Support the Arctic Sea Ice Forum and Blog

Author Topic: Test space  (Read 20074 times)

subgeometer

  • Frazil ice
  • Posts: 362
    • View Profile
    • All in the Name of Liberty
  • Liked: 114
  • Likes Given: 71
Re: Test space
« Reply #150 on: September 24, 2019, 05:54:13 AM »
Nares Strait and Lincoln Sea over the 2019 melt season, encoded in the same batch as the "Beaufort Beard" animation above. Its about 3 times the size of the h265 version, though quality also seems a bit better. H265 outperforms on the bitrate setting while libx264 produces output a bit larger than requested, at least with these ice animations

subgeometer

  • Frazil ice
  • Posts: 362
    • View Profile
    • All in the Name of Liberty
  • Liked: 114
  • Likes Given: 71
Re: Test space
« Reply #151 on: September 24, 2019, 08:27:45 AM »
These are a more reasonable size

The whole arctic basin through the course of the freezing season at 5.4MB by constraining bitrate to 500kbps

subgeometer

  • Frazil ice
  • Posts: 362
    • View Profile
    • All in the Name of Liberty
  • Liked: 114
  • Likes Given: 71
Re: Test space
« Reply #152 on: September 24, 2019, 08:32:19 AM »
Fram strait, A closer view of Nares Strait, the Pacific side and the Laptev Sea

I have some others but I might post those, with all these on youtube at full resolutio

oren

  • Young ice
  • Posts: 4488
    • View Profile
  • Liked: 877
  • Likes Given: 1287
Re: Test space
« Reply #153 on: September 24, 2019, 09:45:50 AM »
Great animation. I can only imagine the volume that was melted in the Beaufort this year, with all these huge floes entering on one side and slush remaining on the other side.

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #154 on: October 07, 2019, 11:42:13 AM »
Reposting here so the library doesn't get clogged up
In the spirit of sharing I stole the genius of the best libraries around.. plotly does more than I could have dreamed of and exports to WebGL, takes some time to load but worth it :D

Smallest example: http://rpubs.com/macid/ITP119_TempC
check the rest on http://rpubs.com/macid/, the biggest ones have crashed my browser on occasion, would like to hear how it works for other people. Could optimize with averaging some data points if needed or other 'd0h' moments.

code updated on https://github.com/macidR/arcticPlotting/blob/master/plotly.R much quicker and awesome
Nearly perfect apart from using 6/8 cpu's and 18GB of memory. Took a long time to load (my broadband is challenged) but hasn't crashed. Actually resource use is dropping now it's loaded. Shame it doesn't show temperature/salinity?? as the slider moves ;)
Tried loading 2 at once, cpu 80% 32GB RAM 2 disks  writing 100% for~2m eventuall rebooted, had a lot of other things running
trying with chrome
Quote
hmm doesn't sound like you're seeing what I see, only tested with chrome, less then GB mem use & 20 secs loading or so, data download should only be 10MB.

The controls are supposed to let you pan/zoom/orbit the thing

Sample of ITP116 attached (one of the nicest)
ok, good with chrome on itp116, trying again with firefox. yup, memory goes to max with firefox.
very stable with 2 running on chrome. itp113 ran lots of 0-200m profiles in between the 0-800m and this shows the 50-100m warm layer well. 3d pan, orbit and slider is smooth.
Thanks macid
edit: works well on microsoft edge too. I disabled all add-ons on firefox but it still hogs memory(tested on 2 PC's)

Anyone else up for testing?  http://rpubs.com/macid/ITP119_TempC
others here: http://rpubs.com/macid/

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #155 on: October 07, 2019, 08:43:43 PM »
itp116 10-50m. Code works in Rstudio out of the box (after installing plotly properly)

macid

  • New ice
  • Posts: 23
    • View Profile
  • Liked: 12
  • Likes Given: 9
Re: Test space
« Reply #156 on: October 08, 2019, 02:41:39 AM »
I changed my username here too.. still think it's dying but less obnoxious

first attempt at adding bathymetry, http://rpubs.com/macid/ITP119_TempC_Bathy
was stuck most of the evening on assigning a separate color scale to this layer but have to do with height as indicator for now. Also intend to add some wind, air temps and ice tendencies indicators as information blobs above the curtains

And I forced x,y ranges to equal dimensions in the new script, else the buoys path was getting stretched in on or the other direction, most paths created yesterday had fairly even x/y range, so not much of an issue but good to get right.
« Last Edit: October 08, 2019, 02:51:37 AM by macid »

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #157 on: October 09, 2019, 01:21:20 PM »
Good progress. I got stuck attempting to change the temperature bar to salinity but my R plotly knowledge is zero. I couldn't see how to change the variable. :( Is it possible to show temp and salinity on the same scene?
I'm also wondering about using your 2d version for the mosaic iatb buoys http://iabp.apl.washington.edu/maps_daily_table.html though manual download is a bit laborious.

