Sensor Web Enablement of environmental monitoring and process control
By Gavin Fleming, Senior
GISc and Sustainable
Development Researcher:
Mintek
Keywords: [acid mine
drainage, EIA, EMP,
environmental impact
assessment, environmental
monitoring, GEOSS,
geospatial, GRI,
instrumentation,
interoperability, Krugersdorp
Basin, mashup, OGC,
pollution, process control,
Python, Sensors Anywhere,
Sensor Web Enablement,
SOS, standards, SWE,
TransducerML]
Adding geospatial information to measurement data provided by
instrumentation provides new and valuable insights for environmental
monitoring. Could this become pervasive in process control?
Abstract
This paper explores the potential of Sensor Web Enablement for environmental
monitoring and process control, illustrated by a mine environmental monitoring prototype
in the Krugersdorp Basin, a mineral processing process control concept and a mining
operations management concept.
The task of measuring and interpreting environmental and process variables is becoming
increasingly large and complex. Many proprietary and open technologies are in use for
observing, communicating, analysing and reporting these variables.
Several international initiatives have created the conditions for interoperability among
these systems at technical, organisation, semantic and political levels. Examples include
the Global Earth Observation System of Systems, the Open Geospatial Consortium’s
(OGC) Sensor Web Enablement (SWE) initiative and the Semantic Web.
Sensor Web Enablement allows for the integration and analysis of streams of sensor data
from multiple and diverse sensors in a standards-based and thus interoperable manner.
For instance, observations from water quality sensors can be fused with those from
weather instruments and satellite remote sensing instruments. Measurements of anything
from process or biophysical variables to higher level indicators such as re-vegetation or
landscape function and even social impact potentially can be Sensor Web Enabled.
Collection, management and analysis of these data can be automated and adaptive,
handling the disruption of service from some sensors or the addition of new sensors.
Sensor Web Enablement of mines and their entire footprints ultimately can enable greater
understanding of the systems at play by capturing their dynamics in time and space.
In the Krugersdorp Basin, disused gold mines, whose groundwater is no longer being
pumped out, are overflowing at several egress points. Acid Mine Drainage is
contaminating surface and groundwater over a large area. Water quality and quantity
sensors have been placed in six clusters throughout the system in an attempt to
characterise it, monitor compliance and to flag events that need urgent attention, such as
flow pulses or dips in pH.
In a Sensor Web Enablement prototype, these sensors have been exposed through an
OGC SOS (Sensor Observation Service), described by OGC SensorML (Sensor Model
Language) and visualised in a SOS graphing client.
1 Introduction Modern mine monitoring requirements are discussed and challenges in achieving these
are raised. Increasingly complex, onerous and costly monitoring requirements necessitate
a new approach to monitoring. Technically, organisationally and semantically isolated
monitoring systems dominate in the industry. These are generally provided by competing
proprietary vendors.
An emerging concept called Sensor Web Enablement (SWE) is introduced and described.
SWE results in interoperable monitoring systems with less redundancy and wider
application. The application of SWE in mine environmental monitoring is explored,
focussing on aspects of closure and rehabilitation. An example of SWE of a water
monitoring system in a closure situation near Krugersdorp is presented to demonstrate
some SWE concepts.
2 Monitoring challenges and solutions Mine environmental monitoring is increasing in importance and complexity. Many
governments are imposing more stringent regulations and compliance with environmental
management programmes. Mining companies monitor more for improved risk assessment
and management; they are showing greater responsibility to mine-affected communities
and are responding to shareholder activism. Comprehensive corporate reporting, such as
GRI (Global Reporting Initiative), requires extensive monitoring.
