Support the Arctic Sea Ice Forum and Blog

Author Topic: interpretation of MODIS imagery  (Read 7971 times)

Andreas T

  • Nilas ice
  • Posts: 1149
    • View Profile
  • Liked: 18
  • Likes Given: 4
interpretation of MODIS imagery
« on: June 05, 2016, 01:27:46 AM »
I feel I need to learn more about interpreting the imagery outside the visible spectrum and  prompted by a question by seaicesailor I have been reading a bit more on the topic. I hope this could be a good place to collect observations which could help us and others who want to make better use of this resource.

To start things off here is a comparison of visible and 7-2-1 bands which shows the ability of 7-2-1 images to distinguish between grey shades due to water on ice (blue arrow) and grey shades due to transparent (thin) ice showing dark water under the ice (red arrow)

Neven

  • Administrator
  • First-year ice
  • Posts: 9470
    • View Profile
    • Arctic Sea Ice Blog
  • Liked: 1333
  • Likes Given: 617
Re: interpretation of MODIS imagery
« Reply #1 on: June 05, 2016, 10:49:09 AM »
Cool stuff, Andreas. I've also been reading a bit about remote sensing lately. Unfortunately I'm too busy keeping an eye on everything, or else I'd be looking into it more, just to understand all the stuff those guys are doing in the Greenland glacier thread.

But I'll be reading this thread to learn about some of the basics.
The enemy is within
Don't confuse me with him

E. Smith

seaicesailor

  • Guest
Re: interpretation of MODIS imagery
« Reply #2 on: June 07, 2016, 08:33:53 PM »
Thanks a lot for this Andreas. I will try to bring images here if anyone wants to discuss it, before misinterpreting them elsewhere...

Andreas T

  • Nilas ice
  • Posts: 1149
    • View Profile
  • Liked: 18
  • Likes Given: 4
Re: interpretation of MODIS imagery
« Reply #3 on: June 08, 2016, 11:43:28 PM »
yes please, as I said elsewhere: I want to collect examples of images where the colours can be related to known features, so we build up a kind of library of situations where parallels can be drawn from known situations to unknown ones. That's how experienced analysts of these thing work I guess, gaining experience from previous correlations.

The first example I posted shows darkened ice on a floe which I know is thinner ice by tracking this floe over a couple of months. In the same shot landfast ice shows similar darkening in the visible but the 7-2-1 image shows water on the landfast ice but not on the floe.

A-Team

  • Young ice
  • Posts: 2977
    • View Profile
  • Liked: 944
  • Likes Given: 35
Re: interpretation of MODIS imagery
« Reply #4 on: June 09, 2016, 04:42:48 PM »
This is a really good project with the potential of becoming a major forum resource.

Put another way, there should already exist a big tutorial with many expert annotation examples associated with each layer provided at Nasa's Worldview, but there isn't. The effect is another very expensive resource goes under-utilized and misinterpreted.

I came across this one that -- along with its 13 later article cites http://tinyurl.com/haxe9f2 -- may be of interest, http://tinyurl.com/hbscow7 with free full text. Neven discussed the product earlier at http://neven1.typepad.com/blog/2013/06/new-map-on-the-block.html and Wipneus used it today on the AMSR2 forum and archives them at https://sites.google.com/site/apamsr2/home/pngcby32/

Quote
Investigating High-Resolution AMSR2 Sea Ice Concentrations during the February 2013 Fracture Event in the Beaufort Sea
Remote Sensing 6(5):3841-3856 ยท April 2014 DOI: 10.3390/rs6053841
Alexander Beitsch et al

Leads with a length on the order of 1000 km occurred in the Beaufort Sea in February 2013. These leads can be observed in MODIS images under predominantly clear sky conditions. Sea ice concentrations (SIC) derived from the Advanced Microwave Scanning Radiometer 2 (AMSR2) using the Bootstrap (BST) algorithm fail to show the lead occurrences, as is visible in the MODIS images.

In contrast, SIC derived from AMSR2 using the Arctic Radiation and Turbulence Interaction Study (ARTIST) sea ice algorithm (ASI) reveal the lead structure, due to the higher spatial resolution possible when using 89-GHz channel data.

The ASI SIC are calculated from brightness temperatures interpolated on three different grids with resolutions of 3.125 km (ASI-3k), 6.25 km (ASI-6k) and 12.5 km (ASI-12k) to investigate the effect of the spatial resolution... Visual comparison with MODIS True Color imagery [shows] ASI-3k are able to reproduce lead structure and size in the sea ice cover, which are not or are less visible in the other SIC data. The results will be valuable for selecting a SIC data product for studies of the interaction between ocean, ice, and atmosphere in the polar regions.
« Last Edit: June 09, 2016, 04:51:42 PM by A-Team »

A-Team

  • Young ice
  • Posts: 2977
    • View Profile
  • Liked: 944
  • Likes Given: 35
Re: interpretation of MODIS imagery
« Reply #5 on: June 09, 2016, 05:06:48 PM »
Here is a thesis on AMSR-E 89 GHz on how to take out atmospheric effects using ECMWF (total columnar water vapor, wind speed, liquid water path, skin temperature and 2 m temperature). It is not at all clear that Worldview applies this algorithm, whether it is feasible for us to do so, or whether something even better is around.

