And we begin

[Mallow]

You sure the graph doesn't mesh with the obs? Bc there looks like a sharp gradient of snow cover in that image Drac put up, with little snow south of northernmost Mongolia latitude. 2012 had a ton of snow south of 60N with anomalous less snow cover north of 60N.

The data being read is what the GFS is initialized with on 1.0 degree grids. If you shift the latitude boundary north to 90N, then this year clearly has the most snow cover when compared to the other years in the analysis. This is just following Cohens threshold.

 

Yeah, I'm sure. The 60N latitude is clearly identified on those plots. Count the 5°x5° grid boxes south of 60N yourself if you don't believe me.

 

You sure there's no bug in the code for the latitude delineation? Perhaps there's an error in the area weighting (by latitude)... I know I've made errors like that before, using cos(lat) but forgetting to convert latitude to radians. That could potentially explain the discrepancy.

[Mike.Ventrice]

Yeah, I'm sure. The 60N latitude is clearly identified on those plots. Count the 5°x5° grid boxes south of 60N yourself if you don't believe me.

 

You sure there's no bug in the code for the latitude delineation? Perhaps there's an error in the area weighting (by latitude)... I know I've made errors like that before, using cos(lat) but forgetting to convert latitude to radians. That could potentially explain the discrepancy.

 

Interesting point- not weighting by latitude at all.. I'll replicate that step and repost when I have some time! Thanks

[Mallow]

Interesting point- not weighting by latitude at all.. I'll replicate that step and repost when I have some time! Thanks

 

Of course, that'll weight the Himalayas (slightly) more than the areas close to 60N, so just weighting vs. not weighting I don't think could account for the discrepancy.

[Mike.Ventrice]

Of course, that'll weight the Himalayas (slightly) more than the areas close to 60N, so just weighting vs. not weighting I don't think could account for the discrepancy.

 

Hmmm then I'm not sure... Code is good... checked it a number of times and I'm very familiar with the format of the file. So yeah... interesting discrepancy. 

[Mallow]

Hmmm then I'm not sure... Code is good... checked it a number of times and I'm very familiar with the format of the file. So yeah... interesting discrepancy. 

 

Yeah, only thing left I can think of is the GFS data itself... but that shouldn't be so different from the other snow cover data sources, I wouldn't think. Can you plot a map of the GFS snow data from 10/22/2012 and 10/22/2014 and post the images on here?

[Mike.Ventrice]

Here's the GFS snow depth data that I convert into coverage. 

[Mallow]

Here's the GFS snow depth data that I convert into coverage. 

 

Thanks! Do you have the same plot for 10/24/2012?

[Mike.Ventrice]

Thanks! Do you have the same plot for 10/24/2012?

I don't at the moment will have to code it up- but I should have mentioned the archive is from the 0.75 era interim reanalysis- not the GFS near real time data

[Mallow]

I don't at the moment will have to code it up- but I should have mentioned the archive is from the 0.75 era interim reanalysis- not the GFS near real time data

 

Hmm... is it possible that there is a large systematic difference between the ecmwf reanalysis data and the gfs real-time data that's leading to the discrepancy?

[griteater]

As Mallow said, there's no comparison between 2012 and 2014 with the maps that Dacula posted.  2014 clearly has more snow south of 60N

[Mike.Ventrice]

Just sharing some objective results with the group. Take it for what it's worth

[Mallow]

Just sharing some objective results with the group. Take it for what it's worth

 

Oh, I and others certainly appreciate it. I'm just noting that it doesn't match the other objective data we've seen (see cfbaggett's plots).

[GaWx]

 Is natice daily data back yet? If not, does anyone have a guess as to when it will be? I'm waiting eagerly for Rutgers to finally update so I'd have apples to apples maps to get a good idea of the current Eurasian snowcover. Fwiw, I estimated that total (not just sub 60N) Eurasian snowcover was already in the 16.5 msk area way back on 10/19 (climo for early November) (based on comparing the 10/19 Rutgers map SCE with past instances of similar looking maps) with a good bit more to add based on the 10/19 model consensus for 12 day snowfall as well as extreme cold as per this post:

 

 http://www.americanwx.com/bb/index.php/topic/44473-and-we-begin/?p=3089971

 

Edit: In my mind, hitting 19.5 for total Eurasian would be a really big milestone as far as getting 2014 into the elite top four of the past 46 years. That would make an Oct. increase of 14, which would only have been exceeded by 1976's 16.75, 2009's 16, and 2012's 15.5. Two others were near 13.75 (1970 and 1968). The three most -AO's for DJF were 2009, 1976, and 1968, all Nino's.