macid

  • New ice
  • Posts: 23
    • View Profile
  • Liked: 12
  • Likes Given: 9
Re: Test space
« Reply #158 on: October 09, 2019, 02:16:29 PM »
Salinity data needs to be taken from the dat files and added to the data frame first,
 tdata$salinity <- as.numeric(txdata$Salinity)
in the .dat import loop, same place temperature is added, then replace color = ~tempC with color = ~salinity in the plotly setup (it takes columns from the data frame specified at the start)

One problem I'm running into is that you can only use one colorscale per plot, still trying to hack my way around that, until then not sure how to display both data, maybe as circle lines, or offset slightly in one direction to make a two sided curtain..

I made the bathy work on ITP 116 which is floating around the north pole and just crossing the Lomonosov ridge, you can see the buoy got stuck for a while in spot where there's an almost 2km vertical cliff / cutout in the ridge. Takes a bit to load but worth it:

 https://rpubs.com/macid/ITP116_TempC_Bathy (hover over a bathy point to see depth in meters, not to z-axis scale..)


uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #159 on: October 09, 2019, 06:03:08 PM »
Quote
I made the bathy work on ITP 116
Tried this in ms edge which seems fastest to load. It works well, though the movement jerks a bit. Similar performance with chrome.

Thanks for the tip on salinity. There were quite a few more references to tempC but easy to find once I got the idea. Names instead of numbers...novel ;)

Why not tempC>2?
« Last Edit: October 09, 2019, 06:10:12 PM by uniquorn »

macid

  • New ice
  • Posts: 23
    • View Profile
  • Liked: 12
  • Likes Given: 9
Re: Test space
« Reply #160 on: October 09, 2019, 07:27:10 PM »
Quote
I made the bathy work on ITP 116
Tried this in ms edge which seems fastest to load. It works well, though the movement jerks a bit. Similar performance with chrome.

Thanks for the tip on salinity. There were quite a few more references to tempC but easy to find once I got the idea. Names instead of numbers...novel ;)

Why not tempC>2?

Nice, I bet i forget a few yes hehe.
the !tempC > 2 was crude error correction, you can skip or set the treshold higher, one buoy had anomalous high temps of 30+ degrees in a few spots.

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #161 on: October 09, 2019, 07:37:49 PM »
I would definitely go higher, maybe 5C

Now to add temp/ salinity label to hover as well as depth. That might be easier than 2 bars
« Last Edit: October 09, 2019, 07:48:09 PM by uniquorn »

macid

  • New ice
  • Posts: 23
    • View Profile
  • Liked: 12
  • Likes Given: 9
Re: Test space
« Reply #162 on: October 10, 2019, 10:20:14 PM »
I've been playing with the mosaic buoy data, projection looks correct using PlotSvalBard's UTM projection, you may also find the bit to download selected buoy csv files from the meereisportal useful, in the R script attached

« Last Edit: October 10, 2019, 11:41:56 PM by macid »

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #163 on: October 10, 2019, 10:23:35 PM »
I think you just got yourself volunteered :) though maybe I can help with the animations

macid

  • New ice
  • Posts: 23
    • View Profile
  • Liked: 12
  • Likes Given: 9
Re: Test space
« Reply #164 on: October 10, 2019, 10:50:50 PM »
I'll volunteer myself to automate further, e.g. how to run this on a server if useful
PlotSvalbard uses ggplot2 as base, gganimate should work with it as well I hope. I'm also looking for other data sources to include


uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #165 on: October 10, 2019, 11:19:30 PM »
I think I have an environment problem (haven't we all?)
object 'rbind.fill' not found
my R installation has been a bit haphazard on win10
dplyr      0.8.3 is installed
will look at it tomorrow
« Last Edit: October 10, 2019, 11:24:40 PM by uniquorn »

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #166 on: October 11, 2019, 07:18:53 PM »
:) fix is good

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #167 on: October 19, 2019, 08:03:19 PM »
Sbuoy distance test. Definitely losing quality on these gifs
hmmm - 600px × 450px (scaled to 580px × 435px)
reposted at 580px wide is better
« Last Edit: October 19, 2019, 08:15:36 PM by uniquorn »

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #168 on: October 20, 2019, 12:55:23 PM »
testing mosaic sbuoys with fade -up to oct20, still no labels

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #169 on: October 20, 2019, 06:05:13 PM »
Deformation test, distances calculated using octave mapping are lower than haversine. I'm not sure what method it uses. removed
« Last Edit: October 21, 2019, 11:56:18 PM by uniquorn »

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #170 on: October 22, 2019, 12:01:16 AM »
delaunay test showing 35 iabp mosaic buoys, naybe some are in similar locations. That will show up with labelling. This only shows 0mins, not the 30m. Quite a few dropouts. Will have to think about how to handle those.
added mp4 which includes the recent distortion.
removed
« Last Edit: October 23, 2019, 08:07:53 PM by uniquorn »

