Research supported by the SCSI is showing how satellites can become a vital part of construction site monitoring. Part I of a two-part article.

Building information modelling (BIM) standards and construction best practice stress the need for timely,
accurate mapping updates to ensure efficient monitoring of progress on site. These updates are traditionally provided by established survey technologies such as total station, laser scanner or, more recently, photogrammetry, using imagery captured by remotely piloted airborne systems (RPAS, aka drones). These techniques provide high-accuracy surveys, but they require personnel or hardware on site (or in the sky overhead) and operations can be restricted in busy, isolated, hazardous or sensitive areas. Satellite remote sensing offers an alternative, rapid, noncontact approach that is exportable and repeatable for any site worldwide, and has been demonstrated recently in developing countries via the World Bank. In this SCSI-funded study, we developed and tested an automated approach using a Dublin coastal site as a test case, demonstrating the suitability of recent advances in satellite technology for a 3D and 4D BIM process in Ireland.

Satellite remote sensing offers an alternative approach that is exportable and repeatable for any site worldwide.


Satellite imagery – a potential solution

Earth observation (EO) satellites broadly fall into two classes; those that record the sun’s energy
reflecting from the earth’s surface, and those that transmit their own energy and measure what
comes back. These satellites can survey large strips during each orbit, with some capable of
surveying areas the size of Leinster in just a few minutes. Other satellites, such as the
commercially available Pléiades constellation, focus on narrower strips but can therefore provide
imagery with a spatial resolution as low as 0.5m2 (each pixel represents an area on the ground
about the size of a flat screen TV). Standard aerial surveys rely on an overlap between images to
create orthophotos or 3D models. Satellite versions are similar but usually rely on pointing
capability to create the overlap (the highest resolution commercial satellite on the market –
WorldView 4 – had a gyroscope failure just before Christmas and now cannot point at different
locations; in fact, it has been written off at a cost of $155 million).

FIGURE 1 (ABOVE)
Tri-stereo Pléiades imagery highlighting the potential
for different environmental conditions visible in the
imagery due to different viewing positions: (a) forward
pointing image; (b) nadir image; (c) rear pointing image;
and, (d) the tri-stereo image acquisition principle.

SCSI project
The Earth Observation and Remote Sensing Working Group of the SCSI’s Geomatics Professional Group applied to the European Space Agency (ESA) for funding, and was awarded satellite tasking time to quantify satellite capability for building measurement. We then specified the optimal date, time and location for each image capture and they would do their best to plan the orbits/point the sensor to accommodate. Pléiades imagery was requested in tri-stereo acquisition mode, which means images of an area from three separate locations during an orbital overpass. Imagery was requested with a four-week window between
subsequent acquisitions, allowing both for weather delays due to cloud cover and also for
sufficient progress on site to have occurred. Figure 1 illustrates the tri-stereo principle; this improves on standard stereo acquisitions from satellite and is ideal for urban or built-up areas, providing an additional image in the centre of the site and reducing data shadowing or occlusions.

FIGURE 2 (ABOVE):
Proposed study site: (a) Dublin bay located on the east
coast of Ireland is part of a designated UNESCO
biosphere; and, (b) conceptual design plans for the
Alexandra Basin Redevelopment in Dublin port.

Earth observation satellites in action
The study area (Figure 2a) is located in the inner part of Dublin Bay on the east coast of Ireland. The Alexandra Basin Redevelopment (ABR) Project (Figure 2b) is located in the approaches to Dublin Airport and 2.5km south west of a nature and bird sanctuary, limiting options for aerial or RPAS surveys. The ABR Project is the first part of the larger redevelopment of Dublin Port that forms part of Dublin Port Company’s Masterplan 2012-2040 (an interim review of the Masterplan was carried out in 2017), and the scale and life cycle of a project of this kind is critically dependent on regular, rapid mapping for monitoring of major infrastructural development in urban environments. Furthermore, the ongoing offshore and onshore development of the Port presents additional opportunities for demonstrating the utility of satellites as a non-contact measurement technology over large areas.
Imagine Photogrammetry (the old Leica Photogrammetry Suite) was used to generate a point cloud of Dublin Port using the Pléiades tri-stereo imagery. In the case of this project the enhanced automatic terrain extraction (EATE) dense matching algorithm was utilised for the purpose of generating the point cloud. The EATE algorithm enabled extraction of high-density 3D information from overlapping imagery by
identifying common points in the images, and then transformed these points into accurate XYZ terrain points. It is important to note that one of the main factors that influence the quality of any
photogrammetric 3D reconstruction process is the amount of overlap between the imagery. The higher the overlap between the imagery, the more robust the image-matching solution will be, and this will result in a more accurate 3D reconstruction of the surface. The tri-stereo imagery used in these tests helped to maximise this. Following creation of block files containing raw imagery and all information essential for
triangulation (including imagery, sensor information and orbital altitude of the satellite), an automated process was then initiated whereby the interior (internal sensor parameters) and exterior (sensor position, pitch, roll, yaw) orientation were determined using the rational polynomial coefficient files, which contained the sensor and image information. Tie points were then also automatically detected in overlapping imagery.

Part II of this article will be published in the Summer 2019 edition of the Surveyors Journal.

Authors
Dr Conor Cahalane FSCSI, FRICS
Department of Geography, Maynooth University
Darragh Murphy
GIS and EO Analyst at Mallon Technology
Aidan Magee
Doctoral candidate at Maynooth University
Dr Avril Behan FSCSI FRICS
Lecturer and Assistant Head of the School of Multidisciplinary Technologies, TUD
Stephen Purcell FSCSI, FRICS, MIPI
Director, Future Analytics Consulting Ltd
Eimear Mcnerney FSCSI, FRICS
GIS and Mapping Specialist, Planning and Asset Management, ESB.