As a Trimble eCognition trainer and now Product Manager, I have a great deal of contact with our users. Over the years, I have often received the feedback that eCognition is “not well suited for multi-temporal data analysis”. I have always struggled to understand this statement; in my previous life as an eCognition user, I created numerous complex rule sets designed specifically for multi-temporal land cover mapping. But in the end, seeing is believing…
I recently had the pleasure to meet Chris Lowe, director of imagery analysis with Land Info Worldwide Mapping based in Colorado, U.S.A. He told me of several fascinating projects they had realized using our Trimble eCognition Developer software. Two of these projects were just featured on the cover of the Spring 2018 edition of APOGEO Spatial. The article, entitled “Meaningful Mapping in Sri Lanka & Dubai – World Bank and National Geographic Society Projects” looks at two different eCognition-based mapping projects: 1) a study of Dubai’s road development over time for “project spotlighting cities’ sustainable solutions for managing urban growth” and 2) an investigation to “how building roof materials can help inform the story of the country’s economic health” in Sri Lanka.
The project in Dubai is a wonderful example of multi-temporal analysis can be automated within eCognition in a production scenario.The study was to focus on “five year intervals between 1984 and 2016… to classify and map Dubai’s road development and expansion” over a 32-year period. According to Lowe, “road networks and how they have changed can enhance the story of a region’s growth”.
Given the considerable length of the study period, the Landsat Program provided optimal temporal coverage of the 6,475 square km (2,500 sq mile) area of interest (AOI). 26 Landsat images from three different missions (Landsat 5, 7 and 8) were acquired and after just a single day of eCognition rule set development and customization, Lowe rand the classification on the data from 1984. After applying an initial classification to distinguish between urban, barren vegetation and hydrology, he integrated vector data from OpenStreetMap (OSM) and “instructed the software to identify just the roads in urban areas and map those”. This two-tier classification approach only took 30 minutes for the entire AOI.
With the initial classification complete, Chris went on to apply a cumulative classification approach to classify the remaining six time periods – “one of the powers of OBIA is its ability to take in any spatial-based data and classify whatever you tell it to” therefore allowing him to “use the previous year’s urban and roads classes as inputs for the next year. Combining each year’s results as inputs for the next year enabled eCognition to build on those and classify only the objects that hadn’t already been done. It’s smarter and more efficient. I couldn’t do that with pixel-based image processing software”.
All seven classifications were completed in about 3.5 hours. The final results showed that Dubai’s road network expanded from 4,692 km (2,915 miles) in 1984 to 13,000 km (8,077 miles) by 2016 – a 277% increase in 32 years.
According to Chris, “NGS can incorporate the city’s road expansion into their other metrics and develop a deeper understanding of Dubai’s urban growth in the last three decades”.
It is always nice to see eCognition used efficiently and successfully! The project in Dubai is a wonderful example of the software’s strengths and these strengths include multi-temporal analysis.