Digital Geology

[bjs]

I'm a retired Earth Scientist and I now work with amateur geologists and OU students trying to get them going on things like geological mapping. The way professionals, like the British Geological Survey, work in the field these days is using ruggedized tablet computers and commercial software. Unfortunately that lot costs around £20,000. A year or so ago MIT? Had a project to make a low cost instrument and got the cost down to ca. £1000. So I have it in mind to try to use the RPi to make a really low cost version, say £100, which would be open source hardware and software and could be built at home.



This is quite a big project but it breaks down into a number of smaller ones which would be of much wider interest to the RPi community. A few of the hardware bits are:





interface RPi with a GPS;



interface RPi with a digital compass;



interface RPi with an accelerometer;



bluetooth data transfer to PC; and,



interface RPi with a small LCD? Screen.





On the software side:





drivers (if necessary);



database for storing field measurements;



GIS front end; and,



data entry screens for section logging etc.





There is already an Android port of the gvSig GIS software to mobile phones (Java source) so it should not be too difficult to port it to the RPi. Similarly most of the hardware steps have already been done for Arduino boards so there is a good starting point there too. I think I will probably start with the Gert board to experiment with the interfacing and software parts but eventually I will have to think about reducing the size so that it is contained in a single hand held waterproof package. That will need quite a bit of engineering input but its a long way off.



I have done a good deal of electronics and programming in the past but I'm now very rusty. I'm just starting to try to learn python which I think is a good easy to read language. My overall aim is to produce a tool that students can use in the field but can also understand how it works and hopefully extend its capabilities. It may also be of value to people in the 3rd World.



I have an RPi on order from Farnell who tell me I will have it by the end of April. Well, I'll believe that when it happens, and since I will be away in Scotland leading a field trip they have some slack before I start looking for it! My skills and resources are limited so I would welcome suggestions, help and collaborations of any sort.

[jamesh]

Bluetooth and GPS should be no problem if you are OK with USB dongles - bluetooth we know works, and I almost got GPS working some time ago, but ran out of time. Bluetooth dongle was about £5, and the GPS dongle was £16.

[winwaed]

I think such a system is going to be memory limited - but out in the field, you probably only need a few tens of square miles of data.

Do you need a bluetooth connection? Keeping costs down, a wire connection to a PC would be cheaper.

The Raspi's very low battery requirements are of course a good thing (as a student I remember lugging LaCoste-Romberg gravimeters over hills - I'm sure they designed the cases to make them as hard to hike with as possible , and about half of the weight was the battery)

I don't know if you've seen the article in the March GeoScientist but there's an item about modern 3d mapping/visualisation by the BGS under provocative titles about the 'end' of geological mapping. Is this something that could be leveraged - in a teaching tool perhaps? The RasPi's GPU is meant to be pretty good...

[bjs]

JamesH said:


Bluetooth and GPS should be no problem if you are OK with USB dongles - bluetooth we know works, and I almost got GPS working some time ago, but ran out of time. Bluetooth dongle was about £5, and the GPS dongle was £16.



I am ashamed to admit that I had not noticed that GPS dongles were available! I now have one and I'm going to get it going on my Ubuntu laptop in the near future. This actually makes things very much easier for a first pass since I probably do not need the USB for anything else. So, many thanks indeed!!!!!!

The blue tooth is a way down the line at the moment but its nice to know that there are cheap solutions.

I'm getting hungry for my raspberry Pi.

[winwaed]

Yes the GPS USB devices are readily available. My day job is maps/navigation: I would probably recommend against the "USB pen drive" type. They look tempting (small, etc) but we've found their reception is not as good due to interference from the host laptop. Some have reported using a USB extension of a couple of feet can cure a lot of these problems for in-car applications.

I've had a lot of success with the Pharos BU-353. This is a magnetic puck that is about 2in diameter with about 1m of wire and a standard USB connector. It even works indoors, which is good for testing!

I don't know about Linux drivers. The manufacturer includes Windows drivers that make it look like a standard NMEA-0183 device (as do all manufacturers - the standard specifies an RS-232, but everyone uses USB, so a software driver is invariably required).

[bjs]

Thank you for your helpful thoughts.

GIS systems are certainly memory hogs but that's way down the line at the moment.

What I have in mind is mapping an area of about 1km x 1km which is about the amount of work that an average student can manage in a weekend. So we are talking about 250 locations per day which is not an excessive data rate!

What I have in mind, for a kick off, is to be able to log position by GPS and then enter geological data manually using simple data input screens. Lithologies, sedimentary structures, structural data, fossils, mineral assemblages and bed attitudes are all finite sets so it should be easy to programme them using check boxes etc. Measurement of attitudes/foliations/lineations etc can be added when an accelerometer has been added. It would be nice to include georeference photos too but memory may then be a limit.

