Friday, 30 October 2015

A possible bug in QGIS 2.12

Working with my maps of a segmented DEM showing average slope direction and magnitude, and curvature with arrows, I noticed it broke once I upgraded to QGIS 2.12.
Unless there has been some change in the syntax of expressions that I didn't get from the QGIS 2.12 changelog I suspect a bug.

I made a minimal example, which I shared via GIS Stack Exchange and QGIS issue tracker.

QGIS 2.10.1. The width of the lines are proportional to the magnitude of the slope, and the colours are proportional to the 'longitudinal curvature' with blue for concave slopes and red for convex. No prizes for guessing the location since it's labelled on the maps at the links above.
QGIS 2.12. Styling is lost for the slope magnitude and longitudinal curvature, but slope direction is still shown.

Friday, 23 October 2015

Georeferencing a map from 1574

I saw the National Library of Wales had digitised a map of Wales produced circa 1574 by Humphrey Llwyd, and uploaded a copy to Wikimedia commons.

I decided to try georeferencing it to a modern map based on OpenStreetMap data using the QGIS georeferencer:

Overlaying 1574 map with transparency shows the variability in the accuracy of the map. full-size version

Removed modern town names to show 1574 map more clearly. full-size version

I chose a variety of places around Wales and the border areas shown on the map, including Cardiff, Newport, St. Davids, Aberystwyth, Holyhead, Liverpool, Shrewsbury, Hereford.

The result looks more accurate than I expected, with Aberystwyth and Shrewsbury being almost exactly referenced, and several other places with small errors.

According to the author Graham Robb, Whitchurch in Shropshire was the prime longitude meridian for the British Druids in the Iron Age, and is named as "Mediolanum" by Ptolemy in the 2nd century. Aberystwyth (or more exactly, the Pendinas hillfort) was also an important location within their navigation system apparently, being both on a longitude meridian and latitude paralell.

Update: QGIS layer blending

Changing the layer blending method to a "Hard light" overlay rather than semi-transparency makes for a slightly clearer result I think:
full-size version

Wednesday, 21 October 2015

The glacial forms of Greg Crater

As an example, in part of the HRSC tile h2224, Colin Souness in his PhD work found a number of glacial like forms in the northern wall of the crater.
Greg crater is located at 38.5°S, 113.1°E and has a diameter of 66km.

Download larger version
The southern wall of the crater tends to be covered in gullies. There is a detailed summary of it on the website of the Planetary Science Institute, Tucson, based on the W. Hartmann et al paper.

Now looking at the classifier, labelling extent areas above ln(K) > 10 and head areas above ln(K) > 12. I have used "Addition" blending in QGIS to show areas that are classified in both in purple.

Download larger version
There are potential glacier source areas above the Souness objects, but there are many other areas that are also marked out as source areas that do not have glacier like forms catalogued by Souness, such as lower down the north face, the east face and the south face of the central mountain.
Download larger version

Google Earth can be useful to visualise Mars, selecting the CTX image mosaic (the 'context' camera with 6m resolution from Mars Reconnaisance Orbiter):

It is also possible to see where HiRISE coverage exists (red outlines):

Tuesday, 20 October 2015

The Green, Green, QGIS of Mars (Part 2)

Here is the southern mid-latitudes of Mars, in the same colour scheme as the previous post showing the northern mid-latitudes:

The lowest elevation area of Mars, the Hellas basin has areas of bright terrain which makes the colour scheme a little confusing. Download larger version.

The Green, Green, QGIS of Mars

There are many blog posts on the Internet and even scientific papers on the evils of the spectral/rainbow colour palette for continuously varying data.

The series of 6 by Robert Simmon on NASA's Earth Observatory website is particularly worth reading and references many interesting and potentially useful sources.

Incidentally while looking for that, I saw this Image of the Day of a landslide onto a glacier in the Yukon in Canada. I wonder how often they have landslides on Mars.

Thus, I now show you the north of Mars, with the elevation encoded in a single hue palette and again overlaid with "Hard Light" on the Mars Global Surveyor image mosaic.

