On clear or partly sunny days, people might look up at the sky and see straight lines of what appear to be clouds or white smoke. These lines are not smoke or natural clouds; they are contrails produced by aircraft. Contrails form because water vapor from jet engine exhaust passes through a cold and humid part of the air at high altitudes. Sometimes the jet that created the contrails is not visible overhead because winds aloft have blown the vapor trail into the observed area after the jet has passed. Naturally occurring high thin cirrus clouds do not form straight lines, they are more diffuse and irregular in shape than a contrail. Can an app be developed to help a ground observer determine the probability that an aircraft made the thin lines of white 'clouds' overhead?
Every day, more than three million people take part in commercial air travel. Given the right environmental conditions, these aircraft can produce stark features against the background sky. These contrails are comprised of different constituents than ordinary cirrus clouds, but may eventually disperse and contribute to the atmospheric amalgam. It is important to know if and where this occurs as the affect on weather patterns as well as climate change cannot be neglected.
Our team has taken steps toward distinguishing clouds from contrails. We do so by utilizing machine learning, training a classifier based on large quantities of image data separated into two categories of either "cloud" or "contrail". We then feed the algorithm a new image of the sky that has not been previously seen to determine the probability with which the new image consists of any contrail pattern. We then send the user back the probability that his or her image contains a contrail. On rudimentary test data, the algorithm appears to be able to classify correctly with roughly 90-95% accuracy.
We integrate this functionality for users who have access to Android devices. Our Android application is capable of sending the picture the user takes to a server to process using the machine learning algorithm described above, and in addition, also determines the user's location in terms of latitude and longitude.
We then use this coordinate information to cross reference with recent flight paths over a specified radius to infer if any flights may have caused any contrail patterns present in the image. The application also allows the user to obtain more information on the flights that pass overhead.
With this information it becomes easier to anticipate local weather patterns, and ultimately, global phenomena.
Watson Bluemix API
Alchemy Vision Recognition
Node.js, Python, AngularJS, etc.