October 2008 – NOAA vs MSU

Today, I’m posting up plots comparing the MSU gridded data and the NOAA land-sea gridded data. On an overall basis, MSU (GLB) is running about 0.2 deg C cooler than NOAA land-sea gridded in October 2008. These differences are quite volatile and this discrepancy is not unusual.

First here is October MSU (centered on 1980-2000) to facilitate comparison with the NOAA land-sea version (which I’ve also centered on 1980-2000.)

Next here is October NOAA land-sea (re-centered on 1980-2000).

While the major patterns are similar in both data sets, there are some intriguing differences. NOAA surface coverage of the Southern Hemisphere looks decidedly scrappy beside UAH. While both NOAA and UAH show a cool “ridge” in the ocean offshore western Europe, it’s much larger in the UAH version. The tropical Pacific and northern Pacific are cooler in UAH. But it’s not all the same direction – UAH western Canada is warmer than NOAA western Canada.


46 Comments

  1. Paul
    Posted Nov 24, 2008 at 10:46 AM | Permalink | Reply

    Would it be possible to grid plot the differences between the two series?

  2. Posted Nov 24, 2008 at 10:54 AM | Permalink | Reply

    This is an important comparison after the emphasis has just been on the bruhaha around the Giss-Nasa October data. NOAA-NCDC is the one with the strongest warming, according to my Excel
    NOAA-NCDC 0.163 K per decade,
    Giss 0.161 K
    HatCrut3 0.16 K
    RSS 0.159 K (corrected down from 0.169 K! in September)
    UAH 0.127 K per daceade.

    This shows that not just UAH tends cooler as of October 2008, even Mears’ RSS satellites fit into a decadal warming of less than Noaa (0.159 vs. 0.163) after their October correction. (Basis period = 1979-2008)

  3. Pierre Gosselin
    Posted Nov 24, 2008 at 11:05 AM | Permalink | Reply

    Different instruments yield different readings.
    Where can I find information on the NOAA measuring instruments?

  4. dh
    Posted Nov 24, 2008 at 11:24 AM | Permalink | Reply

    These maps are for “land centric” observers (ie centered on 0 longitude).
    Some say that the climate is determuned by the oceans.
    How about showing a map centered on the Pacific ocean (ie longitude 180)?

  5. Carl Gullans
    Posted Nov 24, 2008 at 11:50 AM | Permalink | Reply

    I’m not familiar with this data, so I apologize if this is obvious, but is there a reason why the NOAA data shows virtually no anamolies in Antarctica vs. the large hot and cold anamolies in the MSU antarctica data?

  6. Posted Nov 24, 2008 at 12:30 PM | Permalink | Reply

    Steve McIntyre write: “there are some intriguing differences”.
    I already tried to explain those differences here and here are the two figures where the differences are marked.
    I’m posting again my old comment on discrepancies between HadCRU and RSS, hoping someone will note it.

    1) Barrow: all Alaska was very colder than normal with the exception of the northen coast. Possible explanations: more frequent wind from the sea to the north which is usualy warmer; UHI problems; later sea ice formation or snow cover in that area.
    2) In the middle of Labrador: surface station problem.
    3) USA Pacific waters: colder water with warmer air aloft; typical sea-air decoupling, wrong CRU’s assumption regarding near coast waters.
    4) Eastern North Atlantic: cold air advection onto warm sea in the extratropics. Nothing unusual, wrong CRU’s assumption.
    5) Italy-Adriatic sea: colder water with warmer air aloft; typical sea-air decoupling, wrong CRU’s assumption regarding near coast waters.
    6) Eastern Equatorial Pacific: was sst so warmer than normal there? Anyway air temperature was normal. Further investigation.
    7) Waters off Somalia: probable strong upwelling. Colder water with warmer air aloft; typical sea-air decoupling, wrong CRU’s assumption regarding near coast waters.
    8) Waters south of Java: colder water with warmer air aloft; typical sea-air decoupling.
    9) Eastern Australia: cold air advection onto warm sea in the extratropics. Nothing unusual, wrong CRU’s assumption.
    10) Far Indian Ocean to the south of Madagascar: cold air advection onto warm sea in the extratropics. Nothing unusual, wrong CRU’s assumption.
    11)Waters off Chile: cold air advection onto warm sea in the extratropics. Nothing unusual, wrong CRU’s assumption.
    12)Argentina: it’s hard to understand what’s going on there. The only reason to have warmer temperature at the ground and colder aloft, I think, is a problem with the surface stations (UHI, omogenisation).