South Africa’s Mineral and Petroleum Resources Development Act (MPRDA) of 2002
specifies objectives for the monitoring and assessment of Environmental Management
Programmes (EMPs). These are to: Detect short and long term trends; recognise
environmental changes and analyse causes; measure impacts and compare with predicted
impacts; and improve monitoring practices and procedures for environmental protection
(MPRDA).An EIA (Environmental Impact Assessment) defines levels of acceptable
change of various environmental variables. These variables need to be monitored so that
the levels of change are not exceeded. The MPRDA specifies that a monitoring
programme must ‘identify trends, causes and impacts’ and ‘assess performance and
compliance’. Furthermore, Regulation 55 of the MPRDA specifies that monitoring by
mine permit and rights holders must be continuous until closure is granted, that post–
closure risk indicators must be monitored, and that measurements must be stored and
reported to National Department of Minerals and Energy (DME).
New water regulations in South Africa require that the polluter pays according to a
‘waste-discharge charge system’ (WDCS) (Moodley, 2006). A good observation
programme and associated alerts and prediction models will reduce financial risk for
mines.
Mine tailings dams around Johannesburg are a major source of radon, dust and water
pollution. Monitoring for compliance is required on a large scale. It is also required
during reprocessing and moving of the dumps to new superdumps (Anon, 2002).
While it essential to monitor during operations and closure, it is ideal to begin an
observation programme well before mining or even exploration start. The MPRDA
requires that mines be rehabilitated to their natural state. A proper observation
programme can start by helping to define what that natural state is and what indicators
need to be measured to monitor progress back to that state.
These monitoring requirements pose several challenges to the mining industry. What is
the optimum approach to measuring so many variables accurately and continuously?
How can the resulting observation data be stored, managed and distributed? How can
data from multiple different sources be integrated for scientific research and summarised
for communication?
The Sensor Web is an emerging concept that promises to revolutionise measurement and
observation in the mining industry and elsewhere. Sensor Web Enablement, or SWE, of
mine environmental monitoring, while in its early days, has the potential to solve many of
the challenges posed above. I present a vision of a Sensor Web Enabled mining industry
with an example of an early application.
3 The monitoring status quo Figure 1 shows typical components of a monitoring system. Instruments, computers,
people and communications technologies are involved throughout the system. There are
many potential points of failure, inefficiency and unnecessary cost.
There are several dimensions to the complexity of monitoring. One is the range of
phenomena that are measured. There often are separate monitoring systems for each
phenomenon. Another dimension is the users or systems that ‘consume’ the observations.
The environmental reporting department and the risk management department might have
to measure the same phenomena, but because their requirements are different they
maintain separate monitoring systems, causing unnecessary redundancy. Multiple
vendors and lack of standardisation (multiple proprietary data formats for example) are
other important factors. Figure 2 illustrates some aspects of the status quo. There is often
little if any communication among domains. Custom after-market integration software is
usually expensive and problematic.
4 Sensor Web Enablement Sensor Web Enablement (OGC SWE 2007) is an OGC (Open Geospatial Consortium)
initiative that has seen several standards and specifications emerge for describing sensors,
encoding observation data and making these data available via standard web service
interfaces. SWE will facilitate the discovery, exchange and processing of sensor
observations. SWE promises to make a multitude of sensors and their observations
available on the Web. Together with OGC Web Processing Services (WPS) SWE will
exploit distributed computing to fuse and integrate these data through real-time service-
chain composition to generate up-to-date, dynamic and accurate information products
that have been difficult, costly or impossible to access before. Any sensor can potentially
participate in the sensor web, from an in situ borehole water quality sensor to a satellite
remote sensor to a human visual observation or questionnaire.
Figure 1. The monitoring stack.
Each observation in the Sensor Web is tagged with time and location, allowing for
realistic spatio-temporal modelling and a richer understanding of environmental
dynamics.
Extending the scope of OGC SWE, the Semantic Web (W3C 2001) will enable users to
define queries that will automatically be interpreted; automatically locate, interrogated or
task sensors on the Internet, acquire and process observations and return answers.