Similarly, I agree that blue tooth is not a priority. But looking into the future with G4 technology, it is possible that we could have a class of students each mapping a 1km2 square but all maps being available to all students in real time as the mapping progresses. This has already been demonstrated for disease control.

I have seen the GeoScientist article and I agree that 3D mapping is the future. I have also been to various meetings with BGS and seen their work so far. I am very impressed but its not  the kind of thing that junior undergraduates/amateurs are likely to get involved in in the near future. At the moment the issue is in ensuring that all students can make use of modern 2D, traditional mapping and learn how to use the data they collect. There is a lot of very  expensive, although very good, technology available. My concern is to make this kind of thing available to people who do not have the financial resources to make use of the commercial options.

At the end of the day its the observations that the students make that matters. Learning to be a good observer is far more difficult than is often imagined. The object of this project is to remove the distraction of finding out where you are so that students concentrate on the rocks.

I hope that does not sound too negative since I do appreciate your input and its good to have to defend ones position. I'm thinking this all through as I go along but I think that the RPi is going to shake the Earth! Rome was not built in a day. I think the way forward is by making small steps, learning from mistakes and shearing with others for feedback and inspiration.

Thank you for your input

[winwaed]

No not negative at all - I didn't know how far you were and so I was throwing ideas in!

I agree about the importance of observation. And it is all well and good not learning about shading maps (cf. that article), but I think it is still very valuable to be able to visualise the 3d structure from a map or standing in the landscape.

Overall your proposal is very similar to the Trimble Field Computer that my wife and I use for ecology field work (she's a biology prof, and I provide technical support on an annual Costa Rica field trip with students). The Trimble's GPS is going to be more accurate BUT it costs about 10x as much. It was a big investment for her department at the time. A lot of classes will find $100-200 devices affordable - multiple $2000 ones are not!

Could a USB flash drive be added to hold photos? If my DSLR is anything to go by, then even a cheap 2GB drive should go a long way.

I'm not sure how affordable good accelerometers are, but the one in my iPad is pretty bad. I actually have a compass-clino app on it, and to say it is useless would be the polite way of putting it!

Anyway it is an interesting project - keep us posted, and let me know if I can help (I'm more software and maps these days but have two geol degrees under my name and a member of GeolSoc of course).

Other interesting threads on this site also discuss making a seismometer and weather stations - these would be targeted more at high schools / home science though.

[wrhii]

Don't most (hand held) GPS units have some sort of accelerometer included?  i.e. they know direction + speed + altitude +?...

Point being, maybe you will get the compass / accelerometer info from the GPS dongle?  Or is it only able to supply a coordinate?

[error404]

Most do not include an actual compass, they infer a heading from movement. Altitude is provided in the 3D GPS fix (though not very precise).

It seems to me that an inexpensive Android tablet and some software can do everything you're looking for aside from be weatherproof, which you might be able to work around.

[gimliflea]

Bluetooth range is very limited ~100 metres max iirc. So would be of limited use in the field. Otherwise it all sounds doable and many others on the forum are wanting to solve most of these problems. True ruggidizadation  is more than just making it waterproof.

However the Rpi with its low component count should be fairly robust. Connectors are usually the weak point. However Sugru could be you friend there.

[SN]

error404 said:


Most do not include an actual compass, they infer a heading from movement. Altitude is provided in the 3D GPS fix (though not very precise).

It seems to me that an inexpensive Android tablet and some software can do everything you're looking for aside from be weatherproof, which you might be able to work around.


Whilst the project is a good one, any recent Android phone with the £7 app MMTracker Pro and some Memory Map QCT files will do this job beautifully - I have this installed on both my Galaxy S II and Galaxy Tab 7 for use in my car during Off Roading events - the qct files are Standard Ordinance Survey so are perfect for this.

[RMW5]

Out of interest, why do you need the accelerometer and the digital compass? I assume that you want an accurate (1 degree) non-GPS compass.  Between them they would take out $50 of your $100 budget and the USB/wired GPS unit would probably be another $30-40.  A small HDMI touchscreen would be another £150-200. And you need to think about your portable power supply.

Seems to me that an android tablet would probably be a better choice, particularly if you use an external GPS and/or other sensors that the tablet doesn't feature.

[tech_monkey]

@bjs Would you also be looking at basic seismic survey work as well or just surface features.

Very basic seismic might be possible but would certainly need an add on analogue to digital converter board with selectable sampling from about 0.5 to 4ms.

There are loads of open source geophysical and GIS software available with the majority working on linux or windows platforms or both.

If its just GPS mapping then I see no reason why it can't be done. But when a cheap android tablet with at least GPS costs 150 pounds it may be hard to build something for less.

