Challenges of remote sensing


The challenges of mapping the submerged vegetation in the coastal zone using airborne remote sensing.

The mapping of the submerged vegetation using optical remote sensing from airborne sensors is a  bit more difficult than e.g. the mapping of terrestrial vegetation as it has been performed in a number of years.

The water and its constituents limits the usable spectral region
In the mapping of terrestrial vegetation by scanners but also near- infrared aerial photography, the characteristic high reflectance of vegetation in the near-infrared region is utilized to discriminate between vegetation and other targets. Variation in the reflectance properties in both the visible and the near-infrared is used to discriminate between the various vegetation types.
In the mapping of features on the bottom of lakes and the sea, the effective usable spectral region is practically limited to the spectral region below 700 nm and for depth more than a two meters even below 600 nm. See also >

The upwelling signal to the sensor is low and disturbed
The mapping of the bottom features in the optical region relies on that information regarding the bottom features shows up as variations in the radiance directed towards the sensor. The signal received by the sensor is however made up by contributions from a variety of sources of which the radiance reflected at the bottom not necessarily is the majority. In the figure below is a part the complexity seen.

Due to the absorption by the water the radiance levels are low compared to the situation on land. Where both the coastline and the water are to be mapped, there is furthermore a very huge dynamic range between the reflected radiance from a sandy beach to the low radiance levels of deep water. This set high requirements to acquisition and post-processing at the same time to maintain both an acceptable dynamic range and good discrimination of values in the low end of the signal.
A disturbing factor is also the fact that the proportion of the radiance with is scattered from the water column varies with the actual water depth.
For the separation between the reflected radiance from bottom and the upward scattered radiance from the water constituents is it also a problem that some of the characteristic pigments are present in both the water and the vegetation to be mapped.
Reflections in the interface between air and water is a special problem, which can dominate larger regions of an image and almost inhibit the extraction of information regarding the bottom features. For scanner images, a special methodology for sun glint removal has been developed in order to reduce the impact.

Field sampling is difficult and expensive
Sampling of information regarding the target types seen in the images is a necessary element for training of  the classification of the acquired images. The sampling of this information is, compared to the terrestrial situation, both difficult and expensive since boat and diver assistance is required.
A special problem in this relation is the selection of sampling points, since it is difficult to assess the variations to expect and know where to find the characteristic types and densities of the vegetation. 

The geographical orientation is difficult in the images
The geo-orientation of images is in many remote sensing applications based on identifying corresponding points in the image and reference material, or in the case of aero-triangulation, corresponding points in neighboring images.
When performing remote sensing campaigns over water, there is normally no targets that can also be identified in a reference map, and corresponding points in neighboring images can only be found in shallow water. 



Please do not hesitate to contact us if you have comments, questions or suggestions to improve these web pages. 
Please E-mail to the Webmaster
This page was updated: 19. September 1999
These pages are maintained by Michael Stjernholm, NERI

Rescoman - Quotations allowed with reference.