Roberta Kwok interviewed me for a great feature on the use of satellite data and remote sensing tools in ecology, published in Nature yesterday. The article makes a great point: satellite data has become increasingly accessible and we are seeing an explosion of its use to investigate a wide range of ecological research questions.
Take, for instance, our work on tidal mudflats. In 2010 I spent more than a year downloading Landsat images, pre-processing them and running image classifications on an image-by-image basis. These days, I routinely access hundreds of satellite images at a time, apply cloud masks, develop composite metrics and run similar classifications in just a few minutes on Google Earth Engine.
At the global-scale, our new intertidal mapping project used nearly three-quarters of a million Landsat images and about 100 CPU-years of processing (~25 years on a single computer) to develop a high-resolution global time series of the intertidal zone, yet we ran it in parallel in just a few weeks on Earth Engine. This analysis will be published later this year.
This type of workflow has also been implemented in our new remote sensing app, remap (https://remap-app.org). We used more than 300,000 images to produce a single cloud-free image composite at the global extent. Users can quickly and easily classify by applying a random forest classifier, chosen for its ability to produce accurate classifications from a range of predictor layers.
With a range of new sensors recently coming online and several sensors in development and set to be launched in the next few years, the possibilities really are exciting. See the article here: https://www.nature.com/articles/d41586-018-03924-9