SWE uses a subset OGC Web Services (OWS), a Service Oriented Architecture (SOA)
that can operate in any service paradigm, from 'SOAP' to 'REST'.
SWE is an essential component of GEOSS, the Global Earth Observation System of
Systems (GEO 2007). Many governments, including that of South Africa, have agreed to
the GEOSS 10-year Implementation Plan, which aims to coordinate earth observation
activities globally, thereby reducing costs, increasing accuracy and resulting in greater
benefit to society. SWE forms part of the architecture needed to bring this vision about.
Figure 2. Mine monitoring silos.
Sensor Web Enablement currently involves some or all of the following components, all
operating in the framework of the OGC service-oriented reference architecture:
• O&M (Observation and Measurements): XML schema for encoding sensor
observations and measurements (Figure 3).
• SOS (Sensor Observation Service): Web service interface for interrogating
sensors and retrieving observations and sensor descriptions.
• SensorML (Sensor Model Language): XML schema for describing sensors and
processes they can participate in.
• SAS (Sensor Alert Service): Web service interface for subscribing to sensor-based
alerts.
• SPS (Sensor Planning Service): Web service interface for tasking sensor
activities.
• TransducerML: XML schema for describing transducers. Supports real-time low-
level data streaming and actuator control.
• WNS (Web notification service): Web service interface for handling
communication of messages from SAS or SPS.
Figure 3. Example of OGC observations and measurements XML.
Every component within the scope of a Sensor Web intervention is standards-based.
Therefore, perhaps the most beneficial outcome of SWE is interoperability.
In SWE, “any Observation is an Event whose Result is an estimate of the value of some
Property of a Feature of Interest, obtained using a specified Procedure”. The capitalised
terms in the preceding sentence are keywords in Observations and Measurements XML
documents, and imply that observations thus encoded are specified precisely and in the
same way, no matter what is being measured.
5 How does SWE impact on mine monitoring? Figure 4 shows mine monitoring systems integrated through SWE. Only the sensors that
are necessary to describe a phenomenon of interest, need to be deployed. SWE enables
observations from these sensors to be processed and communicated in such a way that all
<om:Observation gml:id="obsTest1"
xmlns:om="http://www.opengis.net/om"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xmlns:xlink="http://www.w3.org/1999/xlink"
xmlns:gco="http://www.isotc211.org/2005/gco"
xmlns:gmd="http://www.isotc211.org/2005/gmd"
xmlns:gml="http://www.opengis.net/gml"
xsi:schemaLocation="http://www.opengis.net/om ../om.xsd">
<gml:description>Observation test instance</gml:description>
<gml:name>Observation test 1</gml:name>
<om:time>
<gml:TimeInstant gml:id="ot1t">
<gml:timePosition>2005-01-11T16:22:25.00</gml:timePosition>
</gml:TimeInstant>
</om:time>
<om:location>
<gmd:EX_GeographicDescription>
<gmd:geographicIdentifier>
<gmd:MD_Identifier>
<gmd:code>
<gco:CharacterString>Subiaco Markets</gco:CharacterString> </gmd:code>
</gmd:MD_Identifier>
</gmd:geographicIdentifier>
</gmd:EX_GeographicDescription>
</om:location>
<om:procedure xlink:href="urn:x-ogc:object:feature:Sensor:OGC:scales"/>
<om:observedProperty xlink:href="urn:x-ogc:def:phenomenon:OGC:mass"/>
<om:featureOfInterest xlink:href="http://some.interested.org/vegetables/instances/banana1"/>
<om:result xsi:type="gml:MeasureType" uom="urn:x-ogc:def:uom:OGC:kg">0.28</om:result>
</om:Observation>
users have access to them in the form they require. These users, or ‘Sensor web clients’
include: Disaster management systems; risk management systems; operational
monitoring; mine planning, modelling and forecasting; reporting to the public; reporting
to government; reporting to global corporate headquarters; and monitoring progress
towards rehabilitation and closure.
