Google Earth

Jul 10, 2017

5 min read

Teaching students across disciplines to detect, map and characterize changes to the Earth

By Nicholas Clinton, Developer Advocate, Google Earth Engine

As satellite imagery and other geospatial data becomes ever more ubiquitous in our changing world, training the next generation to extract information from the expanding archive of imagery is becoming of critical importance. Ran Goldblatt, at the UC San Diego School of Global Policy and Strategy, is one of the educators who recognized that Google Earth Engine would be the perfect tool for the job. Because Google Earth Engine is free for educational use, and because it runs in Google’s Cloud, students wouldn’t need expensive software of powerful computers to conduct novel, impactful and data-intensive analyses in an educational context.

By using Earth Engine to teach remote sensing, Ran’s graduate-level course at UC San Diego prepares students for today’s competitive job market, giving them tools that will allow them to tackle the world’s persistent challenges. The course, now in its third year, attracts students from diverse backgrounds across UC San Diego: from the humanities and social sciences to the more traditional “hard” sciences. But they all have one thing in common: they want to use Google’s Cloud technologies to better understand the world we live in.

Students first learn concepts in Remote Sensing, including types of remotely sensed data and sensors, wavelengths, spectral indices and methods for image interpretation and image classification. They are introduced to the Earth Engine Code Editor and learn how to write code for performing geospatial workflows. They learn basic concepts and terminology of Earth Engine, such as the difference between images and image collections, how to filter image collections and visualize images, how to reduce image collections and map functions, add and analyze vector data and work with feature collections. The most exciting application is, of course, converting data into meaningful information. This is, after all, what Google’s Cloud technologies are designed for. Students learn, for example, how to create charts, calculate statistics over time and space, perform edge detection, linear regressions and more.

The most exciting application is, of course, converting data into meaningful information. This is, after all, what Google’s Cloud technologies are designed for.

Changes in built-up land cover, Ho Chi Minh City, Vietnam (student: Johannes Veerkamp)

Many of the students come from disciplines that traditionally do not use remotely sensed data. Archaeology is one example. When Ran first taught the course in 2016, Brady Liss, an anthropology graduate student in the UC San Diego Division of Social Sciences, tested the potential role of Google Earth Engine in archeological research, identifying archaeological sites in the Faynan region of Southern Jordan. International affairs is another example. William Honaker, a Master of International Affairs student at the School of Global Policy and Strategy, evaluated the spatial dimension of foreign investments in North Korea as observed in satellite imagery. Only when the students start using Earth Engine do they realize the every phenomena in our world has a spatial dimension that can be observed from space.

Students investigate “real-world” problems they care about

Spectral Indices to Identify Rice Paddy Area in the Mekong River Delta in Vietnam (student: Giang Thai)

The students’ final projects are far-reaching and diverse, demonstrating that remote sensing analysis with Earth Engine is fundamental for research in many disciplines: ecology, economics, human rights, urban and regional planning, oceanography, archeology and more, and that there is an increased demand among students to learn these tools:

  • Giang Thai used Landsat data to identify annual trends in the extent of rice paddy in the Mekong River Delta and to estimate levels of salinity and saltwater intrusion in the region.
  • Kyle Navis used forest change data in Earth Engine to detect deforestation along the Santa Cruz-Puerto Suarez highway corridor in Eastern Bolivia and to understand the effect of construction of new roads on deforestation.
  • Johannes Veerkamp identified annual trends in urbanization in Ho Chi Minh City, Vietnam, and compared the performance of different types of classifiers in predicting changes in the extent of built-up land cover.
  • Margo Zlotnick evaluated the potential of remotely sensed data for mapping agricultural production in rural Peru and methods that would potentially replace the traditional expensive ground survey mapping.
  • Seungwan Kim focused on Malaria occurrence in Malawi and used Earth Engine to evaluate the relation between trends in the extent of above-ground water and the distribution of malaria occurrence in the country.
  • Yuwen Xu examined the spatial characteristics of economic development and evaluated the relation between economic development in the Yangtze River Delta Economic Zone, China, and changes in the intensity of nighttime light and the extent of live vegetation.

Ran says that today’s students are curious and passionate to learn how to use novel tools and technology to answer real-world questions. Whether they will end up working in the public or in the private sector, they will have the skills to use geospatial data to benefit for the greater society.

Ran’s lectures and other educational materials are available on Earth Engine resources for higher education webpage. Let us know if you’re using Earth Engine in your teaching or if there is other educational content for Earth Engine you’d like to see!

(Thanks to Ran Goldblatt, Anthony King, Chris Herwig.)