EEO-AGI Scotland Seminar Review : Ed Parsons, Google : 1st Dec

Friday 1st December saw a very healthy turnout for the last EEO-AGI seminar of 2017, given by the engaging and convivial Ed Parsons, Google’s Geospatial Technologist and self-professed geographer-in-residence. Having moved from GIS Applications Manager at Autodesk, to Chief Technology Officer for Ordnance Survey, and acting as Executive Fellow at University of Aberdeen and Visiting Professor at UCL, Ed’s career has carved a path in geospatial data management and visualisation. In his current role, he now seeks to evangelise Google’s efforts to improve the world using geospatial data. As he sees it, Google may seem like a giant frightening techno-monster; but Ed is here to act as the friendly geographer conduit.

Ed’s enjoyable talk focused on his near and distant predictions for the future of technology and the use of spatial data. 56% of all mobile Google searches are for local information, and with Google Maps now serving 1 billion users, an incredible portion of the world’s population is involving spatial information in its daily life.  Suddenly the giant techno-monster analogy didn’t seem so far off…

As Ed joked, ‘only idiots do lectures about the future’, but he made a decent effort in presenting megatrends we can no longer ignore. We are coming into a world of urban living, where residents are comfortable with technology and businesses increasingly make successful use of ‘big data’, APIs and web services. Ambient location is now becoming a natural part of life, with the introduction of Google Maps app for iOS in 2007 helping maps transition from static information to dynamic tools for daily tasks. Maps now operate as egocentric, placing the user at the centre of the data. With such capabilities as travel notifications as you walk into a train or bus station, digital assistants like Alexa and Siri, and lights and thermostats that turn on when you reach the vicinity of your house, ‘science fiction technology’ and the use of locational data is becoming part of ordinary life. Indeed, Ed used a quote from Mark Weiser, chief scientist at Xeroc PARC to portray that “the most profound technologies are those that disappear. They weave themselves into the fabric of everyday life until they are indistinguishable from it.” The question of data privacy was raised, and countered with the reassurance that at the heart of ambient location is the freedom to withhold your information. It is very rare nowadays to get lost with locational data at our fingertips, but we must be offered the choice to get lost if we want to.

Voyaging into the world of virtual reality, Ed acknowledged the criticism and the lack of uptake of the Google Glass headwear but highlighted the potential applications and advantages of 3D modelling and user-friendly augmented reality software. The hardware still experiences limitations in true 3D movement, but with the use of SLAM (simultaneous location and mapping) software, photogrammetry techniques are helping to fill in the blanks of indoor mapping and movement in virtual space using just handheld processing power.

Ed next presented a playful analogy of technology over the last three decades as headgear: the hardhats of 1995-2005 (focused on defence, engineering and inexpensive solutions), the fedora and sunglasses of 2005-2015 (mocking hipsters creating a stylish, entertaining and mobile internet) and the robot head of the present (depicting the development of artificial intelligence). Earth Engine, Google’s cloud-based platform for remote sensing image analysis, now offers a fast, free, up-to-date solution to traditionally slow and clunky remote sensing programs. With over 5 petabytes of data available, Ed gave a demo of the platform’s impressive ability to remove clouds from aerial imagery over the UK on-the-fly. As is the way with live demos, a minor snag required Ed to re-log in to his Google account and then perform two-factor authentication, spawning laughter but allowing him to declare the usefulness of this security measure: “I could have been a malicious person trying to steal Ed’s details!”

The tech evangelist finished with some examples of machine learning, with feature recognition in driverless cars, computer-controlled drone racing, landmark recognition from frequent congregations of people sharing their location (“no human was used in the making of this map”), and Global Fishing Watch taking 10TB of ship tracking data and aerial images to identify illegal fishing hotspots. Ed finally thanked everyone who has ever filled in a CAPTCHA, explaining Google’s improvements of street sign and business name photo-recognition by using the training dataset of millions of CAPTCHA answers entered by unwitting humans. Overall, the seminar proved an enjoyable tour of Google and others’ offerings to the progression of geospatial data usage, and where we may see ourselves (and additionally the pitfalls, legalities, and questions of morality we must be aware of) in stepping into a future of machine-and-human harmony.

Freya Muir

(MSc in Geographical Information Science at University of Edinburgh)

Advertisements

EEO-AGI Scotland Seminar Review : Dr. Lauren Biermann : 17th Nov

[NB Next Talk: 2018 Opener – David Henderson, MD OSGB Fri 26th Jan]

On Friday 17th November we were delighted to welcome Dr. Lauren Biermann, Senior Satellite Scientist at the Centre for Environment, Fisheries and Aquaculture Science (Cefas), for the second in this series of EEO-AGI Scotland seminars, held jointly with the University of Edinburgh’s Hutton Club. Lauren gave us a very engaging seminar entitled ‘Building Bridges to Satellites: how the UK Government is encouraging uptake of remotely sensed data’, in which she talked about her roles as data scientist at Cefas, Marine Lead at Defra’s Earth Observation Centre of Excellence (EO CoE), and Marine Liaison for UK-GEOS, a cross-government Earth Observation service.