Figure 4. The effect of SWE on mine environmental monitoring.
6 Krugersdorp Basin example The decanting situation in the Krugersdorp Basin west of Johannesburg, South Africa, is
a typical future closure scenario for many other mines (Fleming, 2007). Water levels rose
in several old adjoining mines after pumping ceased along with mining operations.
During the life of mine the pumping rate was 60Ml/day, and in late 2007 it was 6-23
Ml/day (pers comm., Council for Geoscience, CGS). Large quantities of contaminated,
acidic water started decanting and are still doing so from several openings. This water is
contaminating surface and ground water over a large area. Some of it is being retained in
buffer dams and treated but much still escapes untreated. The CGS placed six groups of
sensors in shafts, dams, rivers and other locations in the basin, aimed at characterising the
problem so that a suitable solution could be found, as well as to flag potential emergency
situations, such as a large drop in pH or increase in flow. The monitoring system initially
consisted of sensors, data loggers, GSM communications via SMS to a central service
provider and manual downloading of data by CGS staff. Each component in the system
was proprietary and hence closed and inflexible.
6.1 SWE intervention
A data pre-processor was developed to acquire observations from the central service
provider in their proprietary text format at regular intervals, parse them and write them to
a database conforming to the SOS schema. A Python-based SOS implementation adapted
from Bill Howe's at www.oostethys.org exposes these observations via the OGC standard
SOS interface for any client to access. A simple SOS client (Figure 5) fetches data for a
defined period and graphs it. Figure 6 show time series from the Krugersdorp Basin SOS
for six variables over one day. A graphing client such as this is a simple application of the
Sensor Web. If a SOS is public, the world is free to build any application simple or
complex, to use its observations.
Figure 7 shows a more complex SOS client that plots fire observations spatially, in this
case along with a map of South Africa retrieved from an OGC WMS (Web Map Service).
It serves to indicate further application of SWE. To translate this example to a mine,
picture a mine plan showing environmental incidents that need attention during an
emergency.
Figure 5. Simple client for the Krugersdorp Basin SOS (Sensor Observation Service).
Mashup
In web development, a mashup is a web application
that combines data from more than one source into a
single integrated tool. The term Mashup implies easy,
fast integration, frequently done by access to open
API's and data sources to produce results data owners
had no idea could be produced. An example is the use
of cartographic data from Google Maps to add location
information to real-estate data, thereby creating a new
and distinct web service that was not originally
provided by either source.
Source: Wikipedia
Figure 6. Time series of five variables at five sensor pods over one month retrieved from the
Krugersdorp Basin SOS.
Imagine custom dashboards on the computer screens of managers and operational staff,
bringing together in meaningful and useful ways, observations from multiple and diverse
sensors around a mine, all using standard Web technology.
Figure 4. A SOS client showing fire observations from the AFIS SOS (Advanced Fire Information
System, CSIR).
7 Process control While environmental monitoring is the focus of this paper, measurements are made and
used routinely in process control and mining operations as well. The author is
undertaking research with the Cyanoprobe team at Mintek to investigate Sensor Web
Enablement of mineral processes that use cyanide. While SCADA and OPC systems are
routinely used in process control, there is scope for investigating Sensor Web Enablement
to improve interoperability and expand the application of process measurements. With
TransducerML, SWE could be used directly in the process control loop, as part of
cyanide input control and waste treatment control. But outside the normal realm of
control systems, SWE can make process measurements available for operational
warnings and alerts, reporting, compliance monitoring and general environmental
monitoring, without investing in additional, redundant systems.
8 Mining operations The CSIR is investigating the Sensor Web Enablement of mining operations, exposing
production, geology, safety and other routine observations and measurements through
OGC standard web interfaces.
