RICERCA ITALIANA     

Research program

The geomatics in support of the actions of Government of the territory

University Co-ordinator

Università degli Studi di BOLOGNA - Centro di Eccellenza per i Sistemi Elettronici e delle Telecomunicazioni - ARCES - ()

Research Unit Leader

Maurizio Barbarella

Description

The BOLOGNA Research Unit intends to address the problems tied to the monitoring of land areas that are of interest to the public authorities, with a focus on two specific sectors:
1-precision monitoring of buildings or structures prone to movement (for example due to landslides)
2-monitoring of the territory by remote sensing in order to gain knowledge of and govern the evolution, over time, of particular features such as crops, vegetation or other aspects through a multi-temporal approach.
Applications in both thematic areas will essentially be concentrated in Emilia - Romagna .
1) Precision monitoring of structures
The Research Team intends to set up an initially independent test polygon on which to conduct a number of trials involving the use of satellite positioning methods to monitor buildings/structures in landslide areas. The polygon will subsequently be connected to a network of permanent GNSS stations for real-time positioning; if possible this will be achieved using transmission apparatus and sites of mobile radio networks.
The polygon could be made up of a variable number of buildings to be monitored, with movements being determined in relation to a position outside the landslide area assumed to be stable. The polygon must be materialised so as to ensure that the GNSS antennas are integral with the structure to be monitored and for each site a suitable location must be found for the electronic equipment that will collect the data and transmit it to a local control and data storage centre. It would be preferable (albeit not indispensable for an independent polygon) for the control centre to be connected to the Internet or dedicated data transmission networks in order to enable real-time monitoring of the system and uploading of data to a remote site, where a Control Centre could be set up to gather information from different parts of the monitored territory.
Different types of GNSS antennas can be installed on the polygon, starting from low-cost units that work with a single carrier frequency (L1) and then moving on to more sophisticated antennas designed to operate at dual frequencies (L1, L2) and with more than one satellite constellation (GPS+GLONASS). This will make it possible to assess the added value provided, in such a context, by more complete instrumentation, which is both costlier to purchase and entails higher data processing costs.
Another aspect to be considered is the possible synergies with infrastructures that already exist or are under construction. Part of the trial will thus explore the feasibility of integrating the different sensors within a more complex structure, i.e. a network of GNSS stations for real-time precision positioning. One of the sensors (specifically, the one assumed not to be susceptible to landslide movements) could be used as an aid in computing differential correction models for precision positioning and also serve as a reference station for the other sensors located in buildings to be monitored. The platform the research team intends to rely on for this purpose is a network of permanent GNSS stations for experimental real-time positioning (set up, georeferenced and managed by the group itself). If the network is also to be used for monitoring purposes it should be able to guarantee continuous operation, but several trials conducted in recent years have shown that commercial communication channels cannot always assure stability in the transfer of data; as a result malfunctions may occur in the services delivered in real time, even if only for some periods. Where a monitoring capability is required, a continuous flow of data is to be considered indispensable. Therefore, a possible alternative would be to use dedicated channels, including institutional ones, which can fully satisfy this requirement. At present Emilia Romagna Region is building a digital mobile radio network called R3, based on the European standard (TETRA); it is technologically advanced (similar in part to GSM/GPRS at 900 MHz) but specifically conceived for emergency purposes. Since this project is already underway, the research team believes it is strategically worthwhile to experiment with using the same network for the establishment of Permanent GNSS Stations. One member of the research team, a regional official, will guarantee a direct linkup to the R3 structure.
From an operational perspective, a number of test sites will have to be selected in order to perform a feasibility analysis to determine whether the structures can satisfy both geodetic requirements (good stability, wide sky view assuring adequate satellite visibility and absence of electromagnetic waves around the GNSS signal operating frequencies) and technical ones (problems tied to installing the antennas on already existing structures not designed to accommodate sensors of this type, problems tied to the uploading of data to a control centre).
The research activity will be divided into 5 distinct phases:
Phase 1 - estimated duration 4 months – STATE OF THE ART
During this phase the research team plans to carry out a thorough literature review in order to gain insight into the current national and international state of the art, specifically as regards the available instrumentation and methods applicable to this case study. It will also gather information about the technical characteristics of the R3 mobile radio network and the location of the sites, and will moreover select several pilot sites for the implementation of the GNSS antennas.
Phase 2 – estimated duration 4 months – PREPARATION OF FIELD TRIALS
In the second phase sites having appropriate characteristics will be selected to form the test polygon. Suitable GNSS antenna mount adaptors will be devised and requests will be submitted to obtain the necessary authorisations for the research activity.
As regards the implementation of permanent GNSS stations in the R3 network, if the geodetic and technical characteristics of the sites are deemed satisfactory, pilot stations will be selected, suitable GNSS antenna mount adaptors will be built and steps will be taken to create appropriate experimental conditions.
Phase 3 – estimated duration 4 months – EXECUTION OF THE TRIALS
This is the operational phase of the project, during which the trials will be carried out.
Low-cost GPS stations will be installed on the duly set-up test polygon for an initial period (approx. 3 months of data acquisition), whereas dual-frequency geodetic antennas will be applied in the second period (approx. 3 months of data acquisition). During the latter period, involving the use of dual-frequency sensors, tests will also be performed with a view to integrating the monitoring apparatus with the control centre for the delivery of real-time positioning services. In this context the data acquired from the reference station will also be used to compute the spatial correction parameters for real-time precision positioning.
With respect to the implementation of the network of permanent stations via the R3 system, a number of antenna units (3 at most) will be mounted on antennas of the mobile radio network. Wired connections will be set up to allow real-time uploading of data to the control centre, which at this stage will most likely be a computer located in the DISTART Department.
Phase 4 – estimated duration 6 months –DATA PROCESSING AND ANALYSIS
Procedures for processing and analysing the data collected in phase 3 will be defined and implemented. Suitable data treatment and filtering procedures must be formulated, along with methods for detecting bias factors that could mask the actual movements of the monitored structures. A cost-benefit analysis will be conducted to determine the real added value of using more sophisticated equipment. Moreover, an assessment will be made of the performance and efficiency of the system, which, by recording data in real time, produces two types of outputs: spatial correction parameters for real-time positioning, and real-time monitoring of sensors.
Phase 5 – estimated duration 6 months – DOCUMENTATION OF RESULTS
Documentation of the results achieved. Contribution to the Final Report
At the end of the study several months will have to be dedicated to documenting the final results achieved and the methodology adopted.
It will be necessary to organise a presentation of the results of the trials conducted by the Research Team, both before the scientific community and the institutions involved and within the professional sector. Activities in this phase will be carried out in close coordination with the other project partners.

