Support the Arctic Sea Ice Forum and Blog

Author Topic: Maximising asif skills using near real time Mosaic data  (Read 11371 times)

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #150 on: May 19, 2020, 02:25:07 PM »
Several improvements to the webapp

1) Better maps
2) Added new buoys
3) Added function to easily change parameters of the code

uniquorn, i see your point, thats going to be pretty difficult to do though  :-[


Webapp will be kept online for 12-hours. Find it here:

http://52.14.65.105:8501
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

blumenkraft

  • Guest
Re: Maximising asif skills using near real time Mosaic data
« Reply #151 on: May 19, 2020, 04:14:24 PM »
Amazing, Simon! Well done.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #152 on: May 20, 2020, 02:51:24 PM »
Amazing, Simon! Well done.

Thanks!

Understandably I have been very careful about keeping track of usage. At the moment, it looks like if i leave the website live indefinitely, I shouldn't incur any charges.


Random looking ip addresses can be off-putting, I get that. For this reason, I just registered a domain name at "www.mosaic-ice.com", which is being processed by amazon. I can then route the IP of the webapp to that domain name and square the project off.

In the meantime, im going to leave it running for 5-days here;




http://18.222.70.206:8501




Id encourage anyone and everyone to check it out, just so I can be sure that multiple users creating their own cache doesnt go over-and-above my free limits.
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

oren

  • First-year ice
  • Posts: 6291
    • View Profile
  • Liked: 2309
  • Likes Given: 1941
Re: Maximising asif skills using near real time Mosaic data
« Reply #153 on: May 20, 2020, 02:59:16 PM »
Simon thanks so much for creating this.
Isn't there some way for you to set limits on usage such that when it gets to the free allotment it stops for the month?
If not, maybe it would help putting the % of quota used on the home screen, so that users will know to avoid using the site when nearing the limit.

oren

  • First-year ice
  • Posts: 6291
    • View Profile
  • Liked: 2309
  • Likes Given: 1941
Re: Maximising asif skills using near real time Mosaic data
« Reply #154 on: May 20, 2020, 03:20:54 PM »
A question. The dropbox enabled easy download of the processed data files. Will it be possible to do the same from the web app site?
I hope I am not bombarding with requests. Still not sure what of all this I will use, just checking the goods...

Edit: I think we have already discussed download, but I haven't found how to on the current site.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #155 on: May 20, 2020, 03:31:29 PM »
A question. The dropbox enabled easy download of the processed data files. Will it be possible to do the same from the web app site?
I hope I am not bombarding with requests. Still not sure what of all this I will use, just checking the goods...

Edit: I think we have already discussed download, but I haven't found how to on the current site.

I feel amazon deliberately try to lead you down as slope as they dont provide a way to turn off a machine at a cpu-hour threshold.

Thanks for the suggestion about displaying the quota. However, I feel this would potentiate a back-door onto my account, as I would have to write a way of pulling that data to the webapp. I have set a monthly budget of $5, with email alerts at $2.50, so I should be able to stay well on-top of things.

With respect to downloading the datasets. I am sure there must be a way, still working on figuring it out- but its top of my list!
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #156 on: May 20, 2020, 04:06:30 PM »
oren, you should now be able to download csv files straight from the site, let me know how it goes

(I dont own excel so i can only view them as notepad files)
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #157 on: May 20, 2020, 04:57:04 PM »
just downloaded T56. It needs another 12 zeros at the beginning of the 'change_in_thickness(m)' column. Not much of a problem. Site is timing out right now.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #158 on: May 20, 2020, 05:33:14 PM »
Replaced NaNs with 0s to help with downloading and using csv files.

Rerouting traffic to new website/ domain name (at a very small cost in the region of $0.40 per million lookups):

The port at 8501 is unfortunately a must, and wont be going away any time soon


http://www.mosaic-ice.com:8501/
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

SteveMDFP

  • Nilas ice
  • Posts: 1784
    • View Profile
  • Liked: 369
  • Likes Given: 23
Re: Maximising asif skills using near real time Mosaic data
« Reply #159 on: May 20, 2020, 05:42:45 PM »
The port at 8501 is unfortunately a must, and wont be going away any time soon

http://www.mosaic-ice.com:8501/

"Any port in a storm."

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #160 on: May 20, 2020, 06:44:23 PM »
The port at 8501 is unfortunately a must, and wont be going away any time soon

http://www.mosaic-ice.com:8501/

"Any port in a storm."

