Ten Top Tips Learnt Working for a Small Remote Sensing Company

Artist's rendition of a satellite - mechanik/123RF Stock Photo

Artist’s rendition of a satellite – mechanik/123RF Stock Photo

I am approaching the end of my year at Pixalytics, and this blog is summary of what I’ve learnt from working for a small commercial remote sensing company.

The work itself has been a real blessing for me. Remote sensing product development was just the role I had been looking for, so I took it on with relish. During the year I have spent time researching, and supporting the product development of, flood mapping using SAR imagery, vegetation time series and light pollution.

I’ve learnt a huge amount over the past twelve months, and here are my top ten tips on researching & developing remote sensing products:

  1. Keep in mind who your stakeholders are and exactly what they require.
  2. Ensure your ground site is really covered by the satellite image, as coverage tends to be diagonal rather than straightforward latitude and longitude square and can miss a site altogether.
  3. Practise program version control at all times!
  4. Check the images you are using are the best ones for your requirements, i.e., not 16 day composites when daily images are more suitable and available; stopping you wasting a day downloading the wrong images!
  5. Write down problem solving routines, so next time you can do it for yourself!
  6. It’s always important to run pilots and streamline programming. This will save time and effort, and help verify that your end product is statistically robust.
  7. Write down what you find and keep good records of your algorithms and programming, so that you don’t duplicate work.
  8. Write technical notes on your work, so that programs can be easily shared, reviewed and run by others.
  9. Allow sufficient time before deadlines for reviewing and reworking.
  10. Make notes on the data you are using as you go along, including source, dates, locations and any company/organisation credits needed.

These are all lessons I’ll be taking with me when I leave, whether in commerce or academia.

It’s also been an insight into how a business is run, via these activities and hearing (one side!) of Sam’s teleconferences. Plus I’ve been involved in valuable encounters with the Environment Agency on products and have attended conferences, and given a presentation at one, on behalf of Pixalytics.

Plymouth has also been fun to explore. I’ve enjoyed visiting the various arts venues all over the city together with the galleries and museums, festivals and excellent cuisine.

Many thanks to Sam and Andy at Pixalytics for giving me this opportunity. I’m sad to leave and have enjoyed my time here.

Blog written by Dr Louisa Reynolds.

The Day the Lights Dimmed

According to the paper published by Falchi et al in June 2016 around 80% of the world’s population suffer from light pollution. The paper, ‘The new world atlas of artificial night sky brightness’, further noted that in Europe and the USA over 99% of people experience skyglow.

Skyglow is one part of light pollution, and refers to the brightening of the night sky over inhabited areas. The prominence of this feature was demonstrated last week in Puerto Rico when a large fire in the Aguirre Power Plant, in the area of Sanlinas, caused the lights to dim across the island.

The fire began when a power switch overheated causing an oil tank to explode. The resulting fire spread over a three acre area and effected power generation and cut off water supplies. Around one and half million people lost power equating to over 40% of the island’s population and 350,000 people were cut off from water.

This power loss gave a spectacular example of the skyglow effect, as it was possible to produce comparable night time pictures from satellites. Pictures twenty-four hours apart were taken by the Visible Infrared Imaging Radiometer Suite (VIIRS) instrument aboard the Suomi NPP satellite. In a recent blog on the Rio Olympics, we described the instrument in detail.

Image of Puerto Rico, acquired on 21st September 2016 from the VIIRS instrument. Data courtesy of NASA/ NASA’s Earth Observatory

Image of Puerto Rico, acquired on 21st September 2016 from the VIIRS instrument. Data courtesy of NASA/ NASA’s Earth Observatory

On the right is the ‘before image’ acquired at 2.50am local time on the 21st September. On the North coast, just to the right of centre, the bright white concentration shows the light from the capital city, San Juan. This city is the centre for manufacturing, finance and tourism for the island, and the site of its key seaport. Light can also be seen around the edge of the island, which effectively maps the islands interstate highways. The power outage affected the whole island including the westerly cities of Mayagüez and Aguadilla, the southern coastal city of Ponce and Humacao on the east coast.