    What’s the effect of all these assumptions on global mean?

  7. Regular Lurker
    Posted Nov 24, 2008 at 12:53 PM | Permalink | Reply

    The norwegian met office seems to agree more with UAH; western parts of Norway was colder than normal, and northern Norway a bit warmer.

    Se here

  8. Hank
    Posted Nov 24, 2008 at 1:03 PM | Permalink | Reply

    I am guessing GLB is short for global and comes from a filename.

  9. jae
    Posted Nov 24, 2008 at 1:06 PM | Permalink | Reply

    Has NOAA “lost” Australia?

  10. jae
    Posted Nov 24, 2008 at 1:07 PM | Permalink | Reply

    Aaarg, I meant Antartica, not Australia.

  11. Gary
    Posted Nov 24, 2008 at 1:11 PM | Permalink | Reply

    I would love to see the raw NOAA data for Australia, for the southern states had the coldest Oct for a long time. As an aside southern Australia had blizzard conditions 6 days out from start of summer.

  12. Bruce
    Posted Nov 24, 2008 at 1:29 PM | Permalink | Reply

    Steve, your UAH graph looks more dramatic than the UAH actual graph.

    http://climate.uah.edu/maps/1008big.jpg

    Is it just because they chose white for -.5 to +.5? Or is it the baseline you chose?

  13. Jim B
    Posted Nov 24, 2008 at 1:48 PM | Permalink | Reply

    Well for my small part of Western Canada, aka Edmonton. It’s very warm end of November and still no snow. Not that I’m complaining it’s really great! But much warmer than normal.

  14. Posted Nov 24, 2008 at 2:53 PM | Permalink | Reply

    It’s quite annoying to see a .02 to .05 anomaly for Australia and the east coast when New South Wales has had (reported in some media sources) the coldest October in 65 years and now we’ve had unusual snow storms in some areas just days away from the start of summer. That’s what averaging does I guess…

  15. John A
    Posted Nov 24, 2008 at 4:11 PM | Permalink | Reply

    On the eastern coast of Australia, they’re complaining about the unnatural cold for this time of year (end of Spring, beginning of summer). In the last few days its actually snowed in the State of Victoria in the Snowy Mountains.

  16. Posted Nov 24, 2008 at 4:29 PM | Permalink | Reply

    Nice Mollweide projections! (I think someone pointed out last year that the horizontal black bar is caused by the program’s trying to connect the Bering Straights the wrong way around.)

    The pale yellow neutral color in the NOAA graph is a little hard to tell from the white that indicates no information. (Note that the NOAA instrumental temperatures have a lot more missing data than the UAH satellite numbers).

    I hope that future graphics might try a highly chromatic rainbow temperature scale, with bright yellow near neutral — say Purple/Violet/Blue/Teal/Green/Chartreuse/Yellow/Y-Orange/Orange/Red/Maroon/Brick. White would then indicate no information, and/or pale shades of these would indicate little information.

  17. Mark Duffett
    Posted Nov 24, 2008 at 4:37 PM | Permalink | Reply

    For those looking askance at Australia above, here’s the Australian Bureau of Meteorology’s compilation for October 2008:

    Obviously it’s gridded so there’s still some spatial averaging, but much less than in the global CRU/NOAA compilations.