I have a Flytouch3 Android tablet with GPS (cost 150 pounds) and it uses an external antenna so it could be mounted on a pole.

would something like Merkaartor be suitable for the software, don't think this would run on a PI, but data collected from in field PIs could then be downloaded to a laptop.

[tech_monkey]

Found a low cost DGPS solution that may be of interest to you.

http://www.wombat.ie/gps/

Though I reckon you could have a PI logging GPS data for a particular point during the survey time take an average and then compare the average to the current reading then work out the difference. Then use this difference as a correction to the surveyed GPS data.

[bjs]

@Winwaed

@tech_Monkey

It occurred to me this morning that what I am proposing is really a generic Portable-Pie that can be used for any kind of field logging application with appropriate sensors. So it could be used for biology or hydrology with a pH or ion sensors or for geophysics. I have been thinking of a simple proton magnetometer for detecting dykes etc since its a fairly easy build. Seismic could be fun too and the archaeologists use a whole range of geophysics these days.

[bjs]

@Winwaed

@error404

@RMW5

Android Tablets etc.

I thought about this some time ago. My Android htc hero has compass, clinometer, gps and camera apps but they don't seem to be very good. I don't know if this is due to poor sensors or poor software. Its just a black box and is not open source so I can't do much with it. I could of course write my own apps to test but my feeling was that since the Android shell is so limited, just java, I would not be able to reuse any of the good code like gpsd that already exists. A regular Linux distro. running on a smart phone would be a different matter but still quite expensive. I could be wrong about this.

Quite apart from that, I want to play with my RPi!!!!!!!!!!!

[bjs]

RMW5 said:


"Out of interest, why do you need the accelerometer and the digital compass? I assume that you want an accurate (1 degree) non-GPS compass.  Between them they would take out $50 of your $100 budget and the USB/wired GPS unit would probably be another $30-40.  A small HDMI touchscreen would be another £150-200. And you need to think about your portable power supply."

As far as price is concerned I think you are right. The display is the main problem but that is going to be common to any portable RPi  project and for things like the £50 computer for kids that people keep talking about. For this present project I"m thinking or starting with something very simple like the old kind of mobile phone display. They are cheap but very limited. If I only had a GPS on board to start with I would need to add data manually. I think that can be done using a couple of buttons and scrolling menus? Very tedious and it rules out the use of maps/GIS . The BGS product, Sigma Mobile, uses a high end military grade tablet (ca. £8,000). It does not have sensors like the clinometer and that sort of thing has to be entered by hand. When they started using it I"m told that they found it difficult if not impossible to use when it was raining or in very bright sunlight since they could not read the screen.

As for why accelerometers etc, the main things that geologists measure in the field are things like the dip and strike of beds and other geological structures. You need a compass to get the direction of dip or strike and an accelerometer to measure the inclination or dip. GPS will only give you direction when you are moving.

BTW I also need to make this device quite small so that I can orient it with small geological structures; that tends to rule out tablets. One option I have considered is making a small, hand held, device with the sensors and connecting it with bluetooth to a belt mounted computer or tablet for data storage and analysis. Another problem with mobile phones etc is that they tend to have curved cases which can"t be aligned easily with structures; I need a small square sided box.

At the moment my £100 target seems unrealistic but I"ll bet the first RPi cost more than £22!

***************************************************************

Sorry, I don"t seem to be able to get this bock quote to work properly.

[RMW5]

@bjs

A few years, I would have said that the solution would have been to build your small handheld sensor device and link it via bluetooth to an iPaq PDA or similar with a 3x5 inch screen and a few gigabytes of storage on an SD card, but the PDA alone would have set you back £300.  Right now, I would substitute a small Android tablet, not some 10 inch iPad wannabe, but with an adequate screen and battery life.  There are Linux tablets, but I don't recall seeing any with small screens.

In order to get this to work in the field, would it be practical to drop the requirement for a display and rely mostly on audio and a simple LED based output? Likewise, if you could simplify your input so that it could work from a numeric keypad, then you can hook that up via USB for less than £10 and have a practical handheld controller - Google: "Targus Numeric Wired Keypad".

One way to get round the cost of a display would run your app on the R-Pi as a web server and let other devices (i.e. smartphones) nearby connect and view the data using their web browsers. Assuming all your students have smartphones, then you don't need to worry about the cost of the display.

[Andre_P]

@Winwaed

A LONG time ago I was involved with a project that connected to a GPS via a RS232 cable. The was a standard using NEMA Sentences, I presume that kind of thing has become old hat now.

I kind of guess that with the RPi the primary cost will be the instrumentation.

The OTHER issue is going to be powering it all. The RPi itself is going to be pretty good from the hardware point of view.