This will not be perfect for a quantitative visualisation, since it is not possible to distinguish between intrinsically darker areas and areas that are low in elevation.

I choose the green colour, because it leaves red and blue free to plot the classifier function values for areas similar to glacier head and extents.

The various individual HRSC tiles also have the same kind of colour palette of DTM elevation scaled locally overlaid on the nadir image HRSC. Download a larger image of this here.

If there's a North there must be a South

See also previous post for the North.

Here is the southern hemisphere of Mars. Again, I have used the same equirectangular coordinates optimised for 40° latitude.

The Souness et al. glacier like forms (such as QGIS plots them at this scale) are marked and numbered.

Monday, 19 October 2015

Lots of planets have a North

Here is the North of Mars, showing the Mars Express High Resolution Stereo Camera tiles used in my MSc dissertation, colour-coded by elevation, but also with the output of the Bayesian classifier measuring the similarity to Colin Souness's glacial form.

This is possibly a bit too colourful and the classifier output isn't really useful at this scale. The segmented DTM has red indicating similarity to glacier 'head' areas, green for 'context' areas (9x area of extent) and blue for extents.

I use QGIS to blend the elevation with an underlying nadir image also from HRSC with "Hard Light". I am not sure whether the developers of QGIS intended the Red Dwarf reference.

The coordinates are in an equirectangular projection in metres optimised for a standard latitude of 40°.

Tuesday, 13 October 2015

The Lost land of Lyonesse revisited

In a previous post, I mentioned the work of Shennan and Horton (2002) about sea level change around Britain through the Holocene.

I showed a visualisation of the West Cornwall coastline as it may have been around 12,000 years before present.

After downloading further bathymetry via the UKHO INSPIRE portal, I present an updated version covering the whole area around Cornwall.

Much of this came as .csv point data, which I had to grid, and use the QGIS Concave Hull plugin from the Processing toolbox to restrict the interpolated grid to where there was data.

In some areas, there are gaps in what is currently available at high resolution so it falls back to low-resolution data.

Based on my reading off the graphs in Shennan and Horton, the sea level rise at this time should be similar across the area covered here:
Site Number Site Name Lat Long RSL 1kyr RSL 2kyr RSL 3kyr RSL 4kyr RSL 5kyr RSL 6kyr RSL 7kyr RSL 8kyr RSL 9kyr RSL 10kyr RSL 11 kyr RSL 12 kyr

42 Pembroke 51.8 -5.1 -0.8 -1.6 -2.4 -3.2 -4 -9.2 -14.4 -19.6 -24.8 -30 -35.2 -40.4
43 Glamorgan 51.5 -3.7 -2 -4 -6 -8 -10 -13.6 -17.2 -20.8 -24.4 -28 -31.6 -35.2
51 Devon 50.4 -3.5 -1 -2 -3 -4 -5 -10 -15 -20 -25 -30 -35 -40
52 Cornwall 50.2 -5.5 -1.4 -2.8 -4.2 -5.6 -7 -11.6 -16.2 -20.8 -25.4 -30 -34.6 -39.2

It is clear there were substantial areas of land that were disappearing in the millenia after the reoccupation of Britain after the last glacial maximum (with this map corresponding to the cold spell of the Younger Dryas).
In the period of time, between 12kyr BP and 8kyr BP sea level rose by 20m, which corresponds to half a metre every century.
Especially in the large expanses of gently sloping terrain off the north coast, it would be likely that coastal retreat would have been noticeable by people within a single lifetime.

A version speculating about woodland colonising the lowlands:

Sunday, 11 October 2015

Ebrenn y'n Nos mis-Hedra 2015

This is a transcript of the October 2015 episode of Ebrenn y'n Nos (The Sky at Night in Cornish)  which was broadcast in #243 of Radyo an Gernewegva.