[cfbaggett]

So I've found a surprising result after reading through Judah Cohen's 2011a paper about the Snow Advance Index. The proper lat/lon thresholds are 0-60N, 0-180E. After computing the snow-cover % (# of snow covered grid points/total # grid points), the resulting image is quite shocking. 2014 is clearly not as high as 2009 and 2012.  We are also seeing a reduction in the slope of d(snowcover)/dt... Most of the anomalous snow is in the 55-90N band... similar to 2006. I'm not sure what this means exactly, such that the snow-covered forcing is north of the Cohen region. Will need to do the research on this at a later time.

 

Everything mallow said, plus i would add that you have to be really careful when calculating the area represented by each grid point. It is not a straightforward cos latitude weighting. The data on ncep's site is polar stereographic, so not only do you have to adjust your dx according to latitiude, your dy changes as well. I'm not exactly sure what you did, but i am extremely confident my graph is correct, because my values match those that Cohen plots in the auxillary material of his 2011 paper. Plus, from the snow plots shown today, we are clearly higher south of 60N than any year in many decades right now. I just wish i could get new data from ncep so I can update my graph! :-)

[cfbaggett]

Is natice daily data back yet? If not, does anyone have a guess as to when it will be? I'm waiting eagerly for Rutgers to finally update so I'd have apples to apples maps to get a good idea of the current Eurasian snowcover. Fwiw, I estimated that total (not just sub 60N) Eurasian snowcover was already in the 16.5 msk area way back on 10/19 (climo for early November) (based on comparing the 10/19 Rutgers map SCE with past instances of similar looking maps) with a good bit more to add based on the 10/19 model consensus for 12 day snowfall as well as extreme cold as per this post:

http://www.americanwx.com/bb/index.php/topic/44473-and-we-begin/?p=3089971

Based on the snow plots posted today, i am guessing 8.5 msk south of 60N. Nice growth into eastern europe and southeast russia since the 19th. We are looking really good.

[GaWx]

Based on the snow plots posted today, i am guessing 8.5 msk south of 60N. Nice growth into eastern europe and southeast russia since the 19th. We are looking really good.

 

cfbaggett,

 Wow, 8.5 south of 60N already on 10/24 would be hugely encouraging news for strong DJF -AO lovers though not a total shock based on generous model based projections since 10/19.

 

 I said the following at the link below on 10/19:

 

"Estimate of total Oct. sub 60N Eurasian gains (correlates pretty well to # of sub 60N 1.0+ Oct. daily gains):

 

2014: 5.3 (that's only through 10/19)

2009: 7.4

2012: 6.5

2007: 3.9

2013: 3.8"

 

And later I said this:

 

"So, if we could get cfbaggett's graph to show 2014 sub 60N extent reaching 8.5+ as of 10/31 from the current ~6.7, that would be very good news for those wanting a strong DJF -AO. Even getting to just 8.0 extent may be enough."

 

http://www.americanwx.com/bb/index.php/topic/44473-and-we-begin/?p=3090386

 

 But why is Mike's graph suggesting the opposite? I'm confused! Mike, I respect your great contributions here. Could there be an error?

[Mike.Ventrice]

For what it's worth, when I include latitudes north of 60N, 2014 is clearly the king. Just this analysis does not highlight big snow cover south of 60N.

 

  0-90N, 0-180E:

 

 

0-60N, 0-180E:

 

[griteater]

 Is natice daily data back yet? If not, does anyone have a guess as to when it will be? I'm waiting eagerly for Rutgers to finally update so I'd have apples to apples maps to get a good idea of the current Eurasian snowcover. Fwiw, I estimated that total (not just sub 60N) Eurasian snowcover was already in the 16.5 msk area way back on 10/19 (climo for early November) (based on comparing the 10/19 Rutgers map SCE with past instances of similar looking maps) with a good bit more to add based on the 10/19 model consensus for 12 day snowfall as well as extreme cold as per this post:

 

GA - Total Eurasian snow cover is at 16.70 as of 10/23...