  18. Steve McIntyre
    Posted Nov 24, 2008 at 4:38 PM | Permalink | Reply

    HEre’s how I do the Mollweide projection. I make a grid with four columns lat, long, x and col (x being the value of the variable and col being the color calculated from the break points. library(mapproj) is required.

    mollweide=function(grid,options=”cru”) {
    map(“world”,proj=”mollweide”,fill=TRUE,col=”white”);
    index=(1:nrow(grid))[!is.na(grid$x)]
    for(i in index){
    if(grid$long[i]== -172.5) grid$long[i]=grid$long[i]+.2 #weird little patch for Mollweide left boundary polygon(mapproject(list(x=c(grid$long[i]-2.5,grid$long[i]-2.5,grid$long[i]+2.5,grid$long[i]+2.5),
    y=c(grid$lat[i]-2.5,grid$lat[i]+2.5,grid$lat[i]+2.5,grid$lat[i]-2.5) )),col=grid$col[i],border=grid$col[i])
    }
    map(“world”,proj=”mollweide”,fill=FALSE,add=TRUE);
    }

    mollweide(grid)

    There should be a neater way of coloring in the squares than what I’ve done here, but it works. One problem with mapproj is that I don’t know how to control the white space aroud the map which ends up being huge and thus requires manual trimming. The color code was inserted using the points function with pch =16, cex=3 yielding the squares and the legend by the axis command. So it’s a bit of a pastiche.

  19. Steve McIntyre
    Posted Nov 24, 2008 at 4:43 PM | Permalink | Reply

    Here are the Australian sites that are in the GHCN update (distributed by GISS). The vast majority are airports and many are urban. The sites designated R may or may not be “rural” – I’m sure that people can inform us on that. NASA GISS will more or less coerce the overall trend to coincide with the trend from the R-sites.

    country id name lat long airport urban end_raw
    5332 501 50194120000 DARWIN AIRPOR -12.40 130.87 A U 2008.750
    5342 501 50194150000 GOVE AIRPORT -12.27 136.82 A R 2008.750
    5352 501 50194203000 BROOME AIRPOR -17.95 122.22 x R 2008.750
    5356 501 50194212000 HALLS CREEK A -18.22 127.65 A R 2008.750
    5367 501 50194238000 TENNANT CREEK MO -19.63 134.18 A R 2008.750
    5386 501 50194287000 CAIRNS AIRPOR -16.88 145.75 A S 2008.750
    5390 501 50194294000 TOWNSVILLE AM -19.25 146.75 A U 2008.750
    5392 501 50194300000 CARNARVON AIR -24.87 113.67 A R 2008.750
    5393 501 50194302000 LEARMONTH AIR -22.23 114.08 A R 2008.750
    5400 501 50194312000 PORT HEDLAND -20.10 119.57 A S 2008.750
    5412 501 50194326000 ALICE SPRINGS -23.80 133.88 A S 2008.750
    5415 501 50194332000 MT ISA AIRPOR -20.67 139.48 A S 2008.750
    5423 501 50194346000 LONGREACH AIR -23.43 144.27 x R 2008.750
    5432 501 50194367000 MACKAY -21.12 149.22 x S 2008.750
    5436 501 50194374000 ROCKHAMPTON A -23.38 150.47 A U 2008.750
    5442 501 50194380000 GLADSTONE -23.85 151.25 x S 2008.750
    5450 501 50194403000 GERALDTON AIR -28.78 114.70 A S 2008.750
    5461 501 50194430000 MEEKATHARRA A -26.60 118.53 A R 2008.750
    5470 501 50194461000 GILES -25.03 128.28 A R 2008.750
    5485 501 50194510000 CHARLEVILLE A -26.40 146.27 A R 2008.750
    5524 501 50194578000 BRISBANE/EAGLE FARM AUSTRA -27.40 153.10 A U 2008.750
    5543 501 50194610000 PERTH AIRPORT -31.90 116.00 A U 2008.750
    5570 501 50194637000 KALGOORLIE BO -30.78 121.45 A S 2008.750
    5572 501 50194638000 ESPERANCE -33.82 121.88 x R 2008.750
    5578 501 50194653000 CEDUNA AIRPOR -32.12 133.70 A R 2008.750
    5583 501 50194659000 WOOMERA AEROD -31.13 136.82 A R 2008.750
    5596 501 50194672000 ADELAIDE AIRP -34.93 138.52 A U 2008.750
    5614 501 50194693000 MILDURA AIRPO -34.22 142.08 A S 2008.750
    5629 501 50194711000 COBAR -31.48 145.82 x R 2008.750
    5675 501 50194767000 SYDNEY AIRPOR -33.95 151.18 A U 2008.750
    5683 501 50194776000 WILLIAMTOWN -32.78 151.82 A U 2008.750
    5692 501 50194791000 COFFS HARBOUR -30.32 153.12 A S 2008.750
    5694 501 50194802000 ALBANY (ALBANY A.M.O.) -34.95 117.80 x S 2008.667
    5708 501 50194821000 MT GAMBIER AI -37.73 140.78 A S 2008.750
    5746 501 50194865000 LAVERTON AERO -37.85 144.73 A U 2008.750
    5775 501 50194907000 EAST SALE AER -38.10 147.13 A S 2008.750
    5777 501 50194910000 WAGGA AIRPORT -35.15 147.45 A S 2008.750
    5788 501 50194926000 CANBERRA AIRP -35.30 149.18 A U 2008.750
    5811 501 50194968000 LAUNCESTON AI -41.53 147.20 A S 2008.750
    5818 501 50194975000 HOBART AIRPOR -42.83 147.48 A U 2008.750
    5824 501 50194995000 LORD HOWE ISL -31.53 159.07 A R 2008.750
    5825 501 50194998000 MACQUARIE ISL -54.48 158.95 x R 2008.750
    5854 501 50195646000 FORREST (FORREST AMO) -30.83 128.12 A R 2008.750