Dydh da. Ottomma rann nowydh kevres “Ebrenn y'n nos” mis-Hedra.
Dhe mis-Hedra yma an rannnevesow kynyav orth aga omdhiskwedhes.
11 y'n nos, yn kres mis-Hedra
 Onan an moyha kler yw an pedrek Pegasus. Pegasus yw rannneves bras ughel y'n ebrenn dres an kynyav. Ev yw margh-nija diworth henhwedhlow Grek. Nessa dhodho yw rannnves arall Andromeda. Y'n ranneves ma yma an Galaksi Andromeda (Messsier 31). M31 yw an galaksi meur ogassa dh'agan galaksi. Ev yw martesen nebes brassa es agan galaksi, hag a'n jeves galaksiow loor ow resegva a-dro dhodho. Yma M31 dihaval orth an brassa niver a galaksiow, drefenn hi dhe'gan nesa dhe'n kevradh 300km pub eylenn. Dell usys an galaksiow a wra mos dhe-ves diworthyn hag an ollvys dhe omlesa. Yn pymp bilvil blydhen ow tos ev a wra junya gans agan galaksi ha furvya unn galaksi veur. Na borth own drefenn yma an sterennow mar bell diworth aga honan ha'n galaksiow a wra mos dres aga honan, bys dhe'n dhew galaksiow dos ha bos onan. 

Diworth Adam Evans dre Wikimedia Commons
Yma galaksi bras yn resegva a-dro dhe M31, ogas dhe degves rann braster M31. Henn yw an galaksi M33, y'n ranneves Triangulum. M31 yw gweladow yn es dre dhewlagas nooth, ha da yw yn diwlagattell yn ebrenn tewl. M33 yw kalessa dhe weles, mes y hyll bos gwelys dre dewlagas nooth mars yw an ebrenn pur dewl, martesen mars os ta war Woon Bren po Lysardh. Gans pellwelerow pur vras kepar hag an Keck in Hawai'i y hyll bos gwelys an sterennow yn M31 ha M33, ha pella, kammnevesow anedha, dhe dhyski moy a-dro dhe'n galaksiow ma.

An loor: 

Kwartron diwettha yw an 4a mis-Hedra, loor nowydh yw 12ves mis-Hedra, kynsa kwartron vydh 20ves mis-Hedra, ha'n loor leun vydh dhe'n 27ves.


An planetys: 

Mergher yw gweladow yn soth-est y'n ebrenn bora, kyns an howldrevel, gans ystynnans moyha dhe'n West dhe'n 16ves mis-Hedra, mes ny vydh marnas nebes degreow a-vann an gorwel kyns an howldrevel. Gwener yw y'n ebrenn bora ynwedh, marthys ughella ha mos dh'y ystynnans moyha dhe'n West dhe'n 26ves mis-Hedra. Meurth yw yn ebrenn bora ynwedh, ow trehevel y'n Est a-dro dhe 4 eur y'n bora. Ev yw yn rann pell diworthyn yn y resegva ha ganso braster 1.8. Ev vydh marthys splanna Gwaynten nessa.
Planetys kyns an howldrevel (16ves mis-Hedra)

Yma New Horizons hwath ow tannvon imajys ha data diworth y dresnijans Pluton an hav ma. Yma Rosetta hwath ow studhya an komet 67P/Churyumov-Gerasimenko mes soweth ny wrug kestava gans an tirell Philae arta.

Yma China ow pesya hwilas an loor. I a vynn tira robot war denewen pell an loor, erbynn 2020 gans rosyell kepar hag a wrussons i tira Chang'e 3 ha'n rosyell Yutu war an denewen nes yn 2013. I a vynn dannvon robot a wra kemmer sampel ha'y dhehweles dhe'n norvys y'n termyn a dheu, martesen ow lonchya yn 2017.
Y fydh an nessa efanvos Europek dhe Veurth, ExoMars 2016, delatys gans 2 vis. Ev a wra lonchya yn mis-Meurth yn le mis-Genver 2016 ha vyajya 7 mis dhe Veurth, dhe studhya y ayrgylgh ha'n gewer gans an "Trace Gas Orbiter" (Resegvell Gas Tanow) ha tira an direll Schiaperelli.
An efanvos ma yw profyans yn hwir rag an efanvos ExoMars 2018, a wra tira rosyell war eneb Meurth.
Yma neppyth a-dro dhe Veurth diworth NASA, yn Kembrek. Yma gwiasva "Beautiful Mars" gans imajys diworth resegvell "Mars Reconnaisance Orbiter", ha neppyth y'ga hever treylys yn lies yeth, ow komprehendya Kembrek: po
Chons da kavos gewer da hag ebrenn kler. Bys nessa prys.