 

http://www.americanwx.com/bb/index.php/topic/44473-and-we-begin/?p=3095379

[Mallow]

For what it's worth, when I include latitudes north of 60N, 2014 is clearly the king. Just this analysis does not highlight big snow cover south of 60N.

 

  0-90N, 0-180E:

 

 

 

0-60N, 0-180E:

 

 

 

Those are very interesting, but I find them very strange. I still suspect a bug somewhere in the calculation, given that it doesn't seem to match with any of the other subjective or objective measures presented. I'd really like to see a map comparing the 2012 and 2014 snow cover used to generate those values, from whichever day you'd like. I think it would be easier to see where the discrepancy is coming from that way.

[GaWx]

For what it's worth, when I include latitudes north of 60N, 2014 is clearly the king. Just this analysis does not highlight big snow cover south of 60N.

 

  0-90N, 0-180E:

 

 

0-60N, 0-180E:

 

 

Mike/folks,

 Here is one of my problems with Mike's lower graph (sub 60N) that raises my suspicions greatly that something isn't right:

 

 Note that 2014 (black line) is tied with 2009 (light blue line) for the lowest 10/1 sub 60N extent on Mike's graph. Based on Rutgers (assuming it is accurate), this simply can't be the case! 2009's very low sub 60N 10/1 looks right. 2014's very low sub 60N 10/1 looks wrong.

 

I have attached both graphs below:

 

10/1/2009: look how little SCE there is in Eurasia, period, much less sub 60N (meaning Mike's 10/1/09 sub 60N looks right):

 

10/1/2014: look how much more SCE there is in Eurasia, including sub 60N: So. Mike's sub 60N for 10/1/14 has to be off (way too low).

 

 Conclusion: Mike's sub 60N 2014 line is way too low from start to finish.

[The_Global_Warmer]

For what it's worth, when I include latitudes north of 60N, 2014 is clearly the king. Just this analysis does not highlight big snow cover south of 60N.

 

 

Mr. Ventrice, is it possible you could produce one of these times series for the same years for say  October 1st thru June 15th?

 

 

I have spent hours looking for something like this for the tracking of the Spring snow cover melt.  I know this is a different thread.  But unless the Russians or Europeans have some sites I can not find holding snow depth data in a format like this I assume it doesn't exist.

 

I have seen graphics with say 60-100 locations in Eurasia with daily depth updates.  I assume most of those go into the GFS and Euro modeling of snow depth/cover. 

 

 

If it's possible to produce a one time image I can archive it.  If this is something that can be produced everyday like a model derived version of daily snow depth I might have to bite the bullet and learn how to make these graphics from GFS data. 

 

Would be much appreciated. 

 

[DTWXRISK]

 this  cannot   possibly  be close  to being correct...   sorry

 

given this IMAGE     https://9yizja.bl3302.livefilestore.com/y2pS1tAbhl_V9j2uBGPT8XmgAGfljwVVGuzJTJ9kpPhDvq9-Z51ZwxLO9ucL5vO-ytkviSn6VLlKQD__5HokWixdo_BcauRXcgJpjYbuhRTrWo/multisensor_4km_ea_snow_ice_map_syn_fulres_2014292.png

 

for  anyone   to claim that    snow south  of 60N  in 2014 is LESS  than 2o12 is   well ... BS
 

 

So I've found a surprising result after reading through Judah Cohen's 2011a paper about the Snow Advance Index. The proper lat/lon thresholds are 0-60N, 0-180E. After computing the snow-cover % (# of snow covered grid points/total # grid points), the resulting image is quite shocking. 2014 is clearly not as high as 2009 and 2012.  We are also seeing a reduction in the slope of d(snowcover)/dt... Most of the anomalous snow is in the 55-90N band... similar to 2006. I'm not sure what this means exactly, such that the snow-covered forcing is north of the Cohen region. Will need to do the research on this at a later time.

 

[GaWx]

GA - Total Eurasian snow cover is at 16.70 as of 10/23...

 

http://www.americanwx.com/bb/index.php/topic/44473-and-we-begin/?p=3095379

 

Grit,

 The graph suggests about the same SCE on 10/23 as 10/19. So, based on that non growth, 16.7 as of 10/23 would jibe with my 10/19 est. Although possible, I'm a bit surprised that it didn't grow further during then based on what the models were showing. Any opinions?