    Here is the “R(ural)”-subset:
    country id name lat long airport urban end_raw
    5342 501 50194150000 GOVE AIRPORT -12.27 136.82 A R 2008.75
    5352 501 50194203000 BROOME AIRPOR -17.95 122.22 x R 2008.75
    5356 501 50194212000 HALLS CREEK A -18.22 127.65 A R 2008.75
    5367 501 50194238000 TENNANT CREEK MO -19.63 134.18 A R 2008.75
    5392 501 50194300000 CARNARVON AIR -24.87 113.67 A R 2008.75
    5393 501 50194302000 LEARMONTH AIR -22.23 114.08 A R 2008.75
    5423 501 50194346000 LONGREACH AIR -23.43 144.27 x R 2008.75
    5461 501 50194430000 MEEKATHARRA A -26.60 118.53 A R 2008.75
    5470 501 50194461000 GILES -25.03 128.28 A R 2008.75
    5485 501 50194510000 CHARLEVILLE A -26.40 146.27 A R 2008.75
    5572 501 50194638000 ESPERANCE -33.82 121.88 x R 2008.75
    5578 501 50194653000 CEDUNA AIRPOR -32.12 133.70 A R 2008.75
    5583 501 50194659000 WOOMERA AEROD -31.13 136.82 A R 2008.75
    5629 501 50194711000 COBAR -31.48 145.82 x R 2008.75
    5824 501 50194995000 LORD HOWE ISL -31.53 159.07 A R 2008.75
    5825 501 50194998000 MACQUARIE ISL -54.48 158.95 x R 2008.75
    5854 501 50195646000 FORREST (FORREST AMO) -30.83 128.12 A R 2008.75

    • Geoff Sherrington
      Posted Nov 29, 2008 at 1:15 AM | Permalink | Reply

      Re: Steve McIntyre (#19),

      Steve, Thank you for the attention you are giving to Australia.

      The 15 or so stations of the High Quality network in #19 were chosen, we assume, because of a lack of UHI. Their raw data are available but I lack the skill any more to process them.

      A composite graph of these 15 relevant stations should have some resemblance to the shglfulihad of “Mann & Perfect Reconstruction” of Nov 28 or at least it should raise questions about the differences. Anyone wish to help with this simple task?

      I am still confised whether the levelling of NH and SH temps since 1998 has been because of the increasing % use of rural sites. The test above would shed some more light on that confusion.

    • Geoff Sherrington
      Posted Dec 4, 2008 at 7:46 AM | Permalink | Reply

      Re: Steve McIntyre (#19),

      There are 17 rural sites forthe Australian High Quality network. Two are far-distant islands, leaving 15. I have started to do some plotting of the daily raw data, max and min temps. I have not included metadata corrections and I have occasionally infilled a missing value with a value about consistent with surrounding. The infilling makes no significant difference to the result, it just makes my calculations faster.

