A system for managing a database of satellite images for the study of natural object monitoring algorithms has been developed. The prototype of such system consisting of seven interconnected tables was created using Microsoft Access 2007. These tables contain the satellite name, sensor type and its characteristics, information about the snapshots, their parameters and, the image area information, the availability of the vector and raster maps for the snapshot, and information on sub-satellite measurements. For the forest vegetation, a technique for eld measurements has been developed to determine its species composition and coordinates using a GPS receiver. An example of lling the database with information on eight types of vegetation for the sites of the Sverdlovsk region and the environs of Yekaterinburg, as well as satellite images of Terra, Aqua, Landsat and SPOT-4 is given. The developed database allows to solve such tasks as selecting images of a certain area and object for a given shooting time from the selected satellite which is necessary for analyzing algorithms for monitoring and natural objects classi cation by the remote sensing data.
The directory of the forest vegetation images is to be created for investigation of the remote sensing data processing algorithms solving problems of forest cataloguing and monitoring. This also involves the development of methods of collection and storage of information about species composition and the forest areas location.
Creation of the directory of forest vegetation standard parts is possible if there is information about the breed mapping of forest stands. The lack of such information makes necessity to develop the methodology for selection of the standard areas and conduct the eld studies to determine their coordinates and species composition. According to the results of measurement on the ground by means of the GPS receiver, the vector layers (shape- les) are to be formed that contain the forest formula and vegetation characteristics in the attribute information. By this, the directory is created that contains a number of standard areas with di erent species, age composition of the vegetation and satellite images corresponding to them.
The present paper discusses a solution of these problems in Yekaterinburg and Sverdlovsk region, Russia. To determine characteristics of the forest areas, the measurements have been made in several terrain sectors located: near Shartash Lake (Yekaterinburg), in the natural parks Deer Streams (Nizhnie Sergi), Pripyshminskie Bory (Talitsa), and near Revun Rift on the Iset River (KamenskUralskiy).
Formation of such directory of vegetation standard sites greatly simpli es the process of corresponding images selection. In the turn, this speeds up solving the problems of processing the classi cation and vegetation monitoring on the regional level. 2
The areas with uniform vegetation structure have been selected on the terrain
during the formation of the sub-satellite information. The structure of tree types
on the percentage basis (the so-called forest formula) has been recorded for these
areas. Their geographical coordinates were also measured using the GPS receiver
Garmin eTrex Vista Cx [
The measurements have been transferred to the PC memory. As a result, the directory has been created for the shape- les grouped according to the type of vegetation and their location. The fragment of such directory is shown in Fig. 1.
For visualization, the created shape- le (vector layer) of labels made on the route has been combined with the raster layer, i:e:, the space image area, for which sub-satellite measurements have been carried out. For more precise alignment of the vector and raster layers, the automatic or manual mapping by the characteristic points (for example, on lakes or river bends) has been made.
The most important parameter is the shape- le attribute information, which initially includes the coordinates of the point, its projection coordinates, identi cation number, information about measuring device, the date and time of shooting, the height above sea level, and some additional information (Fig. 2).
The information about vegetation has been added to the attribute table
based on the records of its type carried out en-route simultaneously with the
location determination using the GPS receiver. At the next stage, the shape- les
consolidation has been performed according to the areas where measurements
have been taken. These operations have been carried out on the geographical
information system (GIS) ArcGIS [
The labels limiting the elds of the uniform type of vegetation in ArcGIS help to draw corresponding polygons. According to these polygons, parts from the space images are cut for the statistical and spectral analysis of their characteristics. The example of polygons for the Shartash Forest Park with the attribute information is shown in Fig. 3. Clicking any site, the line with the relevant information is highlighted in the table; for example, the birch growth in the selected blue area in Fig. 3.
Correspondingly, the information for any number of the standard areas is formed and entered into the table. This allows creating the directory that contains objects' characteristics that are necessary for studying the properties of their images.
