Detecting and monitoring plumes caused by major industrial accidents with JPLUME, a new software tool for low-resolution image analysis
Introduction
The total number of major industrial accidents reported each year in the EU from 1985 to 1999 shows a steady increase, with the maximum number reported during 1998. This may be due to a number of factors, including increased industrial and other economic activity and increased population densities around potentially hazardous sites, only partly compensated for by increased awareness and safety measures (EEA, 2003). An improved approach to safety and environmental management has been adopted, following the advent of the SEVESO II directive.
The European Council Directive 96/82/EC of 9 December 1996 concerning the control of major-accident hazards involving dangerous substances (SEVESO-II) aims at preventing such accidents and limiting their consequences for man and the environment, with a view to ensuring high levels of protection throughout the Community. This Directive has replaced Directive 82/501/CEE (SEVESO I). It is first to introduce substances considered dangerous for the (aquatic) environment in its scope. Industrial accidents can be connected to a number of different events and processes, including spillage, sudden release of materials, fire, or explosion. The most common effect is the release of gases and liquids used and processed in the installations concerned. Fire and explosion are common effects as well, while a combination of the above is not rare. Releases to land, water or the air may be toxic. Airborne releases usually develop in plumes, which can thereafter be monitored either due to their optical depth or their temperature difference from the ambient air. More often, accidental discharges imply heavier than air gases characterized by a slumping phase giving high and toxic concentrations (Dandrieux et al., 2003). Damages may thus occur both as an immediate and direct consequence of the accident, as well as during the propagation and dispersion of the resulting plume. It should be mentioned that while the immediate ground-level effects in close vicinity to the installation have been examined in detail, limited attention is usually paid to the plume's impacts in the wider geographic region affected during the hours following the accident.
In recent years, and after a number of incidents involving fires in industrial installations and warehouses, research has been oriented towards the definition of the properties and of the amount of the plume particulates generated by different materials, including pesticides, under various fire conditions (Lang, 1993, Bartelds et al., 1993, Atkinson and Jagger, 1994, Miles and Cox, 1994, Grant and Drysdale, 1994, Martins and Borrego, 1994, Marliere, 1996, Porter et al., 1996, Martins et al., 1996, Cozzani et al., 1996).
In several studies, (Kaufman et al., 1990, Ferrare et al., 1990, Kaufman et al., 1992, Cahoon et al., 1994) satellite data have been used for the analysis of the emitted smoke in order to quantify the gaseous output from forest fires. Ackerman and Toon, 1981, Kaufman, 1987 and Fraser et al. (1984), have related the carbon content of the plume to the single scattering albedo, as well as to the light extinction of the plume. Chung and Le (1984) have examined the feasibility of using satellite imagery to detect large-scale pollution episodes. Christopher and Chou (1997) have used a combination of spectral and textural measures in order to visually separate the plume aerosols from the underlying background. Baum and Trepte (1999) proposed a grouped threshold method for scene identification in NOAA/AVHRR (Advanced Very High Resolution Radiometer) imagery that may contain clouds, fire, smoke plumes or snow. Chrysoulakis and Cartalis (2003a) have proposed a software tool for the detection of major fires caused by technological accidents with the use of AVHRR imagery.
There is a specific philosophy in the use of satellite, since they constitute a trustworthy means to obtain information that does not depend upon the local structure. It should be stated that the GMES (Global Monitoring for Environment and Security), which is a joint initiative of the European Commission and the European Space Agency, has currently taken a very important initiative for the identification and standardisation of satellite data services that can be used for environmental management in general, and for risk management in particular (GMES, 2003).
In this paper a new software tool for the detection and monitoring of plumes caused by major industrial accidents is described. The software is named JPLUME and is coded in JAVA2 language using the JDK (Java Development Toolkit) 1.2.2 (Sun, 1999). JPLUME has been evaluated for four past accidents: in Enschede, the Netherlands (May 13, 2003) in Genoa, Italy (April 13, 1991); in Lyon, France (June 2, 1987) and in Kalohori, Greece (February 24, 1986). In this study, an AVHRR image acquired over the broader area of the Netherlands on May 13, 2000 (14.44 UTC) is used to present the functionality of the software tool proposed. This date refers to a massive explosion in a firework factory in the town of Enschede.