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #171 on: October 23, 2019, 08:17:49 PM »
update on mosaic iabp buoys. longer timescale day283-296, no labels. North still up
testing  R still, north down
removed
« Last Edit: October 24, 2019, 12:44:44 AM by uniquorn »

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #172 on: October 24, 2019, 03:09:42 PM »
delaunay deformation test day283-296, random buoys with no dropouts
cosmetic changes using facealpha
« Last Edit: October 24, 2019, 04:12:58 PM by uniquorn »

kassy

  • Grease ice
  • Posts: 776
    • View Profile
  • Liked: 291
  • Likes Given: 338
Re: Test space
« Reply #173 on: October 24, 2019, 04:16:49 PM »
I am sure delaunay got mentioned relating to Elite (the BBC B game) and i can totally see why.

This already looks more like the PC port so that´s good.  :)
Þetta minnismerki er til vitnis um að við vitum hvað er að gerast og hvað þarf að gera. Aðeins þú veist hvort við gerðum eitthvað.

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #174 on: October 24, 2019, 04:35:17 PM »
Right on Commander
Testing 34buoys.
Quote
Compute the Delaunay triangulation for a 2-D or 3-D set of points.
 tri = delaunay (x, y
For 2-D sets, the return value tri is a set of triangles which satisfies the Delaunay circum-circle criterion, i.e., only a single data point from [x, y] is within the circum-circle of the defining triangle. The set of triangles tri is a matrix of size [n, 3]. Each row defines a triangle and the three columns are the three vertices of the triangle. The value of tri(i,j) is an index into x and y for the location of the j-th vertex of the i-th triangle.
Quote
Function File: trimesh (tri, x, y, z)
Plot a 3-D triangular wireframe mesh.
In contrast to mesh, which plots a mesh using rectangles, trimesh plots the mesh using triangles.
tri is typically the output of a Delaunay triangulation over the grid of x, y. Every row of tri represents one triangle and contains three indices into [x, y] which are the vertices of the triangles in the x-y plane. z determines the height above the plane of each vertex. If no z input is given then the triangles are plotted as a 2-D figure.

It looks right but am I only setting 1 vertex?
@macid. Any chance you can knock up the same thing in R to check. Probably only take you 15mins ;)
« Last Edit: October 25, 2019, 12:39:30 AM by uniquorn »

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #175 on: October 25, 2019, 10:52:12 PM »
Changed a couple of lines of code to show distance change as a percentage (dist2-dist1)/dist2 and selected 9 buoys closer to Polarstern to get a more 'local' feel for movement. This also reduces the possible error using octave mapping for distance. day283-298.
switched from a type9 back to cobra mk3

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #176 on: October 29, 2019, 07:59:53 PM »
apple picking, pressing and drying (and eating) so no change to code apart from the axes.
def test day297-302. A big change at 302d3hr. Looking into that.

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #177 on: October 31, 2019, 11:51:07 PM »
3 ascat images per day since 2019206. Here testing how it looks.  Summer season so lots of weather. Also added the mosaic start area graticule in red, transparency25%, but it looks pretty weird after saving as avi->mp4 with -crf29 to reduce size. It's not much better without crf though
Makes me want to go back to gif even if all the ani's have to be shorter.
removed
« Last Edit: November 01, 2019, 12:30:07 AM by uniquorn »

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #178 on: November 01, 2019, 12:16:39 AM »
Temporarily posting ascat interferometry test, shortest timescale, with amsr2 overlay, 100%conc is transparent. Some sort of fault near mosaic. I think the ascat order is wrong - removing
« Last Edit: November 01, 2019, 12:28:33 AM by uniquorn »

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #179 on: November 07, 2019, 09:52:19 PM »
size test- tested
« Last Edit: November 07, 2019, 10:01:43 PM by uniquorn »

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #180 on: November 12, 2019, 09:41:20 PM »
Some moving lines but less than I thought. Will take central buoys out till delaunay is stable. The delaunay calculation is before the distance and mesh so the z axis changes shouldn't affect the lines.

Shelving the 3d idea for a while.
« Last Edit: November 13, 2019, 09:53:25 PM by uniquorn »

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #181 on: November 16, 2019, 06:41:35 PM »
pbuoy motion. That northerly quintet really accelerate away.
Relatively less 'brownian motion' at higher res

uniquorn

  • Nilas ice
  • Posts: 1560
    • View Profile
  • Liked: 765
  • Likes Given: 160
Re: Test space
« Reply #182 on: November 16, 2019, 11:23:41 PM »
small is beautiful. 3 buoys close to PS. day285-320. testing .removed
« Last Edit: November 18, 2019, 06:41:35 PM by uniquorn »