Image of Puerto Rico, acquired on 22nd September 2016 from the VIIRS instrument. Data courtesy of NASA/ NASA’s Earth Observatory

Image of Puerto Rico, acquired on 22nd September 2016 from the VIIRS instrument. Data courtesy of NASA/ NASA’s Earth Observatory

Compare this with the ‘after image’ to the right which was taken approximately twenty four hours later at 2.31am on the 22nd September. Power had already started coming back on by this point, but only 130,000 were connected in the first twelve hours, and so there is still a major outage. The concentration around San Juan is reduced significantly, as are the lights mapping the interstate highways. All the areas are still identifiable, but the reduction in skyglow is apparent and obvious.

Whilst the pictures of cities and islands at night can be amazing, light pollution does have negative impacts on both us and the natural world – particularly nocturnal wildlife.

These images demonstrate the impact of skyglow, and we should all look to try and reduce the amount of light pollution in own lives, cities and countries.

Can Earth Observation answer your question?

The opportunities and challenges of utilising Earth observation (EO) data played out in microcosm in our house over the weekend. On Sunday afternoon, I was watching highlights of the Formula One Singapore Grand Prix which takes place on the harbour streets of Marina Bay and is the only night race of the season. To ensure the drivers can see, there are over 1,500 light projectors installed around the circuit giving an illumination of around 3,000 lux.

Whilst watching I wondered aloud whether we’d be able to see the track from space with the additional floodlights. My idle wondering caught Sam’s interest far more than the actual race and she decided to see if she could answer the question. The entire circuit is just over five kilometres long, but it’s a loop and so an approximate two kilometre footprint; any imagery would need a spatial resolution less than this. The final difficulty is that the data needed to be this weekend, as the circuit is only floodlit for the racing.

Within a few laps Sam had identified free near real time night data available from United States National Oceanic & Atmospheric Administration (NOAA) which covered the required area and timeframe. This was from the Visible Infrared Imaging Radiometer Suite (VIIRS) using it’s Day/Night band with a 750m spatial resolution – this resolution meant we would not be able to see the outline of the track as it would be represented by only three or four pixels, but it would be interesting to see if we could identify the track feature. By the end of the race Sam had selected and downloaded the data, and so we could answer my question. However, it turned out to be not quite that easy.

VIIRS Singapore night time imagery, data courtesy of NOAA

VIIRS Singapore night time imagery, data courtesy of NOAA

NOAA data uses a slightly different format to the image processing packages we had, and we couldn’t initially see what we’d downloaded. Sam had to write some computer code to modify the packages to read the NOAA data. For anyone thinking this is an odd way to spend a Sunday evening, to Sam this was a puzzle to solve and she was enjoying herself! After some rapid coding we were able to view the image, but unfortunately the Saturday data wasn’t useful. On Monday we tried again, the Sunday race took place on a clear night and we’ve got a good image of the area, which you can see above. On the larger image you can clearly the Indonesian Islands with Jakarta shining brightly, up through the Java Sea where the lights of some ships are visible and then at the top of the image is Singapore; the zoomed in version of Singapore is the inset image.

Despite the floodlights used for the race, Singapore and some of the surrounding Malaysian cities are so bright at night that the additional lights simply contribute to the overall illumination, rather than making the track stand out. Hence the answer to my question is that the 2014 floodlit Singapore F1 street circuit can’t be distinguished from the surrounding area at this spatial resolution. Of course if we purchased high resolution imagery we may be able to see more detail, but we thought that was going a bit far for my idle wondering!

EO can answer questions like these quickly; and whilst we know not many businesses are dependent on whether the Singapore Grand Prix can be seen from space, but change this to what is the light pollution in your area, what is happening in terms of deforestation in the middle of the jungle, what phytoplankton are doing in the middle of the ocean or whatever question you might have, then EO might be able to provide the answer in a short space of time.

However, there are two main difficulties in getting the answer. Firstly, you’ve got to know where to find the data and secondly, what do with it when you get it. Currently this can be challenging without specialist knowledge, making it inaccessible for the general population. In the coming weeks, we’re going to write some blogs looking at the freely EO data available, and the easiest way of viewing it. Hopefully, this may to help you answer your own questions. In the meantime if you have questions you want answered, get in touch, we’d be happy to help.