It is necessary to have the appropriate database (DB) for the analysis of e ectiveness of the vegetation classi cation algorithms. The speci ed database includes the images in vector form for seven standard areas in Yekaterinburg and Sverdlovsk region. It also contains the forest formula and other characteristics of vegetation for the areas as the attribute information. The Yekaterinburg region includes the following park zones: Stone Tents (ST), Shartash Forest Park (Sh), Russian Foresters Park (RF), Mayakovsky Park (M), and the Southwest Forest Park (SF). Sverdlovsk region includes the following zones: the forest park Pripyshminskie Bory around Talitsa (T) and vegetation areas near the Revun Rift on the Iset River (R). The vegetation sites in the natural park Deer Streams have been excluded from the consideration because of their small area.
According to this data, the directory containing a su ciently large number of standard areas with di erent species and age composition of the vegetation is created. This data is presented in Fig. 4.
As a result, the following vector les are formed with: pine in seven standard areas; larch in 2 areas; poplar in one area; birch in 3 areas; 5P5B (50% pine and 50% birch) in 3 areas; 7P3B in one area; 9P1B in 2 areas; and 5A5B (50% of aspen and birch 50%) in one area.
Using the known coordinates of the generated standard areas, a database is created. This database contains the catalog of satellite images from the standard areas and sub-satellite vegetation data. This information is obtained from measurements on the ground, from the Internet, public archives, and other sources. This database also includes information about the satellites and instruments used to obtain the constituent snapshots. The database updates the information, which is necessary for executing the vegetation classi cation algorithms by collecting data from the remote sensors.
The database is created using application of the software Microsoft Access 2007. The database consists of several joined tables.
Satellite: contains a description the satellite name, sensor, its spatial resolution, the number of spectral channels, swath, frequency of recording, the start date, as well as a link to the source images.
Snapshots: lists the snapshots and their attribute information. The attribute information includes the name of the snapshot, the date and time of recording, the satellite, coordinates of the upper left and lower right corners of the snapshot, the percentage of cloud cover, region, province or city, location in the image, and allocation of the snapshot. The catalog includes the images of Sverdlovsk region with the percentage cloud cover no more than 80%.
Objects: shows the results of the sub-satellite measurements such as execution date, forest formula, description of the site and its size in pixels, and the name of territory where the measurements are made.
Territories: contains complete information about the natural object, including the location, category, and characteristics.
Vector Maps: includes complete details of the vector maps.
Raster Maps: includes complete details of the raster maps.
Linker: links the tables Snapshots and Raster Maps.
By using this directory, it is possible to perform the following tasks: - selection of images from a predetermined portion; - selection of images from a predetermined time interval; - selection of images from the selected satellite on a certain date; - selection of images, where there is one or more targeted objects.
The developed database simpli es the selection of targeted snapshots, that speeds up the process of their treatment.
Satellite table is shown in Fig. 5, and Snapshots table is shown in Fig. 6. Objects table is shown in Fig. 7, and Territories table is shown in Fig. 8.
Next Vector Maps (Fig. 9) and Raster Maps (Fig. 10) tables are presented. They are intended to view maps where the studied object is located.
The link between the Snapshots and Territories tables is many-to-many. Therefore, this link is carried out it through additional Linker table.
Three requests are performed in our directory. The rst one is Vegetation Type and Subtype Selection (Fig. 11). It forms tables of selected subtype of vegetation (Fig. 12 and Fig. 13).
The request Selection of Snapshots by Date and by Satellite is required to enter the start and end dates, as well as the type of the satellite. The placement of the snapshot and the list of territory objects displayed on it are shown in request results (Fig. 14).
The third request Territory Snapshots for Certain Period Selection is required to enter a date range and territory name (Fig. 15).
The sequence for entering new information depends on the type of information being added. In order to enter new sub-satellite data, it is rst necessary to check whether the Territories table already has the area where the measurements are taken and, if not, add it. After that, the parameters of the objects and their location are entered into the Objects table. Finally, users can link a new territory to the Snapshots table or can link the snapshots to a speci c site in the Territories table. If the territory where the measurements were made already exists in the database, the users only need to add the new sub-satellite information for it.
Upon receipt of the new images, it is rst necessary to check the existence of the satellite in the Satellites table and, if it does not exist, to add it. Then it is necessary to add the new images into the Snapshots table. If cloud cover is more than 80%, the addition of these new images is blocked, and the user will receive a warning message. 4
The article presents the results of the development a imagery database for
natural objects monitoring in Microsoft Access 2007. The database contains a catalog
of images from the satellites Terra, Aqua [