Several models and decision taking help tools exist (Quaranta et al., 2002, Bellasio and Bianconi, 2005) to assess the consequences of the different possible accidents. Software packages aiming at industrial risk assessment have already been developed; some of these software packages are WHZAN (Technica, 1992), RISKIT (VVT, 1993), EFFECTS (TNO, 1991), SAVE (TNO, 1992), MAXCRED (Khan and Abbasi, 1999) and OSIRIS (Tixier et al., 2002). JPLUME differs from the existing software packages in the following areas:
- (a)
it is a detection rather than a forecasting tool;
- (b)
it is not limited to the specific location of an industrial complex, but it is scalable, it may be applied for the monitoring of areas covering a single industrial installation, as well as for extended areas;
- (c)
as a JAVA developed programme it is independent of the hardware and the platform, a fact which provides the software with interoperability; additionally, users can easily modify the source code of the programme using the JDK 1.2.2 which is freeware (Sun, 1999);
- (d)
JPLUME offers a window-based user interface and is user friendly. It carries menus, buttons and a flow chart of the software's algorithm as well as an Image Viewer which gives the user the opportunity to follow the algorithm's various steps.
Section snippets
Data and methodology
JPLUME uses AVHRR images as inputs. AVHRR has a spatial resolution of 1.1 km at the nadir, a swath coverage of 2700 km and a revisiting time of approximately 3 h (synergy of NOAA 12, 15, 16, and 17 polar orbiting satellites). AVHRR records incoming radiation in five spectral channels (μm): 0.58–0.68 (visible), 0.72–1.10 (near infrared), 3.55–3.93 (mid-infrared), 10.5–11.3 (thermal infrared) and 11.5–12.5 (thermal infrared).
An analysis of the physical background of the procedure used to
Design and application of JPLUME software
In this section JPLUME is described in detail with regard to its design and use. Fig. 1 demonstrates the software user interface, which presents the options available in JPLUME. A flow chart of the main stages of the algorithm appears on the right part of this window. The series of buttons appearing on the left part of the main window activates the modules which implement the various parts of the JPLUME algorithm. Each module includes one or more classes of the object-oriented code. Each class
Conclusions
Using satellite data to detect a plume caused by major industrial accidents relates to the plume's characteristics (dispersion, optical thickness and temperature structure) and to the sensor's spatial resolution as well. A new low spatial resolution satellite image analysis software JPLUME, has been developed, as a comprehensive and user-friendly tool for the automatic detection and monitoring of plumes caused by major industrial accidents. JPLUME methodology is based on the development of a
References (37)
- et al.
On line simulation system for industrial accidents
Environmental Modelling & Software
(2005) - et al.
Detection of forest-fire smoke plumes by satellite imagery
Atmospheric Environment
(1984) - et al.
The DVS model: a new concept for heavy gas dispersion by water curtain
Environmental Modelling & Software
(2003) - et al.
Satellite measurements of aerosol mass and transport
Atmospheric Environment
(1984) - et al.
A decision support system for the simulation of industrial accidents
Environmental Modelling & Software
(2002) - et al.
OSIRIS: software for the consequence evaluation of transportation of dangerous goods accidents
Environmental Modelling & Software
(2002) - et al.
Absorption of visible radiation in atmosphere containing mixtures of absorbing and non-absorbing particles
Applied Optics
(1981) - et al.
Plume seeding mechanisms in warehouse fires
- et al.
Emissions from medium scale fires of chemicals
- et al.