      I started with Broome, on the NW coast of Australia, mormal population about 15,000 and swelled by tourism. The weather station at the airport is surrounded by about 2 km of housing on most sides, but is on a small peninsula with sea on 3 sides. Lat -17.95, Long 122.22.

      There follows a plot of the max and min temperatures from Jan 1940 to Dec 2006.

      One does not need to run regressions to see that the maximum temperature annual averages (pink) have stayed about constant and the minimums (green) have decreased. Note that we have a blip upwards in 1998 also, as commonly elsewhere. I wonder why. There is a fairly constant 11 degrees centigrade between the annual max and the annual min.

      The question is, “WHERE HAS ALL THE GLOBAL WARMING GONE?” Did Broome escape the omnipresent GHG warming because film celebrities like to stay there?

      To be continued.

  20. Josh
    Posted Nov 24, 2008 at 5:06 PM | Permalink | Reply

    snip – no need to editorialize.

  21. Mark Duffett
    Posted Nov 24, 2008 at 5:36 PM | Permalink | Reply

    Hmmm, in this instance it looks like the R stands for ‘remote’ (very) rather than just ‘rural’. The ‘R’ sites taken by themselves would severely under-represent Australian climate (only five in southern Australia). To use them alone to estimate trends would be throw out quality data babies with the UHI bathwater. I’d hope they at least use the ‘S’ stations as well (not sure what S stands for), most of which would be at airports well beyond (small) town boundaries, very much within areas Australians would consider ‘rural’. Even some of the ‘U’ (for ‘urban’, presumably) stations are in decidedly rural settings, such as the one I’m familiar with (Hobart airport, which is well away from the city).

  22. James Lane
    Posted Nov 24, 2008 at 6:06 PM | Permalink | Reply

    Assuming that S=Semi rural, the classifications all look reasonable to me.

    What’s interesting is that south-eastern Australia is not well-represented on the list, and there are no rural stations in south-eastern Australia. Further, there is not a single rural station on the entire east coast. It’s weird as there must be dozens of reporting stations at rural (small town) airports on the east coast, and plenty of lighthouses. The east coast is climatically quite different from the rest of Australia as it’s separated from the interior by the Great Dividing Range.

    Almost all of the mainland rural sites are desert locations.

  23. Posted Nov 24, 2008 at 6:25 PM | Permalink | Reply

    Here is the access page for Australia’s Reference Climate Station Network:

    http://www.bom.gov.au/climate/change/reference.shtml

    You will find photo records of the surface stations linked to from there – they are not as well documented as Anthony’s Surface Stations project, but at least they have made a start.

    A bit of time spent with Google Earth should enable better views of the surrounding areas around many of them.

  24. Craig
    Posted Nov 24, 2008 at 7:49 PM | Permalink | Reply

    Re your original post:

    I recently saw some colour maps for CO2 levels in 2003 an article here: http://www.scientificblogging.com/news_releases/forget_oil_says_james_hansen_cleaner_coal_emissions_will_cure_global_warming

    What’s interesting is that concentrated CO2 emissions seem to occur in bands but extend over the ocean (drift?). CO2 absorption has intense blue over antarctica. Maybe the CO2 concentrations and temperature could be compared by someone who can create these colour mappings? It’s beyond me at present.

  25. Craig
    Posted Nov 24, 2008 at 7:56 PM | Permalink | Reply

    Re Post # 19:

    I agree with #22. Esperance (remote southern coastal Western Australia), Ceduna is remote coastal (southern – South Australia), Broome and Carnarvon are remote and coastal but in northern Western Australia. Geraldton could be considered remote from Perth, but it is not shown as “R”. It is less remote than Broome and Carnarvon. Meekatharra and Giles are remote, inland and desert. Giles is much further inland (central Australia, near the Northern Territory border) and is a long-established meteorological station.

