Creating and sharing knowledge for telecommunications

Cloud Base Height and Wind Speed Retrieval through Digital Camera Based Stereo Vision

Janeiro, F. M. ; Wagner, F. ; Ramos, P. M.

Cloud Base Height and Wind Speed Retrieval through Digital Camera Based Stereo Vision, Proc American Geophysical Union Fall Meeting , San Francisco, United States, Vol. 1, pp. 1 - 1, December, 2010.

Digital Object Identifier:

Clouds are an important factor in Earth’s climate system. In general, the height of the cloud determines if the cloud consists of water droplets or ice crystals which have large consequences on
the radiative properties of the clouds. Also, the height plays a role on the interaction between aerosols and clouds. Furthermore, the height of low clouds is very important for air safety, especially if an instrument landing system (ILS) is not available, as is the case in small aerodromes.
Cloud height can be assessed from stereo photography using triangulation methods. However, in the past, the matching of the pictures had to be done manually, in a time consuming and error prone procedure. Recently, the developments on digital consumer electronics have made digital cameras widely available and
at relatively low cost. The use of digital cameras in stereo vision presents the major advantage of allowing for the automation of the picture matching in a process called image registration. In fact, this process can now be performed by a computer in a fully automated way. Another advantage of the stereo digital
photography is that cloud height can be estimated in various horizontal positions in contrast to LIDAR or ceilometer measurements where the cloud height can only be determined in the direction of laser beam.
In a previous work we have analyzed the main sources of uncertainty in the cloud base height and wind speed estimation using a digital camera based system. Here, we present the new developments that attempt to reduce the final system uncertainty. Two digital cameras are placed 40 m away from each other
with a vertical alignment uncertainty estimated to be better than 0.1°. A computer based triggering system is employed to remotely trigger the two cameras within 1 ms of each other. The pictures are automatically downloaded to a single computer for near real time processing.
An overview of the whole system will be presented as well as field measurements. A comparison with available LIDAR measurements will also be included.