 

Edit: I can't wait for Rutgers to return!! This wait is way more painful than staying up past 2 AM EDT to get the Euro.

[griteater]

Yeah, not sure GA...but as DT mentioned, there are a few storms moving through to close out the month (with focus on the area south of 60N).  One in the far east, another rolls through central areas, then another thru central areas near Halloween...so, we should close out pretty well.

[weatherdude]

this  cannot   possibly  be close  to being correct...   sorry

 

given this IMAGE     https://9yizja.bl3302.livefilestore.com/y2pS1tAbhl_V9j2uBGPT8XmgAGfljwVVGuzJTJ9kpPhDvq9-Z51ZwxLO9ucL5vO-ytkviSn6VLlKQD__5HokWixdo_BcauRXcgJpjYbuhRTrWo/multisensor_4km_ea_snow_ice_map_syn_fulres_2014292.png

 

for  anyone   to claim that    snow south  of 60N  in 2014 is LESS  than 2o12 is   well ... BS

 

Call me crazy, but it looks like the graphing error may be in the "0-180E" being the wrong half of the hemisphere. 2012 was ahead of 2014 south of 60N in North America, not Eurasia:

 

[The_Global_Warmer]

Call me crazy, but it looks like the graphing error may be in the "0-180E" being the wrong half of the hemisphere. 2012 was ahead of 2014 south of 60N in North America, not Eurasia:

 

 

 

I'd think 2006 would be much higher if that was the case.

 

[InstantWeatherMaps]

For what it's worth, when I include latitudes north of 60N, 2014 is clearly the king. Just this analysis does not highlight big snow cover south of 60N.

 

  0-90N, 0-180E:

 

 

0-60N, 0-180E:

 

I'm sorry but your graphs are mathematically impossible. The area of a longitudinal portion of spherical cap from a given latitude is equal to (1-sin(lat))*(area of identical longitudinal portion of hemisphere). You are claiming that the snow cover excluding 60N-90N is under 10%, and the snow cover including those latitudes is over 30%. Thus, you are saying that roughly (0.30-0.10*sin(pi/3))/(1-sin(pi/3)) of the area from 60N to 90N is covered in snow. That is around 160%... clearly not possible. If I had to guess I'd say you're not weighting your area correctly, but even that doesn't explain everything. Try remaking these graphs (and your snow depth average graph) using the aave() function in GrADS (which is properly area-weighted) and see what happens. I understand you're not using GrADS here but you may want to try, or see if the software you use has an equivalent function.

[heavy_wx]

Here's the snow cover percentage on 31 October 2012 from the 20th Century Reanalysis v2. The extent listed in the image is calculated by multiplying the area of each lat-lon cell by the fractional percentage of snow cover. Obviously, the resolution is significantly degraded with these data which may account for some of the differences with respect to the Rutgers dataset.

 

 

I've shown October 2009 for comparison, having one of the highest SCE values in recent years.

 

 

To get an estimate of the current snow cover, I used the field initialized in last night's 00z GDAS. The snow cover field in this model has greater resolution but is binary and thus not completely comparable to the above plots. The discrepancies between these values and those posted by cfbaggett are probably due to the process of assimilating snow cover data into the models. Also, I'm not entirely sure how snow cover is defined for a particular location from the remote sensing observations, but it is likely not consistent with the values derived from percentage cover.

 

[Absolute Humidity]

The Oct 2012 map seems a bit off. I for one do not remember having snow cover on Oct 31 2012; Nov 7th yes but that was from the son of Sandy. I don't think Sandy's snowfall was nearly that expansive. Maybe I'm missing something?

[heavy_wx]

The Oct 2012 map seems a bit off. I for one do not remember having snow cover on Oct 31 2012; Nov 7th yes but that was from the son of Sandy. I don't think Sandy's snowfall was nearly that expansive. Maybe I'm missing something?

 

Part of that is the resolution of the model; the small amount of snow cover over PA projects onto some of those grid boxes. Also, NCL draws those cells with the mid point centered on the specific lat/lon coordinate of the data so they will extend somewhat beyond where they "should".