Persoonlijke instellingen


Mapping of tidal flat habitats with digital cameras

Uit Kust Wiki
Ga naar: navigatie, zoeken
Figure 1: Camera system: Olympus 8080 3264 x 2448 pixel at 1000 m altitude, swath: 1569 m, spatial resolution: 48 cm. MUVI IIlc 1280 x 960 pixel, swath: 733 m, spatial resolution: 57 cm, spectral bands: 568, 620, 855 nm, bandwith: 10 nm.


One of the most unique ecosystems in Europe are the tidal flats along the North Sea coast stretching from the Netherlands to Denmark. The vast extension and the changing coverage by water within the tidal cycle make it difficult to observe this area either by foot or by boat and to monitor the distribution of habitats, such as sea grass meadows and mussel beds. Remote Sensing is one possibility to solve this problem. Satellite data have the advantage of covering the whole area at once. But the spatial resolution of 10-30 m is too coarse for some applications. Furthermore, the coincidence of time of overpass, low tide and cloud free weather is rare. Survey aircraft with multispectral scanners or imaging spectrometers can fly at the right time and provide a high spatial and spectral resolution. However, their operation is expensive and flights have to be booked long time in advance with the risk of inappropriate weather during the scheduled period.

Figure 2:Image of Olympus and MUVI camera.
A further, very flexible and cheap option is to use digital cameras from a light aircraft. The data are available right after the flight and can be used to guide the ground work. Within this chapter we will report on our experiences with operating a consumer digital camera and a multispectral digital camera system on a Cessna 172 light aircraft. This test was part of the ORFEW project, which was supported by the environmental authorities of the states of Schleswig-Holstein and Niedersachsen (Lower-Saxony).
Figure 3: Left: RGB image of a macro algae field. Right: Classification of the vegetation density.


The equipment consists of 2 independent camera systems: an Olympus 8080 RGB camera with a 0.7X wide angle lens converter and a system of 4 digital black & white cameras (MUVI, multispectral video imager), each of which is equipped with an interference filter. The wavelength bands of these filters were selected according to the results of spectral reflectance measurements on the ground. Both cameras were mounted in a rack with a vertical viewing direction *(Fig. 1).

Figure 4: right: Mosaic of five images. Right: Seagrass mapping using maximum likelihood classification.
The Olympus RGB camera is fully controlled from a note- book computer via a USB connection. Images are transferred to the notebook right after capture so that a quality check is possible in nearly real-time. Furthermore, a continuous video-stream from the camera is shown on a video display for the fine navigation of the aircraft over small targets. The notebook computer also records the navigational data with an interval of 1 s from a GPS-mouse. The precise GPS time is synchronised with the computer and the camera. The MUVI system includes a laser-gyro, an electronic compass and a GPS for recording position and altitude at the time of image capture. Since the four cameras are not perfectly aligned, the pixel co-registration of the 4 bands has to be performed during the later data processing. Basis for the co-registration is the wider image of the Olympus camera.
Figure 5: Left: RGB and MUVI IR, false colour image of mussels. Right: Mussel classification.
To cover larger areas, mosaics of the images are computed by using the Autopano software, which does an automatic determination of corresponding points in the overlapping areas of the image series. The thematic inter- pretation is performed by visual inspection and by using the supervised maximum-likelihood classification method. It re- quires to identify pixels on the ground with known surface types, such as mud flat, sea grass etc. which can be used as training areas. After classification a is computed as well as the area which is covered by an object such as sea grass. Together with ground samples also the biomass e.g. of seagrass, macro- and micro-algae can be estimated much more accurately than without the aerial survey.

The main author of this article is Doerffer, Roland
Please note that others may also have edited the contents of this article.