Researchers have built a computer programme that is capable of monitoring and analysing large amounts of satellite data which can assist scientists in tracking changing environmental conditions.
James Wang, professor of information sciences and technology at Penn State University, and Jose A. Piedra-Fernandez, assistant professor of information sciences and technology at the University of Almeria in Spain who is currently visiting Penn State, have designed the programme to study a vast amount of satellite data and images to monitor the ever-changing and intricate environmental conditions. They are mainly focused on analyzing mesoscale regional ocean features within the images produced by satellites. Their study was published in IEEE Transactions on Geoscience and Remote Sensing.
“All of the data and information that is continually collected by satellites and sensors can cause tons of problems for scientists, who simply don’t have the time to analyse every pixel of every satellite image,” said Wang. “Our goal has been to provide a tool that would create useful information or knowledge from this large pool of data.”
To make this programme, Wang and Piedra-Fernandez created a database of ocean structures and taught the programme to recognise changes in the ocean. The computer programme is similar to a Bayesian network, which uses probability to make decisions. Wang and Piedra-Fernandez made sure to make the programme as complex as the climate itself by separating ocean regions from land regions, adjusting for possible earth-and-solar-based interference sources, and identifying features from particular regions of ocean. The programme was then able to filter regions of the images by ranking relationships between features on scale based on relevance and strength. This allows the computer to recognise oceanic features like wakes, upwellings and eddies.
Researchers then tested the computer programme on satellite images of oceans in the Mediterranean Coast, the Iberian Atlantic and close to the Canary Islands. The images were provided by the National Oceanic and Atmospheric Administration and the Advanced Very High Resolution Radiometer. The tests consisted of over 1,000 real oceanic features, made up of 472 upwellings, 119 cloudy upwellings, 180 wakes, 40 cyclonic eddies, 10 anticyclonic eddies and 180 “misclassified” regions.
“In almost all cases, the proposed methodology improves the accuracy rate and reduces the number of features necessary to get a good ocean structures classification,” said Piedra-Fernandez. The next step is to add features such as chlorophyll and salinity concentrations. Researchers would also like to improve the image classification system.