Tuesday, February 26, 2013

January 2013 GISS up 0.17°C



I posted earlier about problems with GISS. They have now posted the January anomaly, and it is 0.61°C, up from 0.44°C in December. That's a big rise, but less than the big drop in December. TempLS recorded a similar rise (0.15°C), and I blogged about that and the reasons here. NOAA also rose from 0.42°C to 0.54°C.

As usual below the fold I'll show the comaparative spatial distributions.



Here is the GISS map for January:

And here is the TempLS map (from here), which uses spherical harmonics:



Previous Months

December
November
October
September
August
July
June
May
April
March
February
January
December 2011
November
October
September
August 2011

More data and plots




Monday, February 18, 2013

Hurricanes and SST - movies


I have posted animations of daily high-resolution NOAA/NCDC satellite SST data in various regions, including the North Atlantic. There are patterns there that can be associated with known ocean fluid mechanics, but others that can't. So I extended the movies to include hurricane tracks in the N Atlantic in recent years. I was curious to see whether they would leave footprints of cooler water.

The answer seems to be - sometimes. For example, Gordon in 2012 left a marked trail. But our SST data is daily and hurricanes can move a long way in a day. I've shown just the midday (UTC) positions.

In the process, I've experimented with yet another visual mechanism. This time I've gone back to a sort of animated slide show of jpg's. This gives best quality and control, but no video compression. For the N Atlantic, this seems to work because the number of data points is not so large. The controls are explained below the graphic.

Update - I've added a highlight collection of the best streakers. Click on the bottom selection (below 2005), labelled "Tops".








Choose Year




Pause Loop




Stepmillisec

Details

Hurricanes and storms are marked by their midday (UTC) position. Track data comes from the NCDC 5.5Mb HURDAT2 data file. SST data is as described here.

Each storm is marked by a black circle with central color. The colors are grey for "Low", cyan for "Tropical Depression", pink for "Tropical Storm" and red for "Tropical Hurricane" or "Extratropical Cyclone".

You can select a year from the top dropdown. When you do, the requisite jpg files will be preloaded, which will take a second or two. It's likely the sequence will start at once; otherwise clicking on Pause should get it going. Loop toggles loop mode. The move buttons below move one step forward or back. This works best when paused.

There is a text box where you can enter the time you'd like each frame to show, in millisecs. It works immediately - indeed even as you type. But it won't let the time go below 100 millisec.

You can click on the timeline below the plot to go to a specific date.



Thursday, February 14, 2013

January TempLS Global Temp up 0.15°C


I posted earlier about a delay due to a small hiatus in GHCN data posting. They have posted new data, so I have done the regular January analysis.

The temperature anomaly is up to 0.40°C, which is a rise of 0.15°C from the low December value. The satellite indices showed even steeper rises.

The warm areas in the plot below show up as NW N America (very cold in December), Central Asia, and Australia.

Below is the graph (lat/lon) of temperature distribution for January. I've also included a count and map of the stations that have reported to this date.

Tuesday, February 12, 2013

High Resolution SST regional movies

Recently I posted some screencap videos derived from the WebGL displayed sea surface anomaly temperatures. These excellent data are supplied by NOAA/NCDC, based on their OISST processing of AVHRR satellite data.

I have now been able to mechanize the process of production. I no longer use screen grabs, which was a major holdup. So I have now posted a page which has a much greater number of videos installed, covering various regions of interest. Some of these are up to date, and I plan to maintain them so. You should be at any stage able to replay the last 50 days of ENSO, or the last 100 days at 2-day resolution etc. Same for Arctic, Antarctic and N Atlantic.

I've tinkered a lot to try to get something working in each browser type. There are various difficulties; IE and Safari support for the HTML5 movie tag is patchy. Currently Firefox and Chrome should see OGG videos with controls, while Safari and IE see .swf videos looping without controls. In both cases you should be able to select. There is now a little string at the bottom of the selection area that you can use to link to your current view. I'll use that in this post.

The movie sequences, apart from the ones ending now, are generally full years at 4 day intervals. For ENSO, we currently have 2012,2011,2010,2008,2007,2006,2005,1999 and 1998. For Arctic, 2012,2011,2010,2007 and 2000.

ENSO

The main point of interest is the strong jet which forms directly along the Equator. This is warm in El Nino, cold in La Nina. It shows marked character of an alternating vortex street. It's also interesting to see how far it goes, and how it dissipates. Here's the last 50 days

Update - as an experiment, I've tried longer, faster sequences. They didn't add to the download time nearly as much as I expected, and the results are very good. Here is 2012 at five days a second.

Arctic

The first notable feature is a large region at zero anomaly. This is ice. It isn't true that all ice is at zero anomaly, but it's mostly true that zero anomaly is ice. You can see the ice melt and refreeze. The melt is streaky because melt ponds etc may report a positive temperature, but the refreeze is clear and dramatic. Here's the 2010 annual movie.

You can also see warm SST that may be influencing melting. One thing to watch for recently is the persistence of clear water north of Svalbard, well into the winter.

Antarctic

Again there is the annual melt and freeze, without the geometric intricacy of the Arctic melt. But there are also interesting currents to watch, especially through the Drake passage. Here is the latest period

N Atlantic

The patterns, Gulf Stream etc are familiar, but show up dramatically here. On a finer scale it is possible to see some effects of hurricanes. I'll show soon a video of the SST footprint of Hurricane Sandy.