The talk started with a brief introduction to EO, and the benefits and limitations of satellite data for marine science. An obvious limitation is that the vertical complexity of the oceans is hidden from satellites, which can only measure the surface layer. Satellite data also suffer from reduced accuracy in more complex waters (including those around the UK), though this problem is beginning to be addressed by improved algorithms.

The benefit of EO that Lauren emphasised was one that I had not thought about before, but seemed obvious when it was pointed out: its ability to tie together observations at different spatial scales, from local to global. She gave the example of coastal water quality: this could be affected by very local processes; by more regional processes such as river runoff; by pollutants being carried onshore by ocean currents; by large-scale oscillations such as El Niño; right up to global climate change. Marine scientists study these processes using in situ measurements and computer modelling, and the satellite data is what ties together these different observations and spatial scales. This is perhaps especially true now that high-resolution satellite imagery is becoming more available, as this allows smaller-scale phenomena to be studied from satellite data. A beautiful example of this was shown in this talk: a Sentinel-2 image of a Black Sea algal bloom, with a single vessel and its wake clearly visible.

Lauren then presented one of the projects that she worked on with Cefas: Tracking North East Atlantic Mackerel with Earth Observation Data (aka Finding NEAM-EO – an acronym of which she can be proud!). This project started some time ago, looking at how EO data could be used to predict mackerel distribution in the North Sea, but was then picked up by policymakers in the run-up to the Brexit vote because of the so-called ‘Mackerel Wars’ [1]. This dispute pitted Britain, Norway and the EU on one side against Iceland and the Faroe Islands on the other. Although it has now been resolved, the situation could become more complicated once Britain leaves the EU. In disputes of this kind, it is vital to know exactly where the fish are: the argument that Iceland and the Faroes put forward for unilaterally increasing their quotas was that mackerel had shifted their range northwards.

In the North Sea, the Finding NEAM-EO project had used in situ data on mackerel distribution taken from acoustic surveys, and related this to variables that could be measured from satellites, such as sea surface temperature (SST) and chlorophyll concentration. The results showed that SST was a very good predictor of mackerel distribution, along with bathymetry.

Cefas were then asked to extend the project to the North East Atlantic, but there was no in situ acoustic data available to validate the predictions made from the EO data. However, it was possible to track the mackerel trawlers, which are required by law to carry Automatic Identification System (AIS) transceivers. Mackerel could be assumed to have been present wherever mackerel ships slowed down to trawl. This idea was adapted from Dr. Biermann’s PhD work, in which she investigated the foraging behaviour of tagged elephant seals in the Southern Ocean. In the North Sea, the hotspots identified from the AIS data agreed well with the acoustic data, showing that this was a valid method for tracking mackerel distribution. In the Atlantic, the hotspots showed the same strong relationship with SST and bathymetry as in the North Sea.

After giving us this great example of how satellite data can be used to inform policy, Lauren turned to her role as Marine Lead at Defra’s EO CoE, which involves convincing UK Government to integrate satellite data into monitoring and policymaking. There was initially some scepticism among senior civil servants about the value of satellite data, perhaps because of a lack of information about the range of satellite products available. Although Defra have become more engaged with EO over the last few years, there are still barriers to its uptake. Satellite data are big and require specialist knowledge and software to analyse, so the main task of the EO CoE is to make the data more accessible and easier to use for policymakers across government.

The first step was to understand which data government agencies needed to do their work effectively. Some of this data, such as NDVI/crop maps, SST and chlorophyll-a, could easily be made available in an accessible form. Other data products are currently in development, and Lauren spent some time talking about her role in each of these.

Firstly, AIS and radar maps are being developed in collaboration with Cranfield University to help government agencies understand the shipping pressures on UK waters. By overlaying radar images taken at different times, areas of high shipping intensity can be identified. Combining the radar data with AIS data could potentially allow automatic classification of vessels into different types (fishing vessels, recreational vessels, etc.). This is important in helping to police Marine Protected Areas (MPAs) – particularly in the Overseas Territories, where the UK’s main MPAs are. The aim is not so much to catch individuals as to understand patterns so that limited resources can be allocated more effectively.

Secondly, Cefas and Plymouth Marine Labs are working to develop the first product from Sentinel-3 data: 300 m resolution maps of suspended particulate matter, which will help to understand changes in turbidity in the North Sea as well as UK coastal waters.

Finally, Lauren talked about a project she worked on for UK-GEOS, which was set up to promote the integrated delivery of EO products across government departments. This project investigated the use of satellite imagery to identify river runoff plumes, and predict the types of pollutant they might contain based on the industrial or agricultural facilities that were upstream. This is very relevant for understanding the causes of fish diseases, and for improving the sustainability of aquaculture. Consumption of farmed fish surpassed that of wild-caught fish for the first time last year, so this is of vital importance to the sustainability and security of our food supply.

After making a good case for the integration of satellite data with policy and giving us many fascinating examples, Dr. Biermann ended her talk with some more general remarks about the importance of sea literacy – not just among scientists and policymakers, but in society as a whole – and of greater collaboration both within government and between government and academia. The future health of our seas and coastlines may depend on it, and I’m sure that some of those present in the audience will want to be involved.

Daniel Stow (MSc Earth Observation & Geoinformation Management)

[1] http://britishseafishing.co.uk/the-mackerel-wars/