies and allow them to focus their efforts on analysis.
Our team used the GEE documentation and community tutorials to quickly sift through the hundreds of available data sets and find a data collection that fit our purpose: Sentinel-1 Synthetic Aperture Radar (SAR). This technology acts somewhat like a bat using echolocation, sending out signals and measuring the response to determine elevation and detect objects. It covers each point on earth every six days (there are two satellites that work in conjunction with one another), allowing users to track changes over time. It also isn’t affected by cloud cover in the same way that satellite imagery is. With SAR, we can take a snapshot of the elevation before and after a climate event and calculate the elevation change. We found that a 0.5m threshold is a good starting point to reduce false positives while accurately identifying flooding location to a 10-meter resolution. GEE’s documentation explained how to work with the data and supplied sample JavaScript code to start building our application.
The United Nations Platform for Space-based Information for Disaster Management and Emergency Response (UN-SPIDER) has a set of open-source JavaScript code built in GEE that was used for tracking flood damage. We converted certain components of it into Python and started testing different events, such as Hurricane Beira in Mozambique. Once relatively confident that the tool was working as expected, we supplied the Python code as an open-source tool that anyone can use for disaster management.
Our next step was to incorporate other components from Google Cloud Platform (GCP), including gathering global, historic weather and catastrophic event data from Google Big Query’s public data set. Civil engineering companies keep vast amounts of geospatial data, which Big Query is not only capable of storing, but also optimally querying and accessing. Additionally, we’re using Vertex AI — a data science platform that helps bring all the data and services from Google Cloud together to build sophisticated models and applications.
In summary
The combination of GEE and GCP is a powerful and impressive tool in remote sensing. Driven by Google’s quest to organize the world’s data and simplify it for end users, Slalom is finding ways to innovate and address our most demanding climate and sustainability challenges in partnership with the vibrant and engaged GEE community. Today, we’re well on our way to building an application that replaces several pieces of legacy software, reduces manual operations and — most importantly — improves our ability to solve grand challenges.
Slalom is a global consulting firm that helps people and organizations dream bigger, move faster, and build better tomorrows for all. Learn more and reach out today.