Browser wars


I've been busy making SST movies. It's automated now, and I can produce lots, and keep them up to date. But showing them can be a nightmare. There are technical issues, but they seem to be mainly commercial or patent. So I've been producing movies in swf, flv, ogg, mp4 and it's still hard to get something that works for everyone.

But my latest frustration was kinda amusing. I thought to make with Javascript a different player for each browser. A bit of extra coding, but I've got pretty familiar with them all now. Javascript provides a navigator object which returns a appCodeName and an appName for each browser. So I thought I'll just use that to switch. Just have to find out the names they use.



I use Firefox mostly, so I checked that. appCodeName=Mozilla, as expected. appName=Netscape - not so obvious, but I know the history.

Then I tried IE. Surprise. appCodeName=Mozilla! But at least appName=Microsoft Internet Explorer.

Then I tried Chrome. Worse. appCodeName=Mozilla, appName=Netscape. And Safari the same. So I'm stymied. Navigator will even return a detailed header, but I can't see anything there I can reliably use. There's likely to be more difference between browser versions than browser brands.

Anyway I found this amusing account of the history.

Update - I think I found a solution. Go with their strength!

They don't mind using each other's names. What they won't share is a video format. The new HTML5 video tag allows you to ask whether particular formats are supported.
IE supports MP4 but not WebM or OGG
FF supports WebM and OGG but not MP4
Safari supports only MP4 and
Chrome supports all 3.
So awarding 4 points for MP4, 2 for WebM and 1 for OGG we have the  scores:
IE 4, FF 3, Safari 4 and Chrome 7.

That still leaves IE and Safari. But IE at least uses its own appName.

Actually IE is helpful in another way. They are so non-standard that they provide a metalanguage, buried in HTML comments, that allows you to write IE-only blocks of code.

Temperature tracking issues


I'm late posting on temperature this month. But it isn't for lack of attention. Odd things are happening. GHCN hasn't updated since 5 February. One can get a good estimate on the first few days, but it will be subject to correction when more data comes in.

The other oddity is that GISS, which has been having server trouble, won't respond to my script query for the monthly global average. I can see it. And it has updated for December, but not January. My monitoring page doesn't show their December figure. I could intervene manually, but I'm reluctant to. I also don't intervene to manually put in the UAH numbers which Roy Spencer announces - I wait till the script can see them.



Anyway, the very early result is that TempLS is up a lot, from 0.25°C in December to 0.4°C in January. That's still not back to the November level. Here's the very early (in data) plot:





Friday, February 1, 2013

More on SST movies


This follows an earlier post which has .swf movies derived from the HiRes NOAA SST page. This has the WebGL rotatable globe with a shaded 1/4° plot and lots of daily data, courtesy of NOAA/NCDC and partners.

I've made some improvements, mainly to aid the showing and making of animations. Although the ordinary videos are easy to start up, I'm hoping to make the WebGL version equally accessible. The resolution is much better, especially with zooming, and, well, it's just fun to be able to rotate the globe while the animation is in progress.

So I've added a Pause button and a Step button. I've also systematised the buttons so each has pretty much a toggle function, and turns red when it's on. That helps, because the logic of how they all interact is now complicated, though I hope intuitive.



The mode of movie operation is still that you set up a sequence with the Seq button, prescribing step in days and total frames. This was slow at hi-res; it's better now, but too fast at low res. The Pause button will stop this if you change your mind. There's a counter for the frames.

Then Cycle with set interval in Secs (can be fraction) will show a proper movie in loop mode. Again Pause will pause it, and while paused, Step will go one frame at a time.

I'd recommend that you toggle off any buttons that you don't need.

There's an interesting new button that I've called Manual. It's to help with making flash movies. Like other buttons, it turns red when clicked, and in that state, allows control by keystrokes. j sets it to the start of a movie sequence and k steps by one. You can use it in this mode if you like. Note that it works on a focus system, so if you click elsewhere, focus is lost and the button goes pale and doesn't receive keystrokes any more.

But the real purpose is to allow another process to take over and synchronise. I mentioned in my last post that I found a nifty program, NirCmd, that will set up a loop that takes screenshots at prescribed intervals. It will also generate keystrokes, so I can ensure that each screenshot is of consecutive frames.

You'll notice a new timestamp top left. You can move this anywhere in the plot with a Shift-click.

Here is the complete DOS batch file that I use:

nircmd loop 54 3000 execmd ..\scor ~$loopcount$
convert *.png -crop 780x4000+30+80 -resize 75%% "w%%02d.jpg"
ffmpeg -y -sameq -r 2 -i "w%%02d.jpg" -r 2 -t 50 %dr%.swf

The first is the NirCmd line to run the loop (scor.bat). The second is ImageMagick to convert the resulting png's to cropped jpg's and the third makes the movie. All freely available utilities. The batch file scor.bat is just:

if %1==1 nircmd sendkeypress j
set /A a=%1+100
nircmd savescreenshot C:\mine\blog\experiments\screencap\try\scr%a%.png
nircmd sendkeypress k

The arg %1 is the loop counter; the first line forces the WebGL to the first frame, second just ensures that the output will be in alphabet order, then the screen shot, then the signal to step a frame.

I've added an animation of 2012 at 4-day steps. There will be more.