    I haven’t yet found explanatory information by the authors on what these codes are used for, but it would be worth finding where they do explain their field codes(and this applies to GHCN/GISS data generally). See also the information on Western Australian GHCN files I posted at #79 on “Notes on GISS Station Data” thread. http://www.climateaudit.org/?p=1956.

  26. Craig
    Posted Nov 24, 2008 at 7:59 PM | Permalink | Reply

    There’s a useful tool for examining neighbouring station temperature data. By playing around with the python parameters, you can display on the one graph the GISS temperature data for sites within a range of the station of interest. For example, if you want to see Mandurah (Western Australia), you would go to

    http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=501946050010&data_set=0&num_neighbors=1

    By changing the URL parameter num_neighbours to higher numbers, you can display sites in expanding areas. Putting it at 2 gives you another 30km (approximately), 3 gives you 60km etc.

    Steve, by using the station numbers in your list, you can read off the locations that are listed at the top of the generated graph. For example, put num_neighbors=4 for Mandurah and you will get a dense graph, which includes all the Perth airport stations (501946100000, 501946100001 etc). There are a band of stations in the 16 to 19 degree range, but then only three (Karnet, Dwellingup, Wandering) that are in the 14 to 16 degree range. The 5 degree difference seems to occur within 90km. Most of the higher temperatures are coastal.

    I put num_neighbours = 4 for Geraldton and you get a graph showing stations up to 200km away: http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=501944030000&data_set=0&num_neighbors=4

  27. SteveSadlov
    Posted Nov 24, 2008 at 9:20 PM | Permalink | Reply

    Actual weather patterns experienced suggest that UAH is more accurate.

  28. Craig
    Posted Nov 24, 2008 at 9:46 PM | Permalink | Reply

    Re #19 and #25

    The R,S and U codes seem to be based on population data.

    P =R if rural (not associated with a town of >10,000 population)
    S if associated with a small town (10,000-50,000 population)
    U if associated with an urban area (>50,000 population)

    These codes and the population data to which they related are contained in the station inventory file (shown to the nearest thousand). That is the file I was looking at here #79. The legend in that file is useful. The “P” field contains R, S and U codes that seem to be used in the same way as the “urban” field Steve has cited.

    Steve: This has been discussed before. Interested parties should look at the Surface Record – Hansen GISS category.

  29. Craig
    Posted Nov 24, 2008 at 10:19 PM | Permalink | Reply

    Steve

    Re your comment at #19:

    NASA GISS will more or less coerce the overall trend to coincide with the trend from the R-sites.

    My understanding is that you are referring to what NASA do with the “U” (urban) and “S” (small town) data in step 2 as described here. In that section they say:

    The goal of the homogeneization effort is to avoid any impact (warming or cooling) of the changing environment that some stations experienced by changing the long term trend of any non-rural station to match the long term trend of their rural neighbors, while retaining the short term monthly and annual variations. If no such neighbors exist, the station is completely dropped, if the rural records are shorter, part of the non-rural record is dropped.

    Any town or small town with no rural neighbours is “completely dropped”! This seems to indicate a preference for the kind of data retained before it is understood. I do not know yet, but this might force isolated or fringe towns to be eliminated, with a bias toward rural or towns urban sprawl.

    The station inventory has a separate field for brightness (presumably the brightness of the station area as seen from space at night) ie. A= “dark”, B=”dim” and C=”bright”. At this stage, I do not understand the relevance to Step 2 of the R,S,U (rural/non-rural) categories and the separate station “brightness” category. There would seem to be a need to elaborate on the precise steps taken in Step 2 as described above and as described here.

    Have you or anyone else covered this elsewhere or is it that someone needs to do?

    Steve: This was been partly worked through last year.

  30. Willis Eschenbach
    Posted Nov 24, 2008 at 10:56 PM | Permalink | Reply

    Steve M., the dark lines in your plots above across the top of the drawing are from bits of Russia or Alaska that extend across the dateline. I think they can be removed by

    map(‘world’, projection=’mollweide’,orient=c(90,0,0),interior=F,wrap=T)

    where (if I recall correctly) the “wrap=T” does the work of removing the lines.

