Discrimination of different water layers with TerraSar X images in “La Albufera de Valencia, Spain”.
Pedro Miguelsanz(1) and Miguel Angel García Fernández(1)
(1) Tragsatec, Julián Camarillo nº 6 B, 28037, Spain
This paper will show the methodology and conclusions of a Technologic Innovation Project which main objective has been to check the utilities and applications of TerraSar X images to classify different water layers, used in combination with Spot data and field data, in the area of “La Albufera”. This is a Natural Park located in the province of Valencia (Spain) and it is also a traditional place with rice crops. The farmers’s parcels are under the European and National Agro environmental regulation which obliges them to preserve the habitat and to keep the rice plots flooded a certain period of time between the months of Fall and Winter. Since the year 2000 this measure has been checked and monitored annually with the support of optical Earth observation data.
The input image data used were SPOT and TerraSar X images acquired in the months of November and March, trying to get them as proximate in time as possible. A classification process was carried out in order to get different water layers from the SPOT images and a field campaign was used as ancillary data as well. After that, a classification process with the TerraSar X images was performed using the thresholds and the class information from the previous Spot classification.
Setting up the acquisitions in the times of high inundation (November) and no inundation (March) in the test area, it was very easy to discriminate a class of soak areas, a class of permanent and a class of stationary inundation. Both classifications, from TerraSar X and from Spot images, were spatial analyzed in order to achieve numeric information about the class assignation. In this validation process, Spot classification was used as a reference document in order to check and verify the TerraSar X process.
Therefore, it can be noticed that dry soils or vegetated soils with a very few moisture on the surface are very easily confused in the TerraSar X classification, since the brightness values are very similar and difficult to distinguish, while in the Spot classification both classes are very well separated.
Analyzing the results, it can be seen that TerraSar X images are not better than optical ones for different water layer classification processes. But opposite to this drawback, TerraSar X can get images under bad weather conditions, while the traditional sensors based in optical spectrum could not see through the clouds. Besides, TerraSar X thanks to his images of high spatial resolution and dual polarization possibilities has improve the results in classification processes with respect to the ones we carried out before with ERS and Envisat SAR images.
Finally, it would have to be tested the quad polarization of TerraSar X images in order to check if total polarization returns from the terrain objects can improve the classification results. Besides, other SAR sensor would also have to be analyze, like for example PALSAR L, to see if it can overcome the class assignation where there is a herbaceous stratum above the continuous water layer.