the ice north of the NE corner of Greenland started cracking As PIOMAS gives an average, with all these spaces between the thick floes could this explain why it shows low thickness in that area?
Yes, that could the explanation for areas with large leads. The reason Piomas is always shown as a postage stamp is because both its resolution and accuracy are so-so. For the Lincoln Sea, it does not offer many pixels of coverage. Its bizarre choice of a coordinate system centered in Greenland surfaces in a differencing bias artifact (fig 1 below from #1115 differencing, concentric circles in polar stereographic coords).
In this case Piomas was just plain wrong in the Lincoln Sea, which while minor in area is important to old and thick refugium ice. While some dramatic leads opened up in Sept 2016, these were unremarkable in historic context -- it happens a half dozen times each year and so far has portended little. Maybe some day, opening leads will be the first sign of a total blowout through the Fram.
Hycom has it right -- the ice has been 5-6 m thick there the last 30 days, as it has been for decades. There is no way to average in 10% open leads with 90% 5 m ice to get below 1.5 m as shown in Bliz92's Piomas image. https://www7320.nrlssc.navy.mil/hycomARC/navo/arcticictn_nowcast_anim30d.gif
Now Hycom has some issues of its own. Its palette appears to use 'black' for its heavy-handed text, coordinate grid, ice/land boundary, ice near 1.4 m thick, and ice 5 m thick. Each color in a map is supposed to have a unique meaning so 5 uses of 'black' is a total cartographic no-no.
But in fact the too-dark bin colors are not pure black: the thickest bin is RGB = 25 0 0 and the 1.4 m bin is a peculiar blue, RGB = 2 32 57. These two colors need to be lightened to fall within the human eye color gamut (see below, colorbrewer2.org)
Hycom animations uses 100 color bins to display ice thickness classes between zero and five meters, ie 0.05 m distinctions in thickness. That's already questionable in view of its inherent ground accuracy which would be lucky to get within 20% of true floe thicknesses. Twenty color bins would have been plenty (0.25 m increments); 100 bins is for downstream smoothing aesthetics.
However the Hycom map does not use its palette colors in its ice thickness map to any extent. The image below shows the residual 86 colors remaining after each of the palette colors is replaced by magenta. This means users cannot click on a map pixel itself to look up its thickness
, the error is coloring the data grid
(which isn't provided) instead of coloring the map
AMSR2 UHH maps of sea ice concentration prove the Hycom map error is gratuitous -- every pixel on a AMSR2 map corresponds uniquely to a color on the concentration key, as it should. AMSR2 uses the Arctic standard projection put forth by NSIDC, as it should; the Hycom map can be rotated into this from anglocentric meridional view but only with further degradation of colors.
In the old days of printed journals and CMYK dithering, it didn't matter. For the last 20 years though, we've used the internet for color
. All imagery software has provided a color selecting tool (called 'magic wand' in photoshop) that selects all pixels of the same color as a clicked pixel (or to within a user-set radius of that pixel in RGB color space).
A Hycom frame has 188 colors in total, well within the 255 limit of gif format. It does not dither text (anti-aliasing) like so many products we see, text lifts off when pure black RGB =000 is substituted. Open ocean is represented by a pure gray RGB 254 254 254; lakes are a single tan. The palette also has a bounding box of one pure black pixel.
Importantly for complex areas like the Lincoln Sea, the boundary between land and ocean is also pure black. That means it can be replaced by land brown to reduce confusion between boundary and dark thickness bin pixels.
Thus the template (receiving base map) is done correctly -- except for the fact that two of the color bins are the wrong size, 8 pixels high instead of 6 like the other ninety eight bins. (It also lacks tick marks on the side associated with scale depth number; color bins are actually depth ranges.) This is a very strange error suggesting unsystematic construction of the color key.
The colors are a total hack, looking at them as HSV or RGB progressions or in human eye gamut (below). It is better to construct palettes numerically in a spreadsheet or using precision gradient tricks, as described in previous posts or with visual discrimination ramping. It is easy to crank out palettes by the billions; this one was likely furnished anonymously within the software as a menu option.
The fourth image shows a technical analysis of the Hycom sea ice thickness display. This proves Hycom is showing the very thickest ice in the Lincoln Sea -- not a land boundary or the too dark blue but something above yellow and red. However the display can't be fixed by indexed color table replacement (which may require separate treatment of each frame!); it would have to be redrawn from raw data.