2) Remote sensing
The first step in a research project is to define the context in which the experimental phase will take place and the stakeholders involved. This means identifying, to begin with, the potential users (Municipal Government/Park/Utility Company …) of the system and defining their needs so as to narrow down the field of application of the research (urban environment/natural environment/agricultural area…). For each area of application identified and each stakeholder involved the existing geographic/spatial/qualitative data will be catalogued and an assessment will be made as to whether they can be reused within the system, the aim being to have data that are as standardised as possible and accompanied with an appropriate set of metadata. Attention will be focused on creating a Geographic Information System for crop monitoring and management by integrating data in varying formats and from different sources, including satellite images at different geometric and spectral resolutions, such as those recorded by the QuickBird and Aster sensors.
The design of a database will entail preparatory phases initially centring on the reorganisation of data and, if necessary, their conversion into a digital format, along with georeferencing of the data themselves in a single cartographic reference system and digitisation of QuickBird images for the purpose of defining the individual elements (plots of land).
Database structuring will be followed by the creation of automatic procedures for updating the information content based on the analysis of satellite images.
It is planned to examine the present state of the art as regards techniques for supervised classification of crop pixels applied to multispectral images: the availability of different bands, such as the 14 of ASTER data, which fall between the visible and thermal infrared wavelengths, permits more detailed observations of the earth’s surface and recognition of a large part of the classes making it up. The main difficulty in this phase will be the determination of sample areas – which are necessary both for computing the classification algorithms and verifying the reliability of the results – given the frequent lack, as in the case of herbaceous crops, of up-to-date official records from which to extract useful information. As a consequence, it will be necessary to apply to specific public and private local bodies in possession of some of the required data, which represent a fundamental component of the database. Based on the sample areas it will then be possible to classify the entire area covered by the image.
Given the different variables in play which influence the spectral signatures and hence classifications (such as the complexity of the reality on the ground divided into numerous classes, the seasonal nature of crops, atmospheric factors, etc.), a further contribution may derive from the comparison, and possibly integration, with object-oriented segmentation and classification techniques, which better define the elements making up the territory. There exist rigorous statistical methods for comparing the accuracy of thematic maps derived from different classification algorithms. One of the planned objectives is to experiment with some of these tests, such as the one based on the K coefficient or McNemar’s test, which may be applied to the case in question in order to determine both the algorithm and type of procedure that furnish the best results and thus best guarantee the functionality of the database within the broad context of land monitoring.
Furthermore, the multitemporal character of satellite data, whose frequency of acquisition depends on the characteristics of the respective sensors, offers the possibility of monitoring the evolution of both natural and urban environments and the phenomena that cause them to undergo change.
Another equally significant aspect is the possibility of integrating the result obtained from the analysis of multispectral images with high-resolution images so as to create a multi-scale system that would allow a land analysis to be performed at a different level of detail.
Previous experience in this area has involved the use of commercial software (ArcGis, Envi, Idrisi, eCognition), as such applications are able to satisfy the majority of requirements. Considering that their potentialities often extend beyond the basic functions needed and their high costs, which are not always sustainable, analogous, albeit less sophisticated, open source software could be used in their place to process images and manage all the other data in general.
The procedures defined in the previous paragraph will serve as the basis for the design of a prototype topographical database which will similarly be developed using open source tools. In this context, issues related to the standardisation of data, metadata and formats will also be addressed with the aim of assuring perfect interoperability with proprietary software systems.

High-resolution satellites can also provide useful information for assessing and analysing calamitous events of varying nature.
The research team aims to develop a data management and analysis protocol that uses the data obtained within the field of view of high spatial resolution satellites, combined if necessary with SAR data and on-site surveys, for land monitoring purposes.
One of the most important phases of the research will involve the study of imaging scan models of pushbroom-type systems, where the objective will be to develop algorithms for georeferencing and orthorectification of remote sensing data. It shall be stressed that this aspect of the research is closely linked to the work of other research groups that are concerned with the materialisation and standardisation of different cartographic reference systems.
Procedures will also be devised for the retrieval of information regarding the degree of damage of manmade structures and natural areas; this will involve a simultaneous analysis of scenes of different epochs taken from sensors having different radiometric and spatial resolutions, according to a multiscale and multitemporal approach. For this purpose both the remote sensing images from X and L/S band satellite antennas of ReSLEHM - Remote Sensing Laboratory for Environmental Hazard Monitoring (Remote Sensing Centre of the University of Salerno) and high and medium resolution images (QuickBird, Ikonos, Spot5, etc.) will be used.