Friday, 9 October 2015

Probably not the Aurora Borealis

From time to time, the Aurora Borealis can be visible from the UK, and the Twitter accounts AuroraWatch UK and BGS Aurora Alert can be useful in getting warning of this. It is relatively rare that it is visible as far south as Cornwall, but I did have a look at around 11pm BST on 8th October and took some not very convincing pictures of the northern sky, where there was mist illuminated by artificial lights.

More pictures at my Google+ album.

Maps of Wales with slope arrows

Here are a couple of maps of Aberystwyth, and Cadair Idris, with the RSGISlib segmented digital elevation model.
Slope direction is shown by arrows, and coloured in the same scheme as previously with thicker arrows for steeper slopes, and longitudinal curvature indicated with convex slopes red and concave blue.

1:10000 seems about the limit of the smallest scale that QGIS will cope with the arrowed lines. Beyond that the computer just hangs - slope lines without arrows can be plotted without problem.

Aberystwyth town centre, and the university campus. full-res version

Cadair Idris. This is a textbook example of a glacial landscape in the British Isles. full-res version
Aberystwyth area and Rheidol Valley. full-res version

Dyfi Estuary showing Borth, Tywyn and Machynlleth. full-res version
To readers: are there any requests of areas you would like to see?

The processed data currently includes the following area though I also have the whole of British Isles (except most of Shetland) DEM from 1 arcsec SRTM (and could download anywhere up to 60° latitude) but not processed to create slope curvature layers and segmented.

Pen-y-fan area, Brecon Beacons

Pen-y-fan main summit to the right of centre, showing the slopes convex at the top, but generally concave further down. full-res version
Broader context at 1:40000. full-res version

Thursday, 8 October 2015

A greyscale version of my maps in Cornish

I have made a greyscale version of the maps in Cornish I have been working on.

Here is an extract from the Truro area:

Cornish placenames from MAGA Placename and Signage Panel and the book by Craig Weatherhill. Some names of geographic features are my own translations of English names. Any errors are my own responsibility. large version

Adding arrows to indicate the (downward) slope direction using the segmentation method described in earlier posts. large version
Keeping the map greyscale but showing longitudinal curvature of the slope with convex slopes in red and concave in blue. large version
Penzance. Unfortunately using the arrows tends to make QGIS run very slowly and crash for larger maps. larger version

Saturday, 3 October 2015

Akademi Kernewek

Yth esen vy yn kuntelles an myttin ma, avel 'Forum Yeth Kernewek - Cornish Language Forum' ow keskewsel gans pobel diworth Konsel Kernow, eseli bagaosow yeth ha kernewegoryon erell, a-dro dhe towlennow chanjya furv a wra Konsel Kernow owth oberi a-rag Kernewek.

Nyns o pubonan unnverhes an chanj yw da, nebes a leveris y fia gwell kavoes arghasans hir-dermyn hag ena gorra framweyth nowydh, wosa chanjya an framweyth kyns ambos anodho, mes ny vynnav skrifa meur a'n mater na.

Yma towlenn sevel 'Akademi Kernewek' dhe pesya an ober gans gerlyver FSS, henwyn leow, h.e.

Yma nebes Akademiow Yeth a-dro dhe'n bys, rol diworth Wikipedia. An huni yn Latvia a wrug seulabrys dewis ger rag 'hipster'.

Yma Akademi Yeth yn Pow Chekk, an brassa rann a'ga gwiasva yw yn Chekk mes i a wra gwiasva rag gerlyver ha gidyans gramasek ha taklow an par na.