    Also, I use

    par(xaxp=c(-180,180,12),yaxp=c(-90,90,6),xaxs=”r”)

    which I believe removes some of the white space surrounding the plot.

    You can also use

    par(mai=c(.25,.25,.25,.25))

    or whatever values you want, which sets the margin size in inches. I use this with

    par(mgp=c(2,1,0))

    to bring the titles in closer to the graphs.

    w.

  31. Norm
    Posted Nov 25, 2008 at 12:48 AM | Permalink | Reply

    But it’s not all the same direction – UAH western Canada is warmer than NOAA western Canada.

    Have to agree with Jim B. in #13. The weather all Fall here in Calgary (300 km south of Edmonton) has been extremely nice, and also no snow. Halloween here was the warmest in over 50 years.

  32. Craig
    Posted Nov 25, 2008 at 3:48 AM | Permalink | Reply

    Re #19.

    Okay, I can give an answer to whether the “rural” GISS classifications are accurate for most of Steve’s subset.

    I realise I was not across all previous posts on that general issue, so I have now skimmed the Surface Record posts on this blog. I now understand towns classified as rural had a pop code of “-9″ in the relevant station inventories used by NASA, so it appears they only checked to see if they were over 10,000 and if they weren’t didn’t record anything. We are talking some time prior to 1993?

    I would expect the best population data to be held by the Australian Bureau of Statistics. There is a site that uses this data and lists “urban centres” with population > 10,000. See Thomas Brinkhoff’s site here.

    Brinkoff’s site has Australian urban centre population information for 1996, 2001 and 2006. Based on the information at that site, Ceduna, Learmonth, Longreach, Carnarvon, Meekatharra, Giles, Esperance were still 28,000. Port Hedland was also “small town” with a population figure of 13,000. In 2006 it was still about 11,000. Albany Town was 15,000 but is now >25,000.

    Some other examples not listed by Steve at #19 but which should change from rural to “small town” in the GISS classification: Broome is now >10000 (15,906 from 1996). Karratha has also just nudged past 10,000 (it was listed as rural).

    Also, note that Bunbury, a rapidly growing urban centre (from 25,000 in 1996 to 55,000 in 2006) is not even listed by GISS. I am not sure if it has weather station data, but I will check the B of M and post later.

    In Western Australia there has also been rapid growth in towns in the north west in the last couple of years so it is likely the 2006 population figures will need further checking.

  33. Posted Nov 25, 2008 at 7:22 AM | Permalink | Reply

    The full Australian Daily climatic data archive is now online here:

    http://www.australianweathernews.com/recent_AWN_daydataArchive_element.html

  34. Bob Koss
    Posted Nov 25, 2008 at 8:33 AM | Permalink | Reply

    I decided to download the GISS map projection software and try it out. Works pretty well. Kudos to them for making such a nice tool available. 70+ projection styles are available. Original map must be equirectangular. Doesn’t do animation.

    I then used the GISS maps page to create a land/ocean anomaly map for Oct 2008 using the same 1980-2000 base period as Steve. The graphics below are the original map and an animated Mollweide projection of the same data that rotates in 90 degree increments.

    I added the overlay of lines for the Tropic of Cancer and Capricorn which enclose 40% of the earth surface. Also the Arctic and Antarctic Circle each of which enclose 4% of the earth surface along with a Prime Meridian line. I figure they aid in getting a perspective on the relative size of hot/cold spots to the size of the earth.

    The GISS color range is narrower than the one used in Steve’s Graphics for UAH and NOAA.

    • Aaron Wells
      Posted Nov 25, 2008 at 10:55 AM | Permalink | Reply

      Re: Bob Koss (#34),

      Very nice work Bob. That is far superior to an equirectangular map (IMHO). I wish GISS would offer that as a turnkey option.

    • Geoff Sherrington
      Posted Nov 26, 2008 at 3:17 AM | Permalink | Reply

      Re: Bob Koss (#34),

      There are still distortions, but I’m not saying they are caused by your hard work.

      The Antarctic peninsula sth of Patagonia shows an area of warming about the same size and intensity as the whole of Australia. I think there might be 2 data stations on the former, please correct me if I am wrong.

