Grímsvötn 2011 (Part 2): Effects on aviation of the biggest Icelandic eruption since Katla 1918

The 2011 Grímsvötn eruption was the biggest explosive Icelandic eruption since Katla 1918*, producing twice as much material as Eyjafjallajökull 2010 in around one tenth of the time.  During and after the eruption, scientists measured the effects that it had on the UK.  The final version of our paper containing the findings was published last week.

The paper uses results of a citizen science tape-sampling exercise (co-organised by the British Geological Survey and this blog) along with data from public agencies such as the Department for Environment, Food and Rural Affairs (DEFRA), the Scottish Environment Protection Agency (SEPA) and the Met Office, to show where and when the ash fell.

We found that the north of the UK (mainly Scotland) received a light dusting of ash in the 48-72 hours following the beginning of the eruption, which caused little or no health or environmental problems.  The previous blog post explains how we know this.

As with 13 months earlier, the greatest impact of the Grímsvötn eruption on the UK was the disruption to aviation.  This blog post discusses how new flight rules introduced during the Eyjafjallajökull 2010 eruption meant that the eruption caused much less trouble than it could have done.  It is more complicated than the Ash cloud is a myth! – Oh no is isn’t! story presented at the time.  In short, I think that computer models did a pretty good job of predicting where and when there would be volcanic ash, and that new flight rules kept London airports open, but that initial estimates of contaminated airspace were too large and the currently defined concentration zones are flawed anyway.

Effects on aviation of the Grímsvötn eruption

Given the lack of health or environmental damage from the eruption, perhaps the title of this blog post should have been “Biggest Icelandic eruption since Katla 1918 doesn’t really affect the UK”.  But, of course, there was an impact on aviation.  Nine hundred flights were cancelled between 23 and 25 May and the closing of airports was controversial, with the media giving a significant platform to angry airline bosses venting their frustration.

This figure is less than 1% of the over 95,000 flights cancelled during the Eyjafjallajökull eruption of 2010, despite the Grímsvötn eruption being bigger.  I have explained the main reasons for the limited disruption in a previous post.  An important one was the way that the authorities handled this eruption.  Below I describe what I think went well, and what can be improved.  Although my research is done in collaboration with a number of different organisations, the views here are my own.

Computer models did a pretty good job of predicting where and when there would be volcanic ash…

Our new results show good agreement with the location and timing of ash-affected areas predicted by the Met Office model.  I think that it’s really cool that predictions made using data from some of the world’s most powerful supercomputers can be tested by a bunch of kids with sticky tape.  However, saying where and when the ash will be is the easy bit; the real challenge is estimating concentrations.  As I have previously written, making concentration maps of ash clouds is really hard.

Paths for wind passing over Grímsvötn volcano at 00:00hrs on 22 May calculated by the Met Office computer model. They cross Scotland and the northern UK on 24 May, in good agreement with our results.  See the full paper for more details.

… and new flight rules kept London airports open…

Prior to 2010, the rules were that aircraft should avoid all ash.  When the Eyjafjallajökull eruption caused the widespread closure of airports across Europe, the rules were relaxed to allow aircraft to fly, even if the presence of some ash had been predicted.  Three different zones of ash concentration were introduced.  The zone of Low Contamination (blue in the map below) has a concentration of 200 micrograms per cubic metre (a few grains of fine sand in a bath).  This is the same level that was used to identify contaminated airspace under the old system and is similar to the detection limit for satellite methods.  It is now permitted to fly in this zone.

Zones of Medium (coloured grey) and High Contamination (coloured red), which correspond to concentrations of 2000 micrograms per cubic metre and 4000 micrograms per cubic metre, were introduced to mark the areas of most concentrated ash.  Aircraft can only enter these with special permission.  These rules were in force during the Grímsvötn eruption.

Section of map of predicted ash concentrations on 25 May during G2011. Under the old flight rules, all coloured areas would have been out of bounds. This would have resulted in another closure of London airports. Source: Met Office, Crown Copyright 2011. Click image for full plot.

The map above shows that highest concentrations are expected in the northern UK (as we actually found in our study), and flights were cancelled to and from northern airports such as Glasgow, Edinburgh,  Aberdeen and Newcastle.  But it also shows lower concentrations of ash across southern England and London.  Under the old rules this would have closed the airports there, too, causing massive disruption and losses for the second time in just over a year.  The new rules prevented this and were a big part of the reason that the Grímsvötn eruption had a smaller impact than Eyjafjallajökull, despite being significantly more powerful.

…but the initial estimates of contaminated airspace were too large…

As our findings show, computer models are pretty good at working out where the ash will go.  To estimate the concentration you need to know what is coming out of the volcano.  The mass discharge rate of a typical explosive eruption can be calculated from the height of the eruption column using an equation based on data from many past explosive eruptions.  This is standard practice at Volcanic Ash Advisory Centres across the world.  The equation is very sensitive and if you increase the plume size by 20% it doubles the mass discharge rate.  In Iceland, the height of the eruption column is usually estimated using radar data from the Icelandic Met Office.  The problem is that Grímsvötn 2011 was not a typical explosive eruption.

The eruption plume from the Grímsvötn eruption.  The top part travels north, but is mainly steam.  Most of the tephra travels south in the lower part of the plume.  Photograph by Ólafur Sigurjónsson í Forsæti.

The eruption plume from the Grímsvötn eruption. The top part travels north, but is mainly steam. Most of the tephra travels south in the lower part of the plume. Photograph by Ólafur Sigurjónsson í Forsæti.

The photograph shows the eruption plume.  The upper white part reaches nearly 20 km into the air, but it contains mainly steam and sulphur dioxide.  The steam was produced as the eruption passed through a lake within the ice.  Much of this fell back to Earth as dirty ash-filled hailstones.   Most ash grains (and pumice and other volcanic debris, collectively known as tephra) are in the bottom of the plume.  Sticky from the moisture in the plume, lots of the ash clumped together and fell down quickly to be deposited near the volcano.  This extra fallout meant that less ash left Iceland than was predicted by the standard methods.  This caused dispersion models to predict concentrations that were too high, and so too much airspace ended up in the grey and red zones.

Comparing the predicted ash clouds with information from satellites showed something was wrong.  There was much less ash over Greenland than the model had predicted.  With these observations, and following discussions with volcanologists (disclosure: I was one of them), the Met Office adjusted the model (details provided here, with permission) to reduce the amount of ash in the plume.  This decision certainly limited disruption during the later stages of the eruption.  If it had been made sooner then perhaps some of the flight cancellations during the early stages could have been avoided, too.

This is all much clearer with hindsight, of course, and this was the first eruption of Grímsvötn since it became necessary to estimate ash concentration.  To do better next time, the ability to collect and make use of accurate new data as quickly as possible is important.  The British Geological Survey and Icelandic Met Office are working together to improve volcano monitoring in Iceland, and current hot topics for research in the wider scientific community include finding ways to make it easier to compare model predictions with satellite images and to use satellite data to set model parameters.

…and the currently defined red and grey zones are flawed anyway

The introduction of zones of different ash concentration did a lot to keep aircraft flying during the two recent Icelandic eruptions, but this system still has two big problems.

The first problem is that the zonation scheme gives the impression that flying in the blue zone is ‘safe’, but this is not necessarily the case.  Volcanic ash accumulates continuously within jet engines as they fly through the cloud and so a long flight through the blue zone (or even in lower ash concentrations outside marked areas) may do more damage than a short flight in the red zone.  Furthermore, flying in any amount of ash will result in increased maintenance costs for aircraft operators.  It seems sensible to move to a system that estimates the ‘dosage’ of ash for given flight routes.

The second problem is that it isn’t actually possible to map the boundary between the grey and red zones.  I still can’t find an official justification online for why these levels are set where they are.  The blue zone, which is roughly equivalent to what satellites can detect, is reasonable.  The lower limit of the grey zone (2000 micrograms per cubic metre) is 10 times higher than the blue zone and can perhaps be justified because aircraft operating from dusty airports in desert areas already fly through this level of contamination (but of sand, which has a higher melting point than volcanic ash).

Setting the lower limit of the red zone at double this (4000 micrograms per cubic metre) makes little sense, because comparisons of satellite concentration estimates with measurements made by aircraft and other sensors show that they cannot distinguish between these two levels with any confidence.  Comparing the Met Office model predictions of peak concentration to other measurements (which have uncertainties of their own) shows that they agree within a factor of 2 only around a quarter to a third of the time.  This increases to a half to two thirds of the time if an 80 km wide buffer zone is used.  This is because making maps of ash clouds is really hard.

Given this uncertainty, the huge range in possible ash concentrations and evidence that ash-aircraft encounters that actually stopped engines involved concentrations of over 1,000,000 micrograms per cubic metre, setting the levels of different zones at 200, 2000, 20,000, 200,000 etc. would seem to be more appropriate.

What it all means

There are two main messages from this post:

  • Dispersion models are good at predicting where volcanic ash will go, and this system worked fine for two decades when aeroplanes had to fly around it.  Flying through volcanic ash requires estimates of the concentration, but these currently have large uncertainties.  Improving them needs further scientific research into eruption deposits, on-site monitoring, computer modelling techniques and satellite detection methods.  This is important to remember at a time when budget cuts in the US have severely reduced the capabilities of the Alaskan Volcano Observatory.
  • Things have come a long way since Eyjafjallajökull erupted in 2010 and new flight rules mean that only the very largest eruptions now have the capability to shut down all of European aviation (and there is a lot of it) in such dramatic fashion.  I suspect that the biggest economic threat to the UK in the future is probably from long-lasting eruptions causing short, but frequent and unpredictable, closures of small regions of airspace over periods of many weeks or months.

Further reading

This is the second of two posts about the effects of the Grímsvötn eruption on the UK.  Read the first post to learn where and when the ash fell.

Our study was published in the Journal of Applied Volcanology, which is an open access journal.  This means that anyone can download and read the full report for free by clicking the link below:

  • Stevenson, J. A., S. C. Loughlin, A. Font, G. W. Fuller, A. MacLeod, I. W. Oliver, B. Jackson, C. J. Horwell, T. Thordarson, and I. Dawson (2013), UK monitoring and deposition of tephra from the May 2011 eruption of Grímsvötn, Iceland, Journal of Applied Volcanology, 2(1), 3, doi:10.1186/2191-5040-2-3.

Last year, we published a similar paper in the Journal of Geophysical Research about the deposition of Eyjafjallajökull ash across Europe :

  • Stevenson, J. A., S. C. Loughlin, C. Rae, T. Thordarson, A. Milodowski, J. S. Gilbert, S. Harangi, R. Lukács, B. Højgaard, U. Árting, S. Pyne-O’Donnell, A. MacLeod, B. Whitney, and M.Cassidy, (2012), Distal deposition of tephra from the Eyjafjallajökull 2010 summit eruption, J. Geophys. Res., 117, B008904, doi:201210.1029/2011JB008904.

For other Iceland-volcano related posts, covering topics such as the probability of ash clouds reaching the UK, why volcanoes explode and an account of an expedition to Grímsvötn’s crater, follow the links from my Every Post Ever page.

* Technical point: There are a number of ways to define the size of a volcanic eruption, such as plume height, volume of material erupted, volume of magma involved.  These are incorporated into the Volcano Explosivity Index.  Here we are talking about the volume of widely-dispersed tephra deposited from a (sub-)Plinian eruption column.  The 1963-1967 submarine eruption of Surtsey, and the 1996 subglacial eruption of Gjálp both produced larger volumes of tephra (mainly hyaloclastite), but it was not widely dispersed.

Categories: Uncategorized

Grant applications are hard work (includes LaTeX template)

This post is a taste of (not so) slackademia.  It shows how much work is involved in preparing the funding application for a NERC Standard Grant.  It also includes a LaTeX template for anyone writing their own.  It can be used in any document where there is a tight limit on page numbers.

Grant applications are a lot of work

My recent work on the massive Hekla 3 and Hekla 4 eruptions shows that Hekla 4 produced much more extremely fine ash (<64 microns; this is the stuff that can cross the Atlantic and cause trouble in Europe) than Hekla 3.  We know that explosive eruptions are driven by gases in magma, so I want to get a post-doc for 2 years to look at the bubbles and the dissolved gases in pumice from the eruptions to see why they are so different.

It turns out that these applications are A LOT of work.  The final version required:

  • 13,000 carefully chosen, fully-referenced, words, co-written with 5 other authors (=> 5 sets of corrections/edits/comments) on 2.5 continents (Iceland surely isn’t a whole continent).
  • Detailed spreadsheets with well-researched and up-to-date costs.
  • 2 days of sampling (last summer) and 80 hours of specific analysis (this spring) to get one preliminary dataset; another day including instrument time for another.
  • Letters of Support from heads of departments, lab managers, senior government agency scientists, research professors.  The application is 23,000 words if you include Letters of Support and CVs.
48 sides of carefully thought-out science.

47 sides of carefully thought-out science.  Don’t underestimate how long it takes to put it together.

I reckon that it took around 3 months of full-time equivalent work to prepare.  This is time that I could have been analysing data or writing papers.  You can also add in another fortnight of other post-docs’ time.  Plus a few days of Professor/Senior Scientist time.  While working out the costs, I learned that employing a post-doc costs the tax-payer about £100,000 per year (of which about 1/3 is salary).  So we are talking about at least £25,000 of effort being put into a grant application.  Did I mention that only 1 in 5 get funded?

If it isn’t funded, the time will not have been completely wasted.  The data are useful, I’ve made new contacts and had an excuse to get right up-to-date with the latest published papers.  I understand the importance of being able to find the best-thought out and most useful projects to give funding from a limited pot, but it should also be recognised that every extra section on the application form has a real cost in terms of scientists’ time.

Grant application LaTeX template

LaTeX is an open source document preparation system.  Unlike a word processor, you only have to think about the text and it takes care of the formatting.  No more adding a word and seeing all your pictures jump to different pages.  If you have to write a complicated document with sections, subsections, references, tables and figures (such as a Masters or PhD thesis), then I highly recommend it.

Click here to download a LaTeX template for grant applications.  It is based on the normal article class, using the following extra packages:

  • anysize to set 2cm margins
  • helvet for Arial-like font
  • natbib and multicol with a custom .bst file for a compact reference list (Note: since 2014 NERC applications require references in 11pt font.  Comment out the \tiny{} before the bibliography to make the text full size.)
  • wrapfig to wrap text around images
  • pgfgantt to make a Gantt chart
  • titlesec and a number of other tweaks to make things compact

The packages are fairly common and can be installed on Ubuntu-like Linux systems with a single command (sudo apt-get install texlive texlive-latex-extra texlive-humanities texlive-fonts-extra).  The content of the template comes from this blog post, and the output looks something like this:


Categories: Uncategorized

Ash cloud travel insurance / why scientists should blog

I was quoted on the Daily Telegraph website at the weekend, in an article about the number of travel insurance companies whose policies cover volcanic ash.  I’d answered some questions from the author by email, then he told me when it was published so I could see how it turned out.  You can read the article here:

The article includes lots of good information and gets a big thumbs-up for explaining that the much-bigger 2011 Grímsvötn eruption caused little disruption compared to Eyjafjallajökull 2010, but I thought that the general tone was a bit too alarmist.

I was also interested to see how my quotes were used.  The following line in my email:

Hekla and Katla are both ‘overdue’ for an eruption e.g. the time since their last eruption is longer than the average time between other recent eruptions.

appeared in the article as:

Hekla and Katla are both overdue for an eruption.

The words are mine, but the message is different and without the explanation it seems a lot more urgent.  Katla has been ‘overdue’ since before I was born.  No real harm is done, and perhaps I am being a control-freak, but no volcanologist wants to be associated with scaremongering, either.

This experience sums up a big reason why scientists should blog.  News media have limited space, overly-enthusiastic headline writers and demanding advertisers.  Blogs don’t.  They provide all the space scientists need to explain new research, including all the complexities and limitations, and what they write is exactly the information that the reader gets.

Ash cloud insurance

The article also made me think about the concept of ash cloud insurance.  Some companies now cover volcanic ash-related claims, some don’t, and others charge a special supplement of up to £10.

All insurance is a form of gambling.  If you were to bet £10 that your £1,000 holiday would be cancelled, because the airport was closed by volcanic ash on the day that you were supposed to leave, you would only be better off in the long run if the chances of this happening were more than 1 in 100.  Your 99 ‘losses’ of £10 each would be cancelled out by one ‘win’ of £1000.

Planes have been forbidden to fly through volcanic ash for about 20 years now.  In that time, there have been six eruptions in Iceland (Grímsvötn 1996, Grímsvötn 1998, Hekla 2000, Grímsvötn 2004, Eyjafjallajökull 2010, Grímsvötn 2011).  Because of these, at least some UK airports were closed for about 10 days in total (~8 during Eyjafjallajökull 2010, ~2 during Grímsvötn 2011).  Therefore, the chances of flights being cancelled on any given day are somewhere in the region of 1 in 730.  This is equivalent to just 0.14 in 100, so that extra ash cloud fee is looking pretty expensive.

This is a simplistic analysis, and things are obviously a bit more complicated in real life, but you get the idea.  In gambling in general, the house wins in the end.  In ash cloud insurance, the house can win big.

Further reading

If you want to learn about this topic in more detail, here are my answers to the emailed questions in full:

1. How likely is it that Eyjafjallajokull will erupt in the foreseeable future?

Not that likely.  There has been little seismic activity there since the eruption ended and previous eruptions have been hundreds of years apart.

2. If it does erupt do you think it’s likely that we will see similar levels of disruption to the 2010 eruption?

Definitely not. The changes in rules for aviation during the E2010 eruption mean planes can fly in much higher ash concentrations than they used to be able to.

3. Are there any other volcanos that are likely to erupt in the near future, which could cause major travel and local disruption?

In general, we would expect an eruption in Iceland every ~5 years and a direct hit from ash every ~20 years.

Hekla and Katla are both ‘overdue’ for an eruption e.g. the time since their last eruption is longer than the average time between other recent eruptions.  Check the Smithsonian website for details.  Of these, a Katla eruption could be very damaging within Iceland:

The amount of travel disruption would depend on the length and size of the eruption.  A long eruption would be much more disruptive.

Other explosive eruptions from ice-covered volcanoes could produce a lot of ash, too, but would hopefully be short-lived:

Large lava-producing fissure eruptions, such as Laki (from the Grímsvötn system) and Eldgjá (from the Katla system) are the worst case scenario as they could last months and release large quantities of toxic gas.  But these are rare (e.g. once per 500 years).

4. Finally, to what extent do you believe that the early warning technology that has been developed by the Norwegian Institute of Air Research will prove successful?

I think that this method has potential and I think that Easyjet deserve credit for investing in research and technology.  The idea behind it is the same as satellites currently use to recognise ash clouds and the main scientist involved (Fred Prata) is a real expert in this field.  I’m looking forward to them announcing results of their tests to show how well it works.

Categories: Uncategorized

EGU2013: Dirty volcanic hail, geology blogging, open source science and fracking

Here are some of my highlights from last week’s European Geoscience Union conference.  These were presentations dirty hailstones formed in subglacial volcanic eruptions, a workshop on social media and blogging in geosciences, a splinter meeting on open source software in geoscience and The Great Debate on shale gas/fracking.

Dirty volcanic hail

Þórður (Thordur) Arason presented the first detailed study of volcanic ash-filled hailstones.  These are closely related to subglacial volcanic eruptions, such the two most-recent Icelandic events.   He studied examples from the deposits of the May 2011 Grímsvötn eruption that he collected, still frozen, from layers between the pumice and ash deposits that formed during the eruption, high on the Vatnajökull glacier

These ice-cemented spheres of volcanic ash formed during the Grímsvötn eruption in Iceland in May 2011, and are similar to the hailstones described at EGU 2013. I took this photo when I visited the crater area three months later. It was an adventurous trip, involving monster trucks and crevasses. Click the image to read more.

Arason measured the sizes of the hailstones (mostly 1-2 mm) from close-up photographs.  He weighed a big rectangular block of them, then allowed it to melt so that he could collect that ash grains inside.  The hailstones contained 15-40% ash, with grains from a few microns to over 1 mm in diameter.

Quantifying the contents of the hailstones is important for a number of reasons.  Firstly, by mixing ash grains with ice, you change the particle size and optical properties of the grains.  Arason demonstrated how this can lead to huge errors in measurements of ash plumes made by radar, and ice-covered ash is a problem for satellite measurements, too.  Secondly, by trapping very fine ash, the hailstones stop it drifting off downwind towards Europe.  These processes will be included in the next generation of computer models for ash dispersal.

It must take a lot of water to make so many hailstones.  Thanks to Magnús Tumi Guðmundsson, we have a pretty good idea of how much.  In his talk, he described  changes in the Vatnajökull glacier around the eruption site.  There is a permanent subglacial lake at Grímsvötn that periodically releases meltwater floods (jökulhlaups) out onto the lowlands, so Icelanders have detailed maps of the ice surface and of the bedrock beneath.

In this case, the volume of missing ice is equivalent to the water that went up with the plume, because there was no jökulhlaup during the eruption.  Guðmundsson found this volume was around 0.1 km3, which is about one seventh of the volume of tephra (pumice, ash and rock debris) that were produced during the eruption.  Averaging all this water over the 4 days when the eruption was most powerful gives a discharge of about 290 m3s-1.  This is equivalent to a fountain with 10% of the discharge of the Nile, shooting straight up into the air.

Social media and blogging workshop

The social media and blogging workshop panel included the geobloggers and tweeters Jon Tennant (@protohedgehog), Laura Roberts (@LauRob85), James King (@DrAeolus) and Dave Petley (@davepetley) and was chaired by EGU social media officer Sara Mynott (@EuroGeosciences).  They discussed the advantages of Twitter (finding breaking news first; access to well-informed people on any topic; making contacts from all over the world), and of blogging (explain things in more detail than traditional media are interested in; raise your academic profile; become a contact point for journalists interested in your subject).

I was interested to hear that some PhD student bloggers are writing on blogs that their supervisors had set up but didn’t have time to write themselves.  It was also interesting that some climate scientists are discouraged from blogging by the abuse that they receive in their comments from climate change deniers.

Dave Petley is a professor in the Institute of Hazard, Risk and Resilience at Durham University and runs The Landslide Blog, hosted by the American Geophysical Union.  By professor, I mean in the UK senior-academic-who-runs-his-own-research-group sense of the word, as opposed to the US academic-with-a-permanent-(tenured)-position sense.  His story was especially convincing.  In the six years since he began his blog, he has seen traffic to the site gradually increase to over 1,000 visitors per day.  Over the same period, he showed how citations to his papers had also risen sharply and said he now receives many more invites to meetings and conferences.

Dave also described how a blog post that he wrote about a fatal flash flood in Nepal became the global focal point for people looking for information on the event, including Nepali journalists and visiting tourists.  An ex-Soviet military pilot provided YouTube footage showing that the cause was a landslide from the mountain Annapurna IV, and NASA contributed satellite imagery.  The results of the study will be written up as a scientific publication.

I’ve found a similar benefit from blogging.  In 2011, I wrote a post asking members of the UK public to collect ash fall from the eruption of the Grímsvötn eruption, and posted the request on Twitter.  We received over 130 samples from across the country, and the results, which include a map of where ash was found, will be published in the Journal of Applied Volcanology in the next few weeks.

Free and Open Source Software (FOSS) in the Geosciences

Following an oversubscribed splinter meeting last year, the profile of Free and Open Source Software in the Geosciences continues to increase.  This year featured another Splinter Meeting and a dedicated session featuring both talks and posters.  I made it along to part of the Splinter Meeting, which highlighted the benefits of using free/open source software and displayed the huge and growing range of tools that are available.

The panel highlighted that a great way to test some of these tools out is to download the OSGeoLive DVD, which contains the latest versions of over 50 different packages.  Simply fire up your machine with the disk in the drive and it will boot into a fully-functioning Linux desktop with all the software installed and ready to go.  When you are done playing, shut the machine down and take the disk back out.  Your original operating system will be untouched.

The OSGeo-Live DVD is a great way to try out open source GIS software

The arguments for open source in science were strengthened recently by editorial in the journal Nature, and by articles in the journal Science.  The Scientific Method rests on experiments being tested by different people.  Many advances in science come from computer modelling, but if scientists do not publish their code, how can others test it?

Related to this were issues about where the code should be stored.  Ideally, the code would become citable item so that scientists get recognition when others use it. Reproducibility of results was also discussed.  Using open source software, as opposed to proprietary code whose internal workings are a commercial secret, ensures that the exact versions of software used will always be available to those attempting to reproduce a result.

The Great Debate – Shale gas: to frack or not to frack

Hydraulic fracturing, also known as fracking, involves pumping high-pressure water into underground rocks, forcing them to crack and to release previously-inaccesible natural gas.  It is a controversial process, and the EGU Great Debate was advertised as an opportunity for top scientists do discuss the pros and cons of getting fossil fuels in this way.  These are summarised nicely in the session outline.

I was excited to watch this debate taking place in front of a technical audience, looking forward to getting into the details of charts of production rates in wells, descriptions of changes to rock properties during fracking and projections of future changes in global gas prices.  Many others were, too, and it was standing room only.  Disappointingly, the whole event turned out to be very thin on data.  One guy said that shale gas would all be gone in fewer than 20 years, then another said that it would last more than 100.  Neither presented any evidence for where those numbers had come from.  A wasted opportunity, I think.

[Since getting home, I've found Matt Herod's write-up of the debate.  It seems that he was also disappointed by lack of hard data.  His post contains a link to the video of the debate and some good background information on fracking in general].

Categories: Uncategorized

Two years of volcan01010: Highlights of 2012

This week is the second anniversary of volcan01010.  With this post, I want to give an overview of what I have been writing about over the past year.  If you are new to the blog, then it should give you an idea of what it is all about: Volcanoes, Iceland and (geo)scientific computing.


The volcano posts of this year, as last year, are quite Icelandy.  They cover specific aspects of the recent Grímsvötn and Eyjafjallajökull eruptions, as well has some general volcanology themes.

  • Ten swimming pools of travel chaos
    For all the trouble that it caused, the volume of Eyjafjalljökull magma that formed the ash that reached Europe was actually quite small.

  • Gas, not ice, makes subglacial rhyolite explode
    A feature-length post explaining why volcanoes explode, with an icy twist.  This is one of the posts that I started the blog in order to write.

  • Happy Anniversary Grímsvötn
    How variable wind directions split the Grímsvötn plume and showed that real volcanic eruptions are always more complicated than the theory suggests.

  • Sounds of the Underground
    Sped up seismic data lets you ‘hear’ earthquakes and eruptions from an earthworm’s point of view.  Hear Colima volcano sing.

I spent 18 weeks of last year on fieldwork in Iceland.  It was busy, but I managed a few posts from the field.  For 3 weeks, I had @al_monteith with me, so check out his blog for more Iceland fieldwork photos and stories.

  • Iceland horse fun
    Iceland horses have a lot of personality – a postcard from the field.


I also still get lots of hits on my All the software a geoscientist needs. For free! post, which is good as this is another of the posts that I always wanted to write.

General geology / Environment:

Stories from 2011

If you liked any of these, there is another batch in the One year of volcan01010 post.

Coming up in 2013

It’s been a busy year.  I managed to write something new at least once per month this year and I hope to keep that pace through the next year.  When you follow loads of journalists on Twitter it makes your productivity feel really low, but they don’t have maps to make, samples to analyse or students to teach!  Some things in the pipeline include the results of the Grímsvötn 2011 public sticky tape sampling exercise and a video on how to use Python for science.  Stay tuned.

Subscribe, then tell all your friends

The best way to follow volcan01010 is to subscribe to the RSS feed.  If you’ve never heard of RSS, read this guide.  It lets you keep track of posts that you have read and tells you when a new one is out.  You can also follow me on Twitter (@volcan01010).  This way you get updates with other news and links that I think are cool.  Remember that you can also find old posts on the Every Post Ever page.

Volcan01010 now has 459 followers on Twitter (up from 201 last year), and in the last 12 months the blog scored 20,897 page views from 10,961 unique visitors in 147 countries (with the vast majority in the UK and USA).  The numbers of hits are similar to the previous year, but there were no eruptions in Iceland this year (during the eruption in May 2011, the Grímsvötn eruption – Frequently Asked Questions pulled in 1,400 hits in a single day).  Traffic comes in more steadily now and is spread across more posts.  It is satisfying that there hasn’t been a single day in the past six months when fewer than 10 people visited the blog.

If you find the blog interesting or useful, then please tell all your friends.  Then tell them to tell all their friends, too.

scientist talking to a load of other scientists.  There is no hype, no dramatic music, and no cute baby polar bears.  There are only data, graphs and trying to understand what they all mean.
Categories: Uncategorized

Processing ARSF remote sensing data with open source GIS tools

Orthorectified aerial photographs of the Nesjahraun, Iceland, draped over LiDAR-derived topography visualised in GRASS’s NVIZ module. Click to enlarge.

The UK National Environment Research Council’s Airborne Research and Survey Facility (NERC ARSF) is an aircraft used for scientific research. The on-board instruments include LiDAR, high resolution camera, and multispectral and hyperspectral infrared scanners.  The aim of this post is to share an article and slides from a talk that I gave in 2009 that demonstrate how these data can be processed with free and open source tools.

The slides and articles also illustrate some of the ways that GRASS GIS can be used to process remote sensing data in general.

Getting started with GRASS GIS

Multispectral infrared data channels combined in GRASS to highlight different landscape/vegetation features. If you have visited Iceland, you may recognise the Nesjavellir geothermal plant, with its steam and warm river. Click the image to enlarge.

GRASS GIS is a fully-featured open source GIS package.  In talk and article I describe using it to process LiDAR point clouds, orthorectify aerial photos, calculate slope/aspect maps, digitise geological information, import field data and to visualise and export data in a variety of formats.  From a remote-sensing perspective, the Image Processing features include things like image classification and fusion.

Early versions of GRASS were mainly command-line based, so the learning curve can be a bit steep, but the GUI has improved significantly (especially from 6.4.2 onwards).  The command line has the advantage of easy scripting, and the latest versions are moving to Python which will make it simple to include GIS processing in other workflows.

The following tips will help you get started with GRASS:

  • Install GRASS from the website.  As it was initially developed on the UNIX operating system, the Linux and Mac versions are better.  Windows users should install the OSGeo4W package, which also includes a number of other useful programs e.g. QGIS.
  • Download+print/buy a copy of Open Source GIS: A GRASS GIS Approach and have an actual physical copy beside you as you work.  It is an excellent reference.
  • Visit the pages for first time users on the GRASS website and learn about locations and mapsets.
  • Join the ‘grass-user’ mailing list.  Your questions will be answered very quickly (often by the developers themselves) and you can see examples of how others use it.
  • Bookmark the GRASS Search Engine, where you can often find answers in the mailing list archives.

Further reading

Orthorectified aerial photographs of the Nesjahraun, combined with digitised field data exported to GoogleEarth via GDAL. Click to enlarge.

To see a list of open source tools useful to geoscientists, see All the software a geoscientist needs: For free! There is also a list on Wikipedia of free and open source geophysics software.

The data that I processed was used in a study of an Icelandic lava flow that went into a lake.  You can see how it was used in the following articles:

  1. Stevenson, J. A., N. C. Mitchell, M. Cassidy, and H. Pinkerton (2011), Widespread inflation and drainage of a pahoehoe flow field: the Nesjahraun, Þingvellir, Iceland. Bulletin of Volcanology, doi:10.1007/s00445-011-0482-z.
  2. Stevenson, J. A., N. C. Mitchell, F. Mochrie, M. Cassidy, and H. Pinkerton (2011), Lava penetrating water: the different behaviours of pahoehoe and ‘a‘a at the Nesjahraun, Þingvellir, Iceland. Bulletin of Volcanology, doi:10.1007/s00445-011-0480-1.
Categories: Uncategorized

Gas, not ice, makes subglacial rhyolite explode

A recent study by Jacqui Owen (Lancaster University), Hugh Tuffen and Dave McGarvie shows that the explosivity of a subglacial rhyolite eruption is determined while the magma is still deep beneath the ground. Rhyolite is a particularly thick and gloopy (viscous) type of magma that is especially prone to explosive eruptions.  I’m really pleased to see this study published because I worked on subglacial rhyolite eruptions for my PhD. Where I mapped and measured pumice deposits to show that whether an eruption is explosive or not is not controlled by the ice, Jacqui has used geochemical methods to show what does control it: dissolved volcanic gases.

Explaining how she found out is also a good excuse to discuss a key question in volcanology: Why do volcanoes explode?

Subglacial rhyolite at Kerlingarfjöll

Part of the subglacial rhyolite volcano of Kerlingarfjöll, Iceland. Steep mountains like this, dominated by scree, pumice and ash were formed by explosive volcanic eruptions at times when all of Iceland was buried by ice up to 1000 m thick. Water from melting ice is known to make volcanic eruptions more explosive, but recent findings show that the effect is minor in subglacial rhyolite eruptions. Click to enlarge. Copyright: John Stevenson, 2012.

Volcanoes, explosions and ice

Why do volcanoes explode?

During an explosive eruption, pumice, ash and rock fragments (collectively known as tephra) are blasted from the volcano at high speed, driven by hot, expanding gases. The gases, mainly steam (i.e. water; H2O), have come from within the magma itself. Down in the magma chamber, deep beneath the volcano, the pressure of the weight of the rocks above keeps the gases dissolved, but when the liquid magma rises to the surface, the pressure that it feels decreases. The gases come out of the liquid and form bubbles, just as bubbles appear when you open a bottle of Irn Bru (other, less region-specific, fizzy soft drinks are available). The closer to the surface the magma gets, the more gas comes out, changing the magma from a thick, gloopy liquid, to a thick, gloopy, bubbly liquid, then to a thick, gloopy foam.

Imagine that the conduit (or pathway) through the rocks from the magma chamber to the surface is a rigid pipe, where the same mass of material that enters that bottom (as magma with gas dissolved inside) has to come out at the top (as magma + gas bubbles). This is a closed system. If the volume of a parcel of magma is doubled by forming bubbles, then the speed that it comes out the top must also double. Thus the magma accelerates. Triple the volume, triple the speed.  Quadruple the volume, quadruple the speed. The formation and expansion of gas bubbles drives the magma faster and faster to the surface.

This type of acceleration by expansion is perfectly illustrated by Diet Coke and Mentos eruptions, but volcanoes take things a step further. Once the magma reaches about 70% bubbles, the forces due to acceleration within the conduit become too much for the foam, which rips or shatters into fragments of pumice and ash. The full power of the gas expansion is now unleashed, no-longer limited by the flow rate of the thick, gloopy magma; the mixture accelerates ferociously. When it comes thundering out at the surface, it is travelling at speeds of over 300 metres per second (~1000 km/h, 670 mph).

Explosive eruptions are most common where the magma is so thick and gloopy (viscous) that gas bubbles cannot escape. This is most common with andesite, dacite and rhyolite magma types. The duration of these explosive eruptions depends on the availability of gas-rich magma in the magma chamber.

How does ice affect eruptions?

It is possible to make an eruption become explosive by adding water. The boiling point of water is well below that of magma (800-1000°C) and turning water into steam at atmospheric pressure involves a 1000x increase in volume. This can drive an explosive expansion that fragments magma into pumice and ash. The formation of the island of Surtsey, which grew out of the North Atlantic in 1963-1967, is a good example of an eruption of runny basalt magma that has explosive phases only when water mixed with magma at the vent.

In Iceland, most subglacial eruptions are basaltic. Melting ice adds water to the magma, turning the eruptions explosive.  The eruptions often take place within ice-dammed meltwater lakes.  If the volcano builds above the water surface, Surtsey-like, or if the lake level drops below the vent, then the eruption switches from explosive to effusive (lava-producing). Ice and meltwater therefore control the explosivity of subglacial basalt eruptions and the duration of the explosive phase of an eruption depends on the access of water to the vent.

More recently, researchers have investigated whether the same is true for rhyolite.

Subglacial rhyolite and volcanic gas

Magma chamber water content

Owen and coworkers sampled 5 different subglacial rhyolite mountains in Torfajökull, Iceland.  The first two were piles of pumice, ash and scree up to 1 cubic kilometre in volume, so we know that they formed in large, explosive eruptions. Two were made of lava and were formed in effusive eruptions, and the fifth one was produced by an eruption with both explosive and effusive parts.  The samples contained big crystals (called phenocrysts) that formed in the magma chamber. Some of these had grown in a way that trapped little blobs of melt inside them. These are called melt inclusions and they act as a record of the magma composition before the eruption.

Owen analysed how much water was dissolved in the magma within the melt inclusions.  She found that the samples from the explosive eruptions contained the most (up to 3.9 wt%).  A sample from the explosive part of the mixed eruption had the highest water content ever measured in Icelandic rhyolite (4.8 wt%). Melt inclusions from the effusive eruptions contained the least dissolved water (up to 1.8 wt%). Dissolved melt inclusion gas content therefore correlates with the explosivity of the resulting eruption. The magma “knows” whether it will be an explosive eruption or not while it is still in the magma chamber.

Microscope images of explosively and effusively erupted subglacial rhyolite.

Microscope images of subglacial rhyolite ash from explosive and effusive eruptions. Owen and coworkers found pockets of frozen melt trapped within the big crystals. These could be analysed to see how much magmatic water was in the magma chamber. Tiny crystals, called microlites, are found in samples from effusive eruptions. They formed as magma moved slowly to the surface, losing gas on the way. Image modified from article in Geology.

Gas leaks prevent explosions

It might seem strange, given the huge volume changes and accelerations, that a reduction of just 1-2 wt% in dissolved water content could be the difference between an explosive and effusive eruption, but Owen and coworkers also found differences in what happens to the gas on the way to the surface. The samples from the effusive eruptions showed evidence for ‘open system’ degassing.

Open systems (as opposed to the closed system described previously) allow gas to separate from the magma and to escape along the conduit or through cracks in the rocks. It’s the equivalent of opening the Irn Bru bottle slowly and carefully. SKSSSSSSCH! Sksssssch. Tssssssssssss…

Degassing the magma this way encourages tiny crystals (called microlites) to form. The samples from the effusive eruptions were full of them, while the ones from the explosive eruptions were not. If the gas can leak, then the force for acceleration is lost.  As the  effusively-erupted magmas began with less dissolved gas, they formed bubbles later and more slowly than the explosively erupted ones. This slowed the expansion of the foam, which slowed the rise to the surface, giving the gas time to escape from the magma.  Slower decreases in pressure slowed the release of more gas, which allowed even more time for escape and so on and so on until, instead of exploding, the magma was squeezed from the vent like steaming-hot, rocky toothpaste.

What does it mean?

No-one has ever witnessed a subglacial rhyolite eruption so our best way of working out how one would look, how much ash it would produce, or how long it would last is by looking at what remains of past eruptions.

The work on the pumice, ash and scree mountains of Iceland told us that adding meltwater to already-exploding rhyolite will produce lots of very fine ash, capable of being carried vast distances. This is bad news for aeroplanes. It also shows that the water will make the ash sticky, so the grains clump together and fall to the ground near the volcano. This is good news for the aeroplanes. Including this information in the computer models that predict where ash clouds will go will make their results more accurate.

The new evidence that gas is the main control on explosivity is probably also good news for aeroplanes. It means that subglacial rhyolite eruptions are much like rhyolite eruptions elsewhere. These produce most ash right at the start of an eruption, probably in less than 48 hours. Although the initial disruption may be large, the chances that it lasts more than a few days are small. This is similar to the Puyehue eruption in Chile, whose initial explosive phase produced an ash cloud that went round the world, but quickly settled down to many months of lesser activity.


  • Owen J, Tuffen H, McGarvie DW (2012) Explosive subglacial rhyolitic eruptions in Iceland are fuelled by high magmatic H2O and closed-system degassing. Geology. doi: 10.1130/G33647.1

Follow the authors

Jaqui Owen (@JaquelineOwen), Hugh Tuffen (@HTuffen) and Dave McGarvie (@subglacial) are all on Twitter. Follow them to find out more about rhyolite eruptions, with or without ice, in Iceland and Chile.

Further reading

To learn more subglacial rhyolite, check out these totally awesome papers. ;)

Pumice, ash and lava deposits reveal the mechanisms of explosive subglacial eruptions:

  • Stevenson, J. A., J. S. Gilbert, D. W. McGarvie, and J. L. Smellie (2011), Explosive rhyolite tuya formation: Classic examples from Kerlingarfjöll, Iceland. Quaternary Science Reviews, 30(1-2), 192-209, doi:10.1016/j.quascirev.2010.10.011.

Effusive rhyolite eruption beneath thick ice produces subaqueous-looking deposits:

  • McGarvie, D., J. A. Stevenson, R. Burgess, H. Tuffen, and A. Tindle (2007), Volcano-ice interactions at Prestahnúkur, Iceland: Rhyolite eruption during the last interglacial-glacial transition, Annals of Glaciology, 45, 38-47, doi:10.3189/172756407782282453.

Subglacial dacite and andesite produce pillows and hyaloclastite if the ice is thick enough:

  • Stevenson, J. A., J. L. Smellie, D. W. McGarvie, J. S. Gilbert, and B. I. Cameron (2009), Subglacial intermediate volcanism at Kerlingarfjöll, Iceland: Magma-water interactions beneath thick ice, Journal of Volcanology and Geothermal Research, 185(4), 337-351, doi:10.1016/j.jvolgeores.2008.12.016.

Iceland’s largest volcano records ice conditions during eruptions of different magma types:

  • Stevenson, J. A., D. McGarvie, J. Smellie, and J. Gilbert (2006), Subglacial and ice-contact volcanism at the Öræfajökull stratovolcano, Iceland, Bulletin of Volcanology, 68(7-8), 737-752, doi:10.1007/s00445-005-0047-0.
Categories: Uncategorized

UK Environment Advisor’s talk on climate change

If you are in any doubt that climate change is the biggest issue of our time, then I highly recommend watching the talk given by Prof Robert Watson, the former Chief Scientific Advisor to the UK’s Department for Environment, Food and Rural Affairs (DEFRA), at the American Geophysical Union (AGU) conference last week.

The talk is not too technical, so is understandable to a wide audience.  In it, he summarises the latest scientific data on climate change.  He covers current trends, future projections and the ways that we must adapt to limit its effects.  The outlook is grim.  The main points are that global temperatures will almost certainly rise by more than 2°C and so we must plan for increased droughts and storms, sea level rises, reduced crop yields, displaced people and widespread extinction of many species.  The way we live is going to look very different in 30-50 years time.

The full video is 1 hour long.  You can watch it below, or on the AGU website, where there is discussion on the comments.  If you have less time, I have made links that skip to the various sections (see below).  Of these, the two that you must watch are the sections on observations and projected changes (less than 7 minutes; 05:00-11:22) and on the 2°C limit and how we will miss it (5 and a half minutes; 30:05-35:22).

Academics see talks like this all the time, but for non-academics watching the video, I hope that you will also take away the following points:

  • This is a scientist talking to a load of other scientists.  There is no hype, no dramatic music, and no cute baby polar bears.  There are only data, graphs and trying to understand what they all mean.
  • There has been a huge amount of research into climate change.  Each of those graphs represents months, years, or decades of painstaking work.  The headlines that make the mainstream news gloss over how incredibly detailed our knowledge of current and past climates is.
  • Where there is uncertainty in our understanding, it is acknowledged and discussed.  We do not know if the Earth will heat by 3°C or 5°C.  But there is very little doubt that man’s activities are the main drivers in the current changes.


Some technical terms are used in the talk.  You can look most of them up on Wikipedia or Google.  The following definitions may help understanding the two key sections of the video.

  • IPCC – The Intergovernmental Panel on Climate Change is a body set up by the United Nations to create reports summarising our latest knowledge on climate change.  The last report came out in 2007, the next one is due in 2014.
  • Scenarios – The climate of the future depends on changes in economies and populations.  We do not know what those will be.  So the IPCC reports include estimates for a number of possible scenarios and compare the differences.  Those beginning with A represent more human activity and pollution.
  • Peer review – The process where studies are sent to other scientists before publication.  They check that methods were good and that the conclusions are correct.
  • Aerosols – Tiny particles floating in the air.  These are pollution, and cause acid rain and breathing problems, so we are trying to get rid of them.  But they also reflect sunlight back into space and have kept the climate of the last century lower than it would have been.  Temperatures will rise as we clean them up.

Edit 04 Feb 2013:  if you found this video interesting/useful, here are links to two others.


    * These are the key sections to watch if you have limited time.

    Thanks to Simon Carn (@simoncarn) for tweeting the link to the talk.

    Categories: Uncategorized

    Easily change coordinate projection systems in Python with pyproj

    Python is an easy-to-use programming language which, thanks to a growing number of cool extension modules, is really taking off in the world of scientific data handling.  The Proj4 libraries are a set of programs for performing coordinate system transformations.  Both are open source, so you are free to install them on as many computers as you want and to share them with your friends.  I had been using both for a while but only recently discovered the pyproj module that performs coordinate transformations inside Python itself.  It makes life very easy.

    The easiest way to get pyproj is as part of the Matplotlib Basemap package.  On Ubuntu Linux, this can be installed with the command: sudo apt-get install python-mpltoolkits.basemap

    1.) Setting up coordinate systems

    The first step is to define pyproj ‘objects’ to represent the coordinate systems that you want to use.  These can be defined using the Proj notation (see Proj4 documentation for details) but it is easier to set up commonly-used projections by referring to their standard code numbers.  These are called EPSG codes and can be looked up on

    >>> import mpl_toolkits.basemap.pyproj as pyproj # Import the pyproj module
    >>> # Define a projection with Proj4 notation, in this case an Icelandic grid
    >>> isn2004=pyproj.Proj("+proj=lcc +lat_1=64.25 +lat_2=65.75 +lat_0=65 +lon_0=-19 +x_0=1700000 +y_0=300000 +no_defs +a=6378137 +rf=298.257222101 +to_meter=1")
    >>> # Define some common projections using EPSG codes
    >>> wgs84=pyproj.Proj("+init=EPSG:4326") # LatLon with WGS84 datum used by GPS units and Google Earth
    >>> osgb36=pyproj.Proj("+init=EPSG:27700") # UK Ordnance Survey, 1936 datum
    >>> UTM26N=pyproj.Proj("+init=EPSG:32626") # UTM coords, zone 26N, WGS84 datum
    >>> UTM27N=pyproj.Proj("+init=EPSG:32627") # UTM coords, zone 27N, WGS84 datum
    >>> UTM28N=pyproj.Proj("+init=EPSG:32628") # ... you get the picture

    2.) Forward and inverse transformations

    Now you can convert geographic coordinates into your chosen projected coordinate system.

    >>> # Do the projection
    >>> isn2004(-19.700,63.983)
    (1665725.2429655411, 186813.38847515779)

    Obviously, you want to capture the output:

    >>> # Assign output to variables x and y
    >>> x, y = isn2004(-19.700,63.983)
    >>> x
    >>> y

    You can also do the inverse transform:

    >>> isn2004(x, y, inverse=True)
    (-19.699999999999999, 63.982999999999564)

    3.) Changing between different coordinate systems

    In most cases, you will want to change between coordinate systems.  This is even the case with GPS or GoogleEarth data, which use the specific WGS84 datum.  Coordinate system changes are done with the transform function.

    >>> # Convert x, y from isn2004 to UTM27N
    >>> pyproj.transform(isn2004, UTM27N, x, y)
    (563621.09135458851, 7095768.4648448965)

    And when you have lots of data, transformations can be done with lists/tuples/arrays:

    >>> lon = [-19.5, -19.7, -19.9]
    >>> lat = [63.183, 63.583, 63.983]
    >>> xx, yy = pyproj.transform(wgs84, isn2004, lon, lat)
    >>> xx
    [1674812.314071126, 1665231.4455360402, 1655933.043340466]
    >>> yy
    [97526.592642124306, 142220.30636996412, 186937.31348842022]

    It’s a simple as that.

    Pyproj can be installed with a single command on recent Linux systems (sudo apt-get install python-pyproj).  For other systems, check out the pyproj website.  If you are playing with coordinate transforms, then it is likely that at some point you are going to want plot stuff on a map.  Python can do maps, too;  check out the rest of the Matplotlib Basemap module.

    British National Grid and the OSGB36 datum.

    People working in the UK may have to go through another step to convert their data into the British National Grid format (BNG), which uses two-letter codes to define 100km-wide square regions instead of presenting the full 13-figure coordinates.  Thus Arthur’s Seat, an extinct volcano in the centre of Edinburgh, has a BNG grid reference of NT2755072950.  Below is a Python module that carries out the second step for you. Save the code as ‘’ in a directory where Python can find it.

    Note:  The script below was updated to fix a rounding error on 4 October 2013.  If you downloaded the code before then, you should replace it with the updated version below.

    The function converts between a BNG grid reference as a string, and a tuple of OSGB36 (x,y) coordinates. When converting to BNG coordinates, there is an opportunity to specify how many figures to use.

    BNG coordinates can be converted to GPS coordinates as follows:

    >>> import BNG # import the BNG module
    >>> BNG.to_osgb36('NT2755072950')
    (327550, 672950)

    Thus, you can get the GPS coordinates for Arthur’s Seat as follows:

    >>> x, y = BNG.to_osgb36('NT2755072950')
    >>> pyproj.transform(osgb36, wgs84, x, y)
    (-3.1615548588213667, 55.944109545140932)

    Use Python’s zip function handle multiple values:

    >>> gridrefs = ['HU431392', 'SJ637560', 'TV374354']
    >>> xy = BNG.to_osgb36(gridrefs)
    >>> x, y = zip(*xy)
    >>> x
    (443100, 363700, 537400)
    >>> y
    (1139200, 356000, 35400)

    You can convert OSGB36 coordinates to BNG coordinates like this:

    >>> BNG.from_osgb36((327550, 672950))

    Again, use the zip function for multiple values. You can also specify the number of figures:

    >>> x = [443143, 363723, 537395]
    >>> y = [1139158, 356004, 35394]
    >>> xy = zip(x, y)
    >>> BNG.from_osgb36(xy, nDigits=4)
    ['HU4339', 'SJ6456', 'TV3735']

    Note:  the coordinate transform between WGS84 and OSGB36 is complicated by some distortions in the British National Grid.  This is mainly important for high-precision work such as surveying or construction, and is described by the National Grid Transformation, OSTN02. To get the most accurate results, give Proj a datum shift grid file that describes the offsets.

    Happy mapping!

    Python module:

    #!/usr/bin/env python
    # Filename:
    #  COPYRIGHT:  (C) 2012 John A Stevenson / @volcan01010
    #			Magnus Hagdorn
    #  WEBSITE:
    #  This program is free software; you can redistribute it and/or modify
    #  it under the terms of the GNU General Public License as published by
    #  the Free Software Foundation; either version 3 of the License, or
    #  (at your option) any later version.
    #  This program is distributed in the hope that it will be useful,
    #  but WITHOUT ANY WARRANTY; without even the implied warranty of
    #  GNU General Public License for more details.
    __all__ = ['to_osgb36', 'from_osgb36']
        import numpy as np
    except ImportError:
        print "Numpy not installed.  Numpy comes with most scientific python packages."
    import re
    # Region codes for 100 km grid squares.
    # Reshuffle so indices correspond to offsets
    _regions=np.array( [ _regions[x] for x in range(12,-1,-1) ] )
    def to_osgb36(coords):
        """Reformat British National Grid references to OSGB36 numeric coordinates.
        Grid references can be 4, 6, 8 or 10 figures.  Returns a tuple of x, y.
        Single value
        >>> to_osgb36('NT2755072950')
        (327550, 672950)
        For multiple values, use the zip function
        >>> gridrefs = ['HU431392', 'SJ637560', 'TV374354']
        >>> xy=to_osgb36(gridrefs)
        >>> x, y = zip(*xy)
        >>> x
        (443100, 363700, 537400)
        >>> y
        (1139200, 356000, 35400)
        # Check for individual coord, or list, tuple or array of coords
        if type(coords)==list:
            return [to_osgb36(c) for c in coords]
        elif type(coords)==tuple:
            return tuple([to_osgb36(c) for c in coords])
        elif type(coords)==type(np.array('string')):
            return np.array([ to_osgb36(str(c))  for c in list(coords) ])
        # Input is grid reference...
        elif type(coords)==str and re.match(r'^[A-Za-z]{2}(\d{6}|\d{8}|\d{10})$', coords):
            x_box, y_box = np.where(_regions==region)
            try: # Catch bad region codes
                x_offset = 100000 * x_box[0] # Convert index in 'regions' to offset
                y_offset = 100000 * y_box[0]
            except IndexError:
                raise ValueError('Invalid 100km grid square code')
            nDigits = (len(coords)-2)/2
            factor = 10**(5-nDigits)
            x,y = (int(coords[2:2+nDigits])*factor + x_offset,
                   int(coords[2+nDigits:2+2*nDigits])*factor + y_offset)
            return x, y
        # Catch invalid input
            raise TypeError('Valid inputs are 6,8 or 10-fig references as strings e.g. "NN123321", or lists/tuples/arrays of strings.')
    def from_osgb36(coords, nDigits=6):
        """Reformat OSGB36 numeric coordinates to British National Grid references.
        Grid references can be 4, 6, 8 or 10 fig, specified by the nDigits keyword.
        Single value
        >>> from_osgb36((327550, 672950))
        For multiple values, use the zip function
        >>> x = [443143, 363723, 537395]
        >>> y = [1139158, 356004, 35394]
        >>> xy = zip(x, y)
        >>> from_osgb36(xy, nDigits=4)
        ['HU4339', 'SJ6456', 'TV3735']
        if (type(coords)==list):
            return [from_osgb36(c, nDigits=nDigits) for c in coords]
        # Input is a tuple of numeric coordinates...
        elif type(coords)==tuple:
            x, y = coords
            x_box=np.floor(x/100000.0)  # Convert offset to index in 'regions'
            try: # Catch coordinates outside the region
                region=_regions[x_box, y_box]
            except IndexError:
                raise ValueError('Coordinate location outside UK region')
        # Format the output based on nDigits
            formats={4:'%s%02i%02i', 6:'%s%03i%03i', 8:'%s%04i%04i', 10:'%s%05i%05i'}
            factors={4:1000.0, 6:100.0, 8:10.0, 10:1.0}
            try: # Catch bad number of figures
                coords=formats[nDigits] % (region, np.floor((x-x_offset)/factors[nDigits]), np.floor((y-y_offset)/factors[nDigits]))
            except KeyError:
                raise ValueError('Valid inputs for nDigits are 4, 6, 8 or 10')
            return coords
        # Catch invalid input
            raise TypeError('Valid inputs are x, y tuple e.g. (651409, 313177)')
    Categories: Uncategorized

    A visual estimate of the proportions of mixtures: pumice vs. lithics

    When a volcano erupts explosively, the tephra that comes out is a mixture of material that was molten at the time and bits of other old, cold rock that happened to get caught up in the blast.  These are referred to as the juvenile and the lithic components and consist of bubbly, pumice-like fragments and of chunks of lava and hydrothermally-altered material, respectively.

    Lithic-rich deposits from the explosive Hekla 4 eruption. The yellow-grey-white grains are juvenile pumice, the dark grains are lava lithics that were caught up in the eruption. I reckon that there are up to 40% lithics here. How about you? Click to enlarge.

    The relative proportions of juvenile and lithic grains can tell us something about the eruption.  For example, if the deposits are full of lithics, it might indicate that the vent where the material was erupting was getting wider or that underground water was coming into contact with the magma and turning explosively into steam, breaking up the surrounding rocks in the process.

    The best way to measure the relative proportions of juvenile and lithic grains in a sample is to get out the tweezers and separate them by hand, but this is often impractical, and a lot of time a visual estimate is sufficient.  Visual estimates are a lot better if you have some known values to compare them with.  That’s where the chart comes in.

    Visual estimates of the proportions of mixtures

    The following chart gives examples of what mixtures of different proportions look like.  Download it, print copies, use it whenever you need to make visual estimates of proportions of mixed materials.

    A chart demonstrating the visual appearance of mixtures of different proportions. Click to see the full-size version, or use 'Right-click, Save As...' to download a copy.

    Make your own charts

    The charts were made using the Python programming language.  The source code is given below so that you can make and customise your own plots.  If you are learning Python, try messing around with the code.  First try changing the numbers and sizes of grains; next change the number of plots and give the grey grains different shapes to the black ones; finally, try plotting mixtures of 3 different grain types, each with different size distributions.  Happy hacking!

    import numpy as np
    import matplotlib.pyplot as plt
    x = np.random.rand( number )
    y = np.random.rand( number )
    percentages=[1, 2, 5, 10, 20, 30, 50, 75]
    fig = plt.figure(figsize=(8.27,11.69))
    for i in range(8):
        black = np.round( percent*(number/100.0) )
        color = np.array( black*['black'] + (number-black)*['lightgrey'] )
        np.random.shuffle(color) #randomise so black aren't all at the bottom
        # define a marker shape here to make something that looks more like rocks
        plt.scatter(x, y, color=color, s=60, linewidth=0.25, edgecolors='black',
        plt.title('%i%% black' % (percent), fontsize=12)
    plt.subplots_adjust(left=0.05, right=0.95, bottom=0.05, hspace=0.15, wspace=0.1)
    plt.suptitle('Visual estimates of proportions of mixtures', fontsize=18, y=0.97)
                 x=0.95, y=0.045, horizontalalignment='right',
                 verticalalignment='top', fontsize='x-small')
    plt.savefig('PumiceLithicsProportions.png', dpi=150)
    Categories: Uncategorized