METHODS FOR MULTITEMPORAL ANALYSIS OF SATELLITE DATA AIMED AT ENVIRONMENTAL RISK MONITORING

In the last years the topic of Environmental monitoring has raised a particular importance, also according to minor short-term stability and predictability of climatic events. Facing this situation, often in terms of emergency, involves high and unpredictable costs for public Agencies. Prevention of damages caused by natural disasters does not regard only weather forecasts, but requires constant attention and practice of monitoring and control of human activity on territory. Practically, the problem is not knowing if and when an event will affect a determined area, but recognizing the possible damages if this event happened, by adopting the adequate measures to reduce them to a minimum, and requiring the necessary tools for a timely intervention. On the other hand, the surveying technologies should be the most possible accurate and updatable in order to guarantee high standards, involving the analysis of a great amount of data. The management of such data requires the integration and calculation systems with specialized software and fast and reliable connection and communication networks. To solve such requirements, current satellite technology, with recurrent data acquisition for the timely generation of cartographic products updated and coherent to the territorial investigation, offers the possibility to fill the temporal gap between the need of urgent information and official reference information. Among evolved image processing techniques, Change detection analysis is useful to facilitate individuation of environmental temporal variations, contributing to reduce the users intervention by means of the processes automation and improving in a progressive way the qualitative and quantitative accuracy of results. The research investigate automatic methods on land cover transformations by means of "Change detection" techniques executable on satellite data that are heterogeneous for spatial and spectral resolution with homogenization and registration in an unique digital information environment. In the present work we tested some areas of study particularly interesting for the knowledge of the morphology changes of land cover, in particular the area of Fasano in Apulia Region (Italy) and protected area of the Park of Alta Murgia, both of them with frequent episodes of land transformation. We tested the usability of heterogeneous and freely available images to realize a DEM extraction process to achieve fast and low cost system of analysis. We used archival stereo-pairs Ikonos and LIDAR survey comparing with Aerial photogrammetric DEM extraction.


INTRODUCTION
In recent years research in the field of geometric correction of satellite data has reported remarkable methodological advances, implementing registration and ortho-rectifying algorithms which are now consolidated methods for the international scientific community.Generally, such procedures are carried out to correct or to eliminate image errors due to the bad functioning of sensors and to the atmospheric diffusion effects.Quality of data also depends on the intensity of spectral distribution of energy received by the sensor, with significant variations in its passage through the atmosphere.
The automation and repeatability of the procedure on constantly updated data will permit the development of a monitoring system for land cover transformations with environmental risk, not only to support preliminarily decisions in strategic planning contexts, but also as a tool to verify strategies and accomplishments, allowing modification of actions when the expected effects are not achieved.
To detect and classify a landslide, it is necessary to view the size and contrast of its features and the morphological expression of the topography within and around the landslide.Determining parameters are the type of movement that has occurred, the degree of present activity of the landslide, and the depth to which movement has occurred.The most common remote sensing tools used for the detection and classification of landslides are satellite imagery and aerial photography.
Monitoring landslide movement involves the comparison of landslide conditions over time, including the aerial extent of a landslide, its speed of movement, and the change in its surface topography (i.e.DEM comparison) .
The fundamental merits of the high resolution remote sensing are the ability to perform surveys at regular intervals in the operation, the characteristics of the image and the revisit times.These features are very useful in environmental monitoring especially in case of emergency or also in medium scale cartographic production, particularly in the zones of difficult access and in developing countries.The satellite Ikonos and Geoeye is able to acquire images in stereopair with two different perspectives, during the same orbital transit and with a revisit time of 3 days for every point on the terrestrial surface and it can effect two-way scannings and acquire up to 7 crawled adjoining strips (each of around 11-13 kms) in a single step.

STEREO IMAGES
IKONOS® Pro 1-meter and 4-meter products are perfect for projects requiring high-resolution imagery and positional accuracy when ground control may be costly, difficult, or impossible to acquire.
Providing a strong base for three-dimensional feature recognition, extraction and exploitation, the product provides two images with stereo geometry to support a wide range of stereo imagery applications such as DEM creation and threedimensional feature extraction.Stereo products in epipolar or map projections provide RPC camera model data.
The present study reports on the metric comparison between the DEMs respectively realized with the Stereo aerial pair and with the satellite Ikonos stereo pair, and another test with photogrammetric DEM and a Lidar survey.In a short time we will compare also GeoEye-1 stereo pair.The second test was performed on the area of Fasano in Apulia region with DEM extraction by Aerial photogrammetry with 8 mt.grid provided by regional Gis and a Lidar DEM with 1 mt.grid produced by SIT s.r.l.company in Noci (BA).

COMPARISON
The comparison between the two products obtained in test 1 was carried out by examining some elevation profiles and a number of test points throughout the area.In these first elaborations without the use of the GCP (Figure 8) but only with the satellite RPC, we see a systematic error in the positioning layout.(Table 1) In the second step we repeated operations using GCP by GPS survey, obtaining the results described in the following Table 2 and in Figure 10.

Figure 1 .
Figure 1.Test area Ikonos + Aerial Figure 1 shows the test area, chosen in the Basilicata region for its characteristics of rockfall area, which was part of the comparison indicated.The Ikonos stereo pairs were used for the extraction of DEM by automatic correlation, with different software (LPS, Socet Set, Micromap) by geometric parameters of the satellite with the relevant RPC.The stereo pairs were obtained using a Wild 15/4 UAGA-F aerial camera with C=153.22 mm wide frame of about 1: 16000 suitable for cartography at a scale of 1: 5000.The cartography and its DEM autocorrelation were obtained using standard procedures of Photogrammetric survey.

Figure 11 .
Figure 11.3Dview of DEM

Figure
Figure 16.Differences between two DEMs (test Figure 16.Differences between two DEMs (test