However there will be the instrumentation and the display to drive.

Also to help things it might be an idea to write the software with keeping the power consumption as low as possible, putting the core to sleep and to wake on interrupt, keeping loops as tight as possible to keep them all in cache and stuff like that.

Sounds like an interesting project though and GOOD LUCK .

[bjs]

RMW5 said:


@bjs

A few years, I would have said that the solution would have been to build your small handheld sensor device and link it via bluetooth to an iPaq PDA or similar with a 3x5 inch screen and a few gigabytes of storage on an SD card, but the PDA alone would have set you back £300.  Right now, I would substitute a small Android tablet, not some 10 inch iPad wannabe, but with an adequate screen and battery life.  There are Linux tablets, but I don't recall seeing any with small screens.

In order to get this to work in the field, would it be practical to drop the requirement for a display and rely mostly on audio and a simple LED based output? Likewise, if you could simplify your input so that it could work from a numeric keypad, then you can hook that up via USB for less than £10 and have a practical handheld controller - Google: "Targus Numeric Wired Keypad".

One way to get round the cost of a display would run your app on the R-Pi as a web server and let other devices (i.e. smartphones) nearby connect and view the data using their web browsers. Assuming all your students have smartphones, then you don't need to worry about the cost of the display.





@RMW5

While out cycling it occurred to me to upgrade my Kindle to a colour version and use the old screen for this project. So I looked it up on Amazon and the colour version does not seem to be available in the UK yet. However, what it also found was a NATPC M009S 7inch, 2GB Android tablet with WiFi and a touch screen for £65!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Well it would be difficult to find any screen for that price so thanks a lot for pointing me towards these cheap tablets I had no idea the price had come down that far.

So 'Plan B', I can now start to think of using the RPi to control the instrumentation and a WiFi link to the Android to store and process the data. gvSig GIS has already been ported to the Android so I can use that plus some simple i/o screens etc. gvSig will already have a georefernced data base. This sounds doable. Again many thanks for the idea.

[RMW5]

If you develop your application so that it runs at home when connected to a monitor then you just need the tablet/R-Pi combination to run as a remote desktop client/server (i.e. your input screen is the (800x600?) Android screen rather than the HDMI screen you used at home), but all sensor data would be collected on the R-Pi.

[tech_monkey]

If you are after a cheap Android tablet, have a look here.

I got mine on ebay but be careful, many are being sold as a Flytouch2 or 3 or even 4 when in fact they are not.

[winwaed]

@Andre_P: No, that is still pretty much where things are. All the vendors still claim to support NMEA-0183 - even though their devices are usually USB. So for Windows PCs you typically need a vendor-supplied driver that makes makes the USB look like an NMEA COM Port!  Windows 7 introduced a new "Location & Sensor API" but I have yet to see a production application code directly against it - any commercial application still needs to support earlier versions of Windows.

I don't know the Linux situation, but suspect it is very similar. Perhaps the real choice will depend on which vendor supplies Linux drivers?

[winwaed]

@bjs I like the field idea. Affordable ion probes that don't require re-agents would be ideal for our Costa Rica stuff. We really don't want to take "chemicals" by plane/customs, and try to avoid expensive electronics if possible. We've used "garden" pH meters in the past - they're more likely to break, but it is no loss when they do.

For air/water temperature measurements, we/the ecologists use "HOBOs": http://www.onsetcomp.com/    More ecology/hydrology than geology as you deploy them and they record the temperature periodically over days/months. Simple micro-controller and penny battery, and the price is right (I think less than $50 each, and they're the kind of thing that can be bought a few at a time). A "RasPi HOBO" as it we, could do a lot more - e.g. support simple telemetry, multiple sensors, etc.

With my geological background and the fact we're about 7 miles from Volcan Arenal, I keep trying to think of geological things we can do, and there's not much. Deep soils, lots of clays. There's a pre-European ash layer that outcrops on the study property but in only one location.

If you think a magnetometer could be cheaply built, then that is one possibility, but I suspect we'll only pick up water pipes. Yes I'm sure the property is underlain by dykes, but then virtually the entire country is igneous

The geophysical technique the property is really crying out for is a Bouger inversion to measure the density of the hills. We could drag a physics major along, and do a really neat experiment due to the large elevation changes. Unfortunately a gravimeter is way out of the budget of a small science department.

The other affordable technique (and probably within reach of your RasPi system / our budget) would be a 4 probe resistance meter. We could take that and do some Schlumberger/Wiener networks, but the problem then becomes one of interpretation. I've got a nice wiggle on the graph, but what does it MEAN?

Sorry, I'm digressing.

[Andre_P]

Don't worry about digressing, if I may quote Professor Brian Cox

"It's Brilliant !"