A Grouped threshold approach for scene identification in AVHRR imagery
Journal of Atmospheric and Oceanic Technology
(1999)
Satellite analysis of the severe 1987 forest fires in northern China and southeastern Siberia
Journal of Geophysical Research
The potential for collocated AGLP and ERBE data for fire, smoke, and radiation budget studies
International Journal of Remote Sensing
Hazardous substances formed as a consequence of industrial fires
TAD – a new satellite images analysis software tool for the detection of major fires caused by technological accidents
International Journal of Remote Sensing
A new algorithm for the detection of plumes caused by industrial accidents, on the basis of NOAA/AVHRR imagery
International Journal of Remote Sensing
A new approach for the detection of industrial accidents, using NOAA/AVHRR imagery
International Journal of Remote Sensing
Europe's environment: the third assessment
Environmental Assessment Report No 10
Cited by (19)
Detecting and characterizing upwelling filaments in a numerical ocean model
2019, Computers and GeosciencesCitation Excerpt :Ocean circulation patterns and their impact on marine ecosystems is a topic of interest to scientists (Bakun, 1996; Condie et al., 2005) that presents a technological challenge. This includes detecting features like eddies (Chaigneau et al., 2008), plumes (Chrysoulakis et al., 2005), and fronts and filaments (Cayula and Cornillon, 1992; Nieto et al., 2012) in results from remote sensors and numerical circulation models. These sources of information are often studied with the aid of image detection algorithms (Wang et al., 2010) or geoprocessing tools (Roberts et al., 2010).
Current situation and needs in man-made and natech risks management using Earth Observation techniques
2015, Remote Sensing Applications: Society and EnvironmentCitation Excerpt :Image processing realized with MODIS and MERIS (Medium Resolution Imaging Spectrometer) techniques are also used to detect and monitor smoke plumes and, due to their spatial and temporal resolution, have become important sources of data (Chu et al., 1998; Kaufman et al. 2003; Kaufman and Tanre, 1998; King et al., 1999, Stowe et al., 1997). Moreover air emissions that occur as "plumes" can be monitored by their temperature difference with the surrounding air (Chrysoulakis et al., 2005) and often emissions coming from an industrial accident are composed by gases heavier than air that could lead to high concentrations of toxic substances at ground level. If the event has been caused by a structural failure in the facility the quantity of toxic substances released and their concentration will be extremely abundant and easily visible in remote airborne images (Dandrieux et al., 2003).
An improved algorithm for the detection of plumes caused by natural or technological hazards using AVHRR imagery
2007, Remote Sensing of EnvironmentCitation Excerpt :However, while the detection of natural or technological fires using satellite images has been given much attention and is now performed on a routine basis (Cahoon et al., 1994; Christopher et al., 1996; Chrysoulakis & Cartalis, 2000, 2003b; Huang & Siegert, 2005; Kaufman et al., 1990; Prins et al., 1998; Pu et al., 2004), fewer interest has been given to automatic plumes detection. Methods have been developed for plume detection based either on image photointerpretation using pseudo-coloured compositions of Advance Very High Resolution Radiometer (AVHRR) spectral channels (Chung & Le, 1984; Kaufman et al., 1990; Randriambelo et al., 1998), or on retrieving of optical depth of heavy aerosol plumes (Wong & Li, 2002), or on multi-threshold methods based on spectral indices (Baum & Trepte, 1999; Chrysoulakis & Cartalis, 2003a; Chrysoulakis et al., 2005) and textural features (Christopher & Chou, 1997; Christopher et al., 1996), or on multi-threshold methods based on neural networks (Li et al., 2001). Recently, MODIS (Moderate Resolution Imaging Spectroradiometer) and MERIS (Medium Resolution Imaging Spectrometer) imagery have become important sources of data of appropriate spatial and temporal resolution to be used for plume detection and monitoring (Chu et al., 1998; Kaufman et al., 2003; Kaufman & Tanre, 1998; King et al., 1999; Stowe et al., 1997).
Quantifying effects in two-sample environmental experiments using bootstrap confidence intervals
2007, Environmental Modelling and SoftwareCitation Excerpt :Finally, also real-time monitoring systems could benefit from incorporating a “bootstrap tool” into their data analysis. For example, Chrysoulakis et al. (2005) developed a software for low-resolution image analysis to detect the occurrence of major industrial accidents using satellite imagery data. In this case, the control sample would come from the data before the accident and the treatment sample would come from thereafter.
An analysis of the operation mechanism of chemical industry park ecosystem based on theory of ecological organization
2021, Energy Engineering: Journal of the Association of Energy Engineering