9 Barriers to Sensor Web Enablement Several standards and specifications for SWE are mature and have been implemented.
Others are in early stages of development. Currently there is no operational example of
all the components of the Sensor Web working together. The OGC’s own Test Beds (e.g.
www.opengeospatial.org/projects/initiatives/ows-5) are cutting-edge environments where
implementations of SWE components are tested and demonstrated. The ‘real-world’
domain that has progressed furthest with SWE is the Global Ocean Observing System
(www.ioc-goos.org). Large international research projects like Sensors Anywhere
(SANY 2006) are making inroads into SWE technology and application.
10 Recommendations The Mining industry can learn from these pioneers and start by adopting mature
components of SWE while assisting with Research and Development of SWE for less
mature components, or where special development is required for the mining
environment.
Behavioural and organisational changes are required to adopt something new like SWE.
Thus, implementing the standards upon which SWE is based will take foresight in the
industry. Stakeholders in the monitoring industry need also to participate in the
development of or upgrading to standards-based software applications.
Acknowledgements
Members of the Sensor Web Alliance (www.sensorweb-alliance.org) have been
instrumental in developing many of these ideas and associated technologies. Thanks to
Joan van Genderingen and Bridget Fleming for the original illustrations.
References
• Fleming, G. (2007) Sensor web enablement in the mining industry. Internal report number 40088,
Mintek, South Africa.
• Fleming, G., N. King, M. Mugonda, S. Dlamini (2008). Human-environment interactions: Integrated
indicators of sustainable development and sensor web enablement. Internal report number 4892,
Mintek, South Africa.
• Moodley, N. (2006) New waste water levy to impact mining. Mining Weekly, 22 March.
• Anon (2002) The disappearing act of Gauteng’s golden dumps. Mining Weekly, 7 June.
• MPRDA (Minerals and petroleum resources development Act 28 of 2002). Government of South
Africa.
• SANY (2006) Sensors anywhere. EU IST FP6 integrated project: http://sany-ip.eu/
• GRI (2006) Global reporting initiative. Sustainability reporting guidelines, G3 Version.
www.globalreporting.org
• W3C (2001) World Wide Web Consortium. Semantic Web. www.w3.org/2001/sw
• OGC SWE (2007) Open Geospatial Consortium Sensor Web Enablement working group.
http://www.opengeospatial.org/projects/groups/sensorweb.
• OGC (2006) Observation and Measurements. Document reference OGC 05-087r3.
• GEO (2007) Group on earth observations. Global earth observation system of systems.
http://earthobservations.org/
For more information contact Gavin Fleming, Mintek, +27 (0)11 709 4668,
[email protected], www.mintek.co.za
1 About the author
Gavin is a Senior GeoInformation Science and Sustainable Development Researcher
at Mintek, in the Mineral Economics and Strategy Unit (MESU). He qualified from
Wits with an MSc in population genetics, after a BSc and Hons in genetics and
microbiology.
In 1996 Gavin switched to GIS, starting at GIMS. In 1999 he moved to the GIS lab in
Environmentek (now Natural Resources and the Environment) at the CSIR. Still at the
CSIR he spent a year at the Satellite Applications Centre and then a year at ‘ICT4EO’
(ICT for Earth Observation) at the Meraka Institute before moving to Mintek in 2006.
Gavin heads up Mintek’s GIS department which captures, manages and publishes
spatial information and products such as paper and online maps. He is a staunch
advocate of Free and Open Source Software (FOSS) as well as of open standards and
interoperability and applies these wherever possible. He led the organisation of ‘Free
and Open Source for Geospatial 2008’ (www.foss4g2008.org), an international
conference that is happening in Cape Town in spring.
A specific niche of GIS that overlaps with various other fields is ‘Sensor Web
Enablement’ (SWE), which is the subject of his MMP presentation. He is running a
research project to develop Sensor Web applications for mine environmental
monitoring. That topic overlaps with his other work interest which is sustainable
development through mining. In that arena he is working on projects investigating
water and mining and local economic development in mine-affected communities.