10/10, had a giggle
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #161 on: May 26, 2020, 11:02:24 AM »
http://www.mosaic-ice.com:8501/
data currently stops at may21 (needs a reboot occasionally?)
suggestions:
add links to the original files used (temp_proc, ts, TEMP120)
Is there any chance to show all the 6hourly therm1 temperatures? If not a daily average might be better.
Sorry, have no time to look at streamlit at the moment.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #162 on: May 28, 2020, 10:29:46 PM »
python ice thickness update, may28. Posting here until I'm sure they are reliable.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #163 on: May 30, 2020, 11:56:57 PM »
Checking Tbuoy thickness
T73 definitely has a problem. The last few rows of T62.
« Last Edit: May 31, 2020, 12:07:16 AM by uniquorn »

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #164 on: June 01, 2020, 03:49:49 PM »
T77 temperatures, heat 30sec and heat 120sec, may10-jun1
melted snow refreezing?

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #165 on: June 01, 2020, 05:32:32 PM »
The t77 chart above shouldn't be showing that much bottom melt. Here is a close look at the ice/ocean thermistor which stays at, or very close to therm130, possibly moving to 129.
Possibly related to the maximum ocean temperature we are using.
The problem may get worse as the temperature gradient levels off. Tricky as there may be super cooled water beneath the ice. edit:I doubt that with warmer air temps.

updated below
« Last Edit: June 01, 2020, 07:52:40 PM by uniquorn »

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #166 on: June 01, 2020, 07:24:26 PM »
The t77 chart above shouldn't be showing that much bottom melt. Here is a close look at the ice/ocean thermistor which stays at, or very close to therm130, possibly moving to 129.
Possibly related to the maximum ocean temperature we are using.
The problem may get worse as the temperature gradient levels off. Tricky as there may be super cooled water beneath the ice. edit:I doubt that with warmer air temps.

Spring (summer?) really is messing with these buoys, I have a feeling the reliability of any measurements will be decreasing exponentially. Just look at their raw data.

By the way I cant edit the website myself (im on the clock for a project these days) but I just found out its imperative that one clears the cache to see the most up-to-date data.

Clearing the cache can be done in the tab on the right of the main page

Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #167 on: June 01, 2020, 08:03:16 PM »
Thanks, streamlit updates with cleared cache. Still checking data but I see the old python version I was using had mean ocean temp of < -2.5C. Editing that to -2.1C gives closer alignment to the bottom melt thermistor (like the streamlit version using -2.2C). I'd like to go lower but there are a few -2.06C deeper down. That drives up ice thickness.
examples below.
Will keep tinkering. maybe <-2.06C ....
« Last Edit: June 02, 2020, 08:50:45 PM by uniquorn »

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #168 on: June 01, 2020, 10:55:39 PM »
Agree completely about the supercooled idea (well not "supercooled")- there is probably a phase-equilibrium across several thermistors at the moment. Possibly supplement with HEAT data? Ill look into it
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #169 on: June 02, 2020, 02:54:00 PM »
Webapp update.

Amazon emailed me suggesting I may be eligible for 300$ account credit based on my activity. Should I qualify ill be able to significantly ramp-up the scope of;

www.mosaic-ice.com:8501

Adding several more features, including the ability to handle more than one dataset at any one time (due to migrating to a faster machine). Fingers crossed.
« Last Edit: June 02, 2020, 06:52:17 PM by SimonF92 »
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #170 on: June 02, 2020, 04:40:53 PM »
Good luck :) as long as you don't block adblockers.
Continuing analysis of T77 here are the temperatures from may18-jun2. I think there was snow between therm39 (possible therm37) and 43 and based on the animation above, I don't think there is bottom melt yet.
Not sure which parameters can be adjusted to match.
click for a gif to get rid of the lines.

blumenkraft

  • Guest
Re: Maximising asif skills using near real time Mosaic data
« Reply #171 on: June 02, 2020, 04:51:30 PM »

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #172 on: June 02, 2020, 06:52:36 PM »

www.mosaic.ice.com:8501


What's happening here? Is ice.com your domain?

Oops! Fixed


Great figure uniquorn, so much data
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #173 on: June 02, 2020, 08:23:58 PM »
T76 almost makes me want to give up this idea. One of the heaters looks like it's broken and there appears to be an air cavity that has been iced over. Drained before it could freeze?
The estimation seems too low at the moment.  117-41=76, *2=152cm plus maybe 24cm snow(less since the warm spell) click to play
« Last Edit: June 02, 2020, 08:31:53 PM by uniquorn »

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #174 on: June 02, 2020, 10:52:39 PM »
I didnt know this existed, but reassuringly we use somewhat similar methods

https://www.seanoe.org/data/00485/59709/
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

blumenkraft

  • Guest
Re: Maximising asif skills using near real time Mosaic data
« Reply #175 on: June 03, 2020, 11:24:18 AM »

Oops! Fixed

Haha, ok! That would have been amazing.  ;D

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #176 on: June 03, 2020, 02:24:58 PM »
I didnt know this existed, but reassuringly we use somewhat similar methods
https://www.seanoe.org/data/00485/59709/
Winter only. I think our winter results are ok though snow needs a bit more work.
Might have to be more manual for summer. If the buoys last that long.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #177 on: June 03, 2020, 03:57:24 PM »
I take it when they do leave the pack they will get scooped back up and put further north again? Iv been watching their relentless and almost unprecedented race toward the fram with bemusement, half of them will be out the ice within a month at this rate
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #178 on: June 03, 2020, 04:17:08 PM »
There doesn't seem to be an 'afterstory' for our courageous buoys. If any parts do get re-used they are anonymously renumbered. They should 'do a Musk' and go big on reusability. Whoi itp116 (not part of mosaic)already making its way down the greenland coast.
Hopefully they have a new batch ready for deployment further north.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #179 on: June 03, 2020, 04:46:57 PM »
There doesn't seem to be an 'afterstory' for our courageous buoys. If any parts do get re-used they are anonymously renumbered. They should 'do a Musk' and go big on reusability. Whoi itp116 (not part of mosaic)already making its way down the greenland coast.
Hopefully they have a new batch ready for deployment further north.

If all else fails, I will take one and put it in my living room. I think it would be a good conversation starter. Infact, at 5metres it might do better in the garden- my neighbours might think im conducting surveillance and stop secretly feeding my cat
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #180 on: June 04, 2020, 10:54:40 PM »
Checking T64, the most southerly.
2020-05-10   28   45   144
algo is detecting the snow peaks from heat, pretty good. Arguably the ice should be thicker at th150 but we have to make a cut off temperature.

There is a loss of snow event by may26 22:00. Algo begins to detect it on may30 though the chart shows it before then.
2020-05-29      28   45   141
2020-05-30    33   45   140
snow drops to 2cm, should be closer to 10cm.

I don't see any bottom melt. algo detects 6cm from may24. Experimenting with the old py code I think I see better results with mean ocean temp of -2.05 during summer. Less 'phase-equilibrium' perhaps.
2020-05-24   28   45   144
2020-05-25   28   45   142

Something is wrong near the end because we lose 0.1m. Incomplete day perhaps?
2020-06-03   43   44   140   -0.06   T64   0   -1.8   0.02
2020-06-04   43   44   139   -0.1   T64   0   -1.7   0.02





« Last Edit: June 04, 2020, 10:59:59 PM by uniquorn »

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #181 on: June 05, 2020, 03:54:16 PM »
Checking T64, the most southerly.
2020-05-10   28   45   144
algo is detecting the snow peaks from heat, pretty good. Arguably the ice should be thicker at th150 but we have to make a cut off temperature.

There is a loss of snow event by may26 22:00. Algo begins to detect it on may30 though the chart shows it before then.
2020-05-29      28   45   141
2020-05-30    33   45   140
snow drops to 2cm, should be closer to 10cm.

I don't see any bottom melt. algo detects 6cm from may24. Experimenting with the old py code I think I see better results with mean ocean temp of -2.05 during summer. Less 'phase-equilibrium' perhaps.
2020-05-24   28   45   144
2020-05-25   28   45   142

Something is wrong near the end because we lose 0.1m. Incomplete day perhaps?
2020-06-03   43   44   140   -0.06   T64   0   -1.8   0.02
2020-06-04   43   44   139   -0.1   T64   0   -1.7   0.02

I get a day off Sunday. Im going to take some time to properly use python to debug the model, will post the comparison results here
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #182 on: June 05, 2020, 03:57:24 PM »
PS iv been thinking, the ocean is more saline in the region the buoys are migrating to. This would in theory increase bottom melt, even though to us it would appear not much had changed in terms of ocean temp? Could this be a factor?

I cant find a current figure, but I know you and blumenkraft are pretty keyed in with your maps. Heres where T64 is. Im pretty sure thats close to a sharp salinity gradient

Scrub that i found one :)
« Last Edit: June 05, 2020, 04:02:53 PM by SimonF92 »
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #183 on: June 05, 2020, 06:39:34 PM »
whoi itp94 is our active profiler in the array. It's a touch more saline but presently able to be much a bit colder to match, mixing down to 150m. Besides, if there was bottom melt we'd see it on the animation.
Don't spoil your weekend for it.
« Last Edit: June 05, 2020, 10:14:45 PM by uniquorn »

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #184 on: June 07, 2020, 06:56:46 PM »
Ok so it looks like that the recent temperatures are causing the value of the Th(ice) to change, not Th(ocean), causing these spurious "melt" events due to the way I have coded the final figures, will continue to work on it

this is T64
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #185 on: June 07, 2020, 09:44:31 PM »
Thanks. If you have time it would be good to see the py code before it goes streamlit

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #186 on: June 08, 2020, 12:04:18 AM »
 ::)
Thanks. If you have time it would be good to see the py code before it goes streamlit

No problem, will post a link here when its done
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #187 on: June 09, 2020, 10:27:27 AM »
Ok so I have been really working at this, using algos for both HEAT deviation from top and bottom separately, then in combination. The readings are just too sporadic to return meaningful and consistent values. Th(ice) is fluctuating no matter what I try.

So I came up with something else.

How about we SET th(ice) as the winter measurement. Now, when temperatures are more variable and harder to work with, I suggest we use this set value for th(ice), but change it based on the ambient air temperature.

 If ambient air temps are above 0degc for a sustained period, we infer top-melt. Here's the if statement long-hand.

if (ambient air temperature is >0degC for 24hours) then (Th(ice) = Th(ice) - 1)

I need some expert opinions.

1) is this reasonable?
2) How do we calibrate? ie is 24hrs of +1degC enough to cause 2cm of top melt?

In the image, you can actually see the measurement blip that is throwing off our Th(ice), unfortunately there really is not much I can do to correct for this.
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

oren

  • First-year ice
  • Posts: 6291
    • View Profile
  • Liked: 2309
  • Likes Given: 1941
Re: Maximising asif skills using near real time Mosaic data
« Reply #188 on: June 09, 2020, 11:03:17 AM »
Yes, Th(ice) needs tp be fixed at a stable value derived from winter. This represents a physical attribute of the buoy installation. Since ice does not grow from the top, once installed Th(ice) remains the same. It just need to be discovered once for each buoy, using the various algorithms.
This, combined with limited time invested, has been hindering me from the start. I can't concentrate enough to decide once and for all on this Th(ice)(installed) for each buoy, as each algorithm gives a slightly different number. IMHO the installers should have published this number as part of buoy data, but oh well, never mind.
As for top melt, this can be identified by temperatures below Th(ice)(installed) being equal to air temperatures above that. Air temps fluctuate. Ice is an insulator. If a thermistor is in ice, it will fluctuate less and remain more rangebound than air temp. (The same is even truer for snow). In addition when air temps are extreme in any direction, the ice (and snow) should show a rather smooth gradient. Thus whether each thermistor is exposed to air or not can be calculated from the data, and no need for formulas which really cannot give a true answer as ice melt depends on more than just air temps - sunlight or clouds, wind speed, humidity, drainage, ice salinity, and more.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #189 on: June 09, 2020, 12:16:36 PM »
Yes, Th(ice) needs tp be fixed at a stable value derived from winter. This represents a physical attribute of the buoy installation. Since ice does not grow from the top, once installed Th(ice) remains the same. It just need to be discovered once for each buoy, using the various algorithms.
This, combined with limited time invested, has been hindering me from the start. I can't concentrate enough to decide once and for all on this Th(ice)(installed) for each buoy, as each algorithm gives a slightly different number. IMHO the installers should have published this number as part of buoy data, but oh well, never mind.
As for top melt, this can be identified by temperatures below Th(ice)(installed) being equal to air temperatures above that. Air temps fluctuate. Ice is an insulator. If a thermistor is in ice, it will fluctuate less and remain more rangebound than air temp. (The same is even truer for snow). In addition when air temps are extreme in any direction, the ice (and snow) should show a rather smooth gradient. Thus whether each thermistor is exposed to air or not can be calculated from the data, and no need for formulas which really cannot give a true answer as ice melt depends on more than just air temps - sunlight or clouds, wind speed, humidity, drainage, ice salinity, and more.

I agree with what you are saying, human eye for this may actually be best. However, if we do set th(ice) as immutable we reduce dimensionality through the assumption that top-melt does not exist?

At the moment, being exasperated as I am with the recent changes to reliability, I might just make it an immutable value.
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

oren

  • First-year ice
  • Posts: 6291
    • View Profile
  • Liked: 2309
  • Likes Given: 1941
Re: Maximising asif skills using near real time Mosaic data
« Reply #190 on: June 09, 2020, 12:34:13 PM »
The fixed value should be Th(ice)(installed) or similarly named. Th(ice) should vary when top melt begins, but never fall below (physically above...) Th(ice)(installed). Th(snow) could still make an appearance even during summer.

Thus we have air-snow-ThIcIn-ice-ocean thermistors in winter.
And air-ThIcIn-air-ice-ocean thermistors in summer when there is no snow
Thickness graphs should best be aligned at Th(Ice)(Installed) as they were in winter, thus showing top melt as well as bottom melt visually.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #191 on: June 09, 2020, 12:46:22 PM »
The fixed value should be Th(ice)(installed) or similarly named. Th(ice) should vary when top melt begins, but never fall below (physically above...) Th(ice)(installed). Th(snow) could still make an appearance even during summer.

Thus we have air-snow-ThIcIn-ice-ocean thermistors in winter.
And air-ThIcIn-air-ice-ocean thermistors in summer when there is no snow
Thickness graphs should best be aligned at Th(Ice)(Installed) as they were in winter, thus showing top melt as well as bottom melt visually.

ok, ill add a class variable Th(ice)[installed] on a per buoy basis. This will definitely clean up spurious data
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #192 on: June 09, 2020, 01:51:17 PM »
Not sure what to suggest. Perhaps it will become clearer once we see some top melt.
Thanks Simon, I hope it's still fun.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #193 on: June 09, 2020, 04:08:07 PM »
Python is a joy to write (when I get the chance).

Will distribute a code update soon.

PS Amazon gave me that $300. I'm suspicious it might be an entrapment to get me hooked. Dont look a gift horse in the mouth? A Trojan horse perhaps
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

SimonF92

  • Frazil ice
  • Posts: 293
    • View Profile
  • Liked: 121
  • Likes Given: 62
Re: Maximising asif skills using near real time Mosaic data
« Reply #194 on: June 12, 2020, 10:39:10 AM »
Hate to say this, so decided to wait a few days before I did.

I really cant see a way of getting th(ocean) anymore. Uniquorn I believe if you run the code on your end youll see what i mean. The temps at the ocean:ice interface are just too variable now.

That being said, if anyone has any data they would like to see graphed, let me know
Im working on a satellite-miner to detect changes in small ice-caps/ snow-fields. Send me recommendations to optimise the program with.

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #195 on: June 13, 2020, 08:58:29 PM »
It's surprised me how shallow the gradient is.
Much easier to define the points by eye than by maths.
« Last Edit: June 13, 2020, 09:09:53 PM by uniquorn »

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #196 on: June 13, 2020, 09:34:00 PM »
another way of looking at it.
This method makes it easier to spot snow and ice thickness. click to run
air where the lines diverge on the left.
Snow or melt ponds?  T56, T68 and T75 reach +3C close to ice surface.
ice is the thickening band in the middle. The thickness is calculated from the number of thermistors along the x-axis. Multiply by 2.
ocean on the right.
« Last Edit: June 13, 2020, 09:47:57 PM by uniquorn »

oren

  • First-year ice
  • Posts: 6291
    • View Profile
  • Liked: 2309
  • Likes Given: 1941
Re: Maximising asif skills using near real time Mosaic data
« Reply #197 on: June 14, 2020, 08:39:31 AM »
Isn't T58's data indicative of a deepening meltpond?

uniquorn

  • Young ice
  • Posts: 2733
    • View Profile
  • Liked: 1231
  • Likes Given: 243
Re: Maximising asif skills using near real time Mosaic data
« Reply #198 on: June 14, 2020, 02:32:24 PM »
I'm not sure. If so that is very rapid melt. jun8-14
Or is that what 'phase change' looks like?
« Last Edit: June 14, 2020, 02:45:43 PM by uniquorn »

oren

  • First-year ice
  • Posts: 6291
    • View Profile
  • Liked: 2309
  • Likes Given: 1941
Re: Maximising asif skills using near real time Mosaic data
« Reply #199 on: June 14, 2020, 03:02:47 PM »
Yeah I didn't notice the numbers. 80cm of top melt in 4 days is impossible. But once the ice reaches 0C I think melt progress should be rapid nonetheless.
I still wonder how 0C can propagate so deeply if the ice is solid, especially considering it's supposed to be somewhat salty. Above my pay grade unfortunately.