I'm also wondering since ECMWF by all accounts is markedly better than GFS, how big a job it would be to run a version of nullschool based on ECMWF. Easiest would be a button switch on the existing nullschool site to facilitate comparisons.

Atmospheric Correction of Brightness Temperatures for Sea Ice Concentration Retrieval using 89 GHz Algorithms
Junshen Lu
http://www.iup.uni-bremen.de/PEP_master_thesis/thesis_2014/Thesis_Lu.pdf

Sea ice has large impact on climate changes. An accurate retrieval of the spatial
and temporal distribution of sea ice is thus essential to understand and predict
the weather and climate. Taking advantage of the high resolution of AMSR-E 89 GHz
channel, the ASI (Artist Sea Ice) algorithm has a higher spatial resolution, but
is more sensitive to the atmospheric impact. In this study, the influence of atmospheric
parameters on sea ice concentration retrieval in the Arctic is studied, and
a new version of ASI algorithm that includes atmospheric correction is developed.
The correction is carried out by simulating TB contributed by atmosphere with
a linear forward model developed for the frequencies of AMSR-E. ECMWF data,
co-located with AMSR-E measurements, are used as the atmosphere profiles. The
included parameters are: total columnar water vapor, wind speed, liquid water
path, skin temperature and 2 meter air temperature. The combined correction of
TWV, WS, LWP and Tskin effectively screens out most atmospheric influences.

Andreas T

  • Nilas ice
  • Posts: 1149
    • View Profile
  • Liked: 18
  • Likes Given: 4
Re: interpretation of MODIS imagery
« Reply #6 on: June 17, 2016, 12:33:26 AM »
The melting of snowcover in the NWP gives an opportunity to look what different frequency band images show. The ice between the islands of the CAA has the advantage that much of it does not move during the winter which makes it easy to trace the history of its development.
The area shown is west of Resolute which is partly shown bottom right.
First image is an ASAR of 9.4. when snow covered the ice and all of it looked the same in visible MODIS images. http://www.polarview.aq/images/106_S1jpgsmall/201604/S1A_EW_GRDM_1SDH_20160409T133203_4EE4_N_1.jpg
Second is a 7-2-1 band image of the 14.6. where the same features can be recognized.
http://go.nasa.gov/21sskZD


« Last Edit: June 18, 2016, 06:43:40 PM by Andreas T »

Andreas T

  • Nilas ice
  • Posts: 1149
    • View Profile
  • Liked: 18
  • Likes Given: 4
Re: interpretation of MODIS imagery
« Reply #7 on: June 17, 2016, 12:35:44 AM »
The next images  show how these features originated during the winter.
In October when new ice begins to form floes which have drifted into the passage from further North become multiyear ice among the thinner shifting ice cover.
In December these floes are recognizable because their thickness and deeper snow cover means their surface temperature is lower (band 31 of 11.12.2015) A recently opened polynya shows up as relatively high temperatures of the young ice surface.
In January (band 31 of 9.1.2016) shows part of the previous polynya has opened again, forming another stage of more recent ice.
https://worldview.earthdata.nasa.gov/?p=arctic&l=MODIS_Terra_CorrectedReflectance_Bands367(hidden),MODIS_Terra_CorrectedReflectance_Bands721,MODIS_Aqua_CorrectedReflectance_Bands721,MODIS_Aqua_CorrectedReflectance_TrueColor(hidden),VIIRS_SNPP_CorrectedReflectance_TrueColor(hidden),MODIS_Terra_CorrectedReflectance_TrueColor,Graticule,Coastlines,AMSR2_Sea_Ice_Brightness_Temp_6km_89H(hidden),MODIS_Terra_Brightness_Temp_Band31_Night(palette=rainbow_1,min=220.5,max=269.8,squash)&t=2015-12-11&v=-1620567.7782986271,-1106880.8757620812,-1140823.7782986271,-818112.8757620812
edit: added link (shortened links have stopped working in the past) as suggested by A-team
same band31 setup used for both dates
« Last Edit: June 18, 2016, 06:39:53 PM by Andreas T »

A-Team

  • Young ice
  • Posts: 2977
    • View Profile
  • Liked: 944
  • Likes Given: 35
Re: interpretation of MODIS imagery
« Reply #8 on: June 18, 2016, 01:54:05 PM »
This is quite useful to develop a line of interpreted examples. It might be a good idea to always include the url (or the tinyurl) to your WorldView set-up so that visitors can quickly replicate the view and, for example, look around the immediate area for other features or see how band 31 etc correlate with other Worldview channel choices.

Andreas T

  • Nilas ice
  • Posts: 1149
    • View Profile
  • Liked: 18
  • Likes Given: 4
Re: interpretation of MODIS imagery
« Reply #9 on: June 18, 2016, 06:46:37 PM »
Thanks for the reminder I have said the same myself  :-[
ammended the comment above

Andreas T

  • Nilas ice
  • Posts: 1149
    • View Profile
  • Liked: 18
  • Likes Given: 4
Re: interpretation of MODIS imagery
« Reply #10 on: June 25, 2016, 02:27:23 PM »

seaicesailor

  • Guest
Re: interpretation of MODIS imagery
« Reply #11 on: August 20, 2016, 01:29:23 PM »
Burning question: what is the best layer of Worldview to NOT see through the clouds and reveal more clearly storm structure / rain? I mean similarly of what is used to reveal hurricanes ...
Thanks if anybody knows :-)