      Re: James Lane (#22),

      Too many desert locations. This has been reported before. More pronounced if you subtract urban Lord Howe Is and Macquarie Island from the set, as both are about 2,000 km from the nearest mainland Australia and are Australian by accident of history rather than geography.

      Of the remaining 15 urban in Steve McIntyre (#19), only 3 (Broome, Gove and Esperance, all on the coast) receive significant annual rainfall. Which makes one wonder about the definition of a drought year in a desert. And stats distributions which have a cutoff of zero some years, not normal.

      Regarding the urban locations that need UHI adjustment, about half are on the coast, which roughly means that a radius search for nearest neighbours will find only 50% of stations because half the search area is unpopulated sea.

      It’s a problem. As soon as the BOM finds a nice rural site, people go to live there and it becomes urban before adequate years of records can be gathered. (Heard in a new shopping centre: “I’m sure this centre would be more popular if there were not so many people here.”)

      Final problem. Who knows what GISS or NOAA or Hadley do with the data that comes from the Australian Bureau of Met? Nobody seems to want to tell.

      • Bob Koss
        Posted Nov 26, 2008 at 6:45 AM | Permalink | Reply

        Re: Geoff Sherrington (#41), actually there are about a 1/2 dozen closely located stations at the end of the peninsula.

        Early 2008 I made this KML file for use in Google Earth. Anyone with Google Earth can can download it and see where all the GISS stations used in 2006-2007 are located. There is a readme file along with it that contains a usage explanation.

        I posted a link to it about 6 months ago, but that link no longer works.

  35. rafa
    Posted Nov 25, 2008 at 10:18 AM | Permalink | Reply

    Re.: 7, on the contrary the spanish met office seems to agree more with GISS. See Steve’s graphs and here for October temp’s in Spain. AFAIK data for Spain is centered 1971-2000, don’t ask me why

    best

  36. Gary
    Posted Nov 25, 2008 at 1:08 PM | Permalink | Reply

    Steve, thanks for the Australia list of sites. I got the raw data to compare Oct 07 to Oct 08.
    The result was Oct 07 average T 20.87 C
    Oct 08 average T 20.99 C
    A rise of 0.12C
    So why did we all feel colder? About 90% of the Aust population lives east of longitude 144. East of 144 the difference is
    -0.6. West of 144 it is +0.74. A few more points in SE Australia and suddenly the difference becomes negative. Suggests the rise is not statistically significant.
    How does 0.12C compare to the GISS anomoly difference between Oct 07 and 08?

  37. Sam Urbinto
    Posted Nov 25, 2008 at 3:34 PM | Permalink | Reply

    Why does mother nature hate us so? Look at all the environmental catastrophe all around us! Oh, sorry. Wrong blog. Ahem.

  38. Jeff Norman
    Posted Nov 25, 2008 at 3:54 PM | Permalink | Reply

    Bob Koss,

    Brilliant map work. Thank you.

  39. Craig
    Posted Nov 26, 2008 at 3:48 AM | Permalink | Reply

    Re #32

    Based on the information at that site, Ceduna, Learmonth, Longreach, Carnarvon, Meekatharra, Giles, Esperance were still 28,000.

    Note sure what happened with that post. What I meant to say was that based on the information at that site, the respective populations for Ceduna, Learmonth, Longreach, Carnarvon, Meekatharra, Giles, Esperance were still <10,000 in 2006. They were classified as rural on that basis pre 1993 and it seems that category is still appropriate.

    Thanks to Carl and Len’s links, I’ve confirmed that Bunbury (Western Australia) has a long temperature record going back to 1887, even though a post office, and power station served as station locations before the present site. However, it’s not part of the GISS data.

  40. John M
    Posted Dec 4, 2008 at 6:40 PM | Permalink | Reply

    November RSS v 3.2 is 0.216.

    Oct was 0.181 and Nov 2007 was 0.131.

Post a Comment

Required fields are marked *

*
*

Follow

Get every new post delivered to your Inbox.

Join 2,882 other followers

%d bloggers like this: