December 2016 Global Surface (Land+Ocean) and Lower Troposphere Temperature Anomaly Update – With a Look at the Year-End Annual Results

This post provides updates of the values for the three primary suppliers of global land+ocean surface temperature reconstructions—GISS, HADCRUT4 and NOAA NCEI (formerly NOAA NCDC)—and of the two suppliers of satellite-based lower troposphere temperature composites—RSS and UAH—all through December 2016. It also includes a few model-data comparisons.  Also included is a quick look at the annual data.

This is simply an update, but it includes a good amount of background information for those new to the datasets. Because it is an update, there is no overview or summary for this post. There are, however, simple monthly summaries for the individual datasets. So for those familiar with the datasets, simply fast-forward to the graphs and read the summaries under the headings of “Update”.  


Lots of chatter from the mainstream media about the “record high” annual surface temperatures in 2016, along with the claims of “three in a row”…meaning 2014, 2015 and 2016 were all warmest in sequence. Figure 00 compares the three primary land+ocean surface temperature reconstructions and two primary lower troposphere temperature composites on an annual basis, with the temperature anomalies referenced to a common period of 1981-2010.


Figure 00

Most of the news stories include mention of the record highs in 2016 being caused by the aftereffects of the 2014/15/16 El Niño. Not one that I’ve run across, however, have mentioned that the record high global sea surface temperatures in 2015 were driven primarily by the 2014/15 portion of the 2014/15/16 El Niño and by “The Blob” (a prolonged, naturally occurring, coupled ocean-atmosphere weather event) and that “The Blob” was the primary reason for record high sea surface temperatures in 2014. See General Discussions 2 and 3 of my most recent free ebook On Global Warming and the Illusion of Control (25MB).

For those wanting a brief look at the individual results on an annual basis, see Animation 1. It includes the comparison shown above, followed by the individual surface and lower troposphere temperature results.  The final two graphs are separate views of the 3 surface temperature reconstructions and the two lower troposphere temperature composites.


Animation 1

The additional individual graphs that make up the animation:

Back to your regularly scheduled monthly update.


We discussed and illustrated the impacts of the adjustments to surface temperature data in the posts:

The NOAA NCEI product is the new global land+ocean surface reconstruction with the manufactured warming presented in Karl et al. (2015). For summaries of the oddities found in the new NOAA ERSST.v4 “pause-buster” sea surface temperature data see the posts:

Even though the changes to the ERSST reconstruction since 1998 cannot be justified by the night marine air temperature product that was used as a reference for bias adjustments (See comparison graph here), and even though NOAA appears to have manipulated the parameters (tuning knobs) in their sea surface temperature model to produce high warming rates (See the post here), GISS also switched to the new “pause-buster” NCEI ERSST.v4 sea surface temperature reconstruction with their July 2015 update.

The UKMO also recently made adjustments to their HadCRUT4 product, but they are minor compared to the GISS and NCEI adjustments.

We’re using the UAH lower troposphere temperature anomalies Release 6.0 for this post as the paper that documents it has been accepted for publication. And for those who wish to whine about my portrayals of the changes to the UAH and to the GISS and NCEI products, see the post here.

The GISS LOTI surface temperature reconstruction and the two lower troposphere temperature composites are for the most recent month. The HADCRUT4 and NCEI products lag one month.

Much of the following text is boilerplate that has been updated for all products. The boilerplate is intended for those new to the presentation of global surface temperature anomalies.

Most of the graphs in the update start in 1979. That’s a commonly used start year for global temperature products because many of the satellite-based temperature composites start then.

We discussed why the three suppliers of surface temperature products use different base years for anomalies in chapter 1.25 – Many, But Not All, Climate Metrics Are Presented in Anomaly and in Absolute Forms of my free ebook On Global Warming and the Illusion of Control – Part 1 (25MB).

Since the July 2015 update, we’re using the UKMO’s HadCRUT4 reconstruction for the model-data comparisons using 61-month filters.

And I’ve resurrected the model-data 30-year trend comparison using the GISS Land-Ocean Temperature Index (LOTI) data.

For a continued change of pace, let’s start with the lower troposphere temperature data. I’ve left the illustration numbering as it was in the past when we began with the surface-based data.


Special sensors (microwave sounding units) aboard satellites have orbited the Earth since the late 1970s, allowing scientists to calculate the temperatures of the atmosphere at various heights above sea level (lower troposphere, mid troposphere, tropopause and lower stratosphere). The atmospheric temperature values are calculated from a series of satellites with overlapping operation periods, not from a single satellite. Because the atmospheric temperature products rely on numerous satellites, they are known as composites. The level nearest to the surface of the Earth is the lower troposphere. The lower troposphere temperature composite include the altitudes of zero to about 12,500 meters, but are most heavily weighted to the altitudes of less than 3000 meters. See the left-hand cell of the illustration here.

The monthly UAH lower troposphere temperature composite is the product of the Earth System Science Center of the University of Alabama in Huntsville (UAH). UAH provides the lower troposphere temperature anomalies broken down into numerous subsets. See the webpage here. The UAH lower troposphere temperature composite are supported by Christy et al. (2000) MSU Tropospheric Temperatures: Dataset Construction and Radiosonde Comparisons.  Additionally, Dr. Roy Spencer of UAH presents at his blog the monthly UAH TLT anomaly updates a few days before the release at the UAH website. Those posts are also regularly cross posted at WattsUpWithThat. UAH uses the base years of 1981-2010 for anomalies. The UAH lower troposphere temperature product is for the latitudes of 85S to 85N, which represent more than 99% of the surface of the globe.

The UAH lower troposphere data are now at Release 6. The paper that supports the latest release has been accepted for publication (no date yet set for publication), and the Release 6 data are no longer being published with a “beta” identifier. See Dr. Roy Spencer’s post here. Those Release 6.0 enhancements lowered the warming rates of their lower troposphere temperature anomalies. See Dr. Spencer’s blog post Version 6.0 of the UAH Temperature Dataset Released: New LT Trend = +0.11 C/decade and my blog post New UAH Lower Troposphere Temperature Data Show No Global Warming for More Than 18 Years. The UAH lower troposphere anomaly data, Release 6.0, through December 2016 are here.

Update: The December 2016 UAH (Release 6.0) lower troposphere temperature anomaly is +0.24 deg C.  It took a nosedive in December (a whopping decrease of about -0.21 deg C).


Figure 4 – UAH Lower Troposphere Temperature (TLT) Anomaly Composite – Release 6.0


Like the UAH lower troposphere temperature product, Remote Sensing Systems (RSS) calculates lower troposphere temperature anomalies from microwave sounding units aboard a series of NOAA satellites. RSS describes their product at the Upper Air Temperature webpage.   The RSS product is supported by Mears and Wentz (2009) Construction of the Remote Sensing Systems V3.2 Atmospheric Temperature Records from the MSU and AMSU Microwave Sounders. RSS also presents their lower troposphere temperature composite in various subsets. The land+ocean TLT values are here. Curiously, on that webpage, RSS lists the composite as extending from 82.5S to 82.5N, while on their Upper Air Temperature webpage linked above, they state:

We do not provide monthly means poleward of 82.5 degrees (or south of 70S for TLT) due to difficulties in merging measurements in these regions. 

Also see the RSS MSU & AMSU Time Series Trend Browse Tool. RSS uses the base years of 1979 to 1998 for anomalies.

Note: RSS recently release new versions of the mid-troposphere temperature (TMT) and lower stratosphere temperature (TLS) products.  So far, their lower troposphere temperature product has not been updated to this new version.

Update: The December 2016 RSS lower troposphere temperature anomaly is +0.23 deg C.  It also dropped a good measure (a noticeable downtick of -0.16 deg C) since November 2016.


Figure 5 – RSS Lower Troposphere Temperature (TLT) Anomalies


Introduction: The GISS Land Ocean Temperature Index (LOTI) reconstruction is a product of the Goddard Institute for Space Studies. Starting with the June 2015 update, GISS LOTI uses the new NOAA Extended Reconstructed Sea Surface Temperature version 4 (ERSST.v4), the pause-buster reconstruction, which also infills grids without temperature samples. For land surfaces, GISS adjusts GHCN and other land surface temperature products via a number of methods and infills areas without temperature samples using 1200km smoothing. Refer to the GISS description here.   Unlike the UK Met Office and NCEI products, GISS masks sea surface temperature data at the poles, anywhere seasonal sea ice has existed, and they extend land surface temperature data out over the oceans in those locations, regardless of whether or not sea surface temperature observations for the polar oceans are available that month.  Refer to the discussions here and here. GISS uses the base years of 1951-1980 as the reference period for anomalies. The values for the GISS product are found here. (I archived the former version here at the WaybackMachine.)

Update: The December 2016 GISS global temperature anomaly is +0.81 deg C. According to the GISS LOTI data, global surface temperature anomalies made a noticeable downtick in December, a -0.12 deg C decrease.


Figure 1 – GISS Land-Ocean Temperature Index


NOTE: The NCEI only produces the product with the manufactured-warming adjustments presented in the paper Karl et al. (2015). As far as I know, the former version of the reconstruction is no longer available online. For more information on those curious NOAA adjustments, see the posts:

And recently:

Introduction: The NOAA Global (Land and Ocean) Surface Temperature Anomaly reconstruction is the product of the National Centers for Environmental Information (NCEI), which was formerly known as the National Climatic Data Center (NCDC). NCEI merges their new “pause buster” Extended Reconstructed Sea Surface Temperature version 4 (ERSST.v4) with the new Global Historical Climatology Network-Monthly (GHCN-M) version 3.3.0 for land surface air temperatures. The ERSST.v4 sea surface temperature reconstruction infills grids without temperature samples in a given month. NCEI also infills land surface grids using statistical methods, but they do not infill over the polar oceans when sea ice exists. When sea ice exists, NCEI leave a polar ocean grid blank.

The source of the NCEI values is through their Global Surface Temperature Anomalies webpage. Click on the link to Anomalies and Index Data.)

Update: The December 2016 NCEI global land plus sea surface temperature anomaly was +0.79 deg C.  See Figure 2. It made an uptick (an increase of about +0.04 deg C) since November 2016.


Figure 2 – NCEI Global (Land and Ocean) Surface Temperature Anomalies


Introduction: The UK Met Office HADCRUT4 reconstruction merges CRUTEM4 land-surface air temperature product and the HadSST3 sea-surface temperature (SST) reconstruction. CRUTEM4 is the product of the combined efforts of the Met Office Hadley Centre and the Climatic Research Unit at the University of East Anglia. And HadSST3 is a product of the Hadley Centre. Unlike the GISS and NCEI reconstructions, grids without temperature samples for a given month are not infilled in the HADCRUT4 product.  That is, if a 5-deg latitude by 5-deg longitude grid does not have a temperature anomaly value in a given month, it is left blank. Blank grids are indirectly assigned the average values for their respective hemispheres before the hemispheric values are merged.  The HADCRUT4 reconstruction is described in the Morice et al (2012) paper here. The CRUTEM4 product is described in Jones et al (2012) here. And the HadSST3 reconstruction is presented in the 2-part Kennedy et al (2012) paper here and here. The UKMO uses the base years of 1961-1990 for anomalies.  The monthly values of the HADCRUT4 product can be found here.

Update: The December 2016 HADCRUT4 global temperature anomaly is +0.59 deg C. See Figure 3.  It also made an uptick from November to December 2016, an increase of about +0.07 deg C.


Figure 3 – HADCRUT4


The GISS, HADCRUT4 and NCEI global surface temperature anomalies and the RSS and UAH lower troposphere temperature anomalies are compared in the next three time-series graphs. Figure 6 compares the five global temperature anomaly products starting in 1979. Again, due to the timing of this post, the HADCRUT4 and NCEI updates lag the UAH, RSS, and GISS products by a month. For those wanting a closer look at the more recent wiggles and trends, Figure 7 starts in 1998, which was the start year used by von Storch et al (2013) Can climate models explain the recent stagnation in global warming? They, of course, found that the CMIP3 (IPCC AR4) and CMIP5 (IPCC AR5) models could NOT explain the recent slowdown in warming, but that was before NOAA manufactured warming with their new ERSST.v4 reconstruction…and before the strong El Niño of 2015/16. Figure 8 starts in 2001, which was the year Kevin Trenberth chose for the start of the warming slowdown in his RMS article Has Global Warming Stalled?

Because the suppliers all use different base years for calculating anomalies, I’ve referenced them to a common 30-year period: 1981 to 2010. Referring to their discussion under FAQ 9 here, according to NOAA:

This period is used in order to comply with a recommended World Meteorological Organization (WMO) Policy, which suggests using the latest decade for the 30-year average.

The impacts of the unjustifiable, excessive adjustments to the ERSST.v4 reconstruction are visible in the two shorter-term comparisons, Figures 7 and 8. That is, the short-term warming rates of the new NCEI and GISS reconstructions are noticeably higher than the HADCRUT4 data. See the June 2015 update for the trends before the adjustments.


Figure 6 – Comparison Starting in 1979



Figure 7 – Comparison Starting in 1998



Figure 8 – Comparison Starting in 2001

Note also that the graphs list the trends of the CMIP5 multi-model mean (historic through 2005 and RCP8.5 forcings afterwards), which are the climate models used by the IPCC for their 5th Assessment Report.  The metric presented for the models is surface temperature, not lower troposphere.


Figure 9 presents the average of the GISS, HADCRUT and NCEI land plus sea surface temperature anomaly reconstructions and the average of the RSS and UAH lower troposphere temperature composites. Again because the HADCRUT4 and NCEI products lag one month in this update, the most current monthly average only includes the GISS product.


Figure 9 – Average of Global Land+Sea Surface Temperature Anomaly Products


As noted above, the models in this post are represented by the CMIP5 multi-model mean (historic through 2005 and RCP8.5 forcings afterwards), which are the climate models used by the IPCC for their 5th Assessment Report.

Considering the uptick in surface temperatures in 2014, 2015 and now 2016 (see the posts here and here), government agencies that supply global surface temperature products have been touting “record high” combined global land and ocean surface temperatures. Alarmists happily ignore the fact that it is easy to have record high global temperatures in the midst of a hiatus or slowdown in global warming, and they have been using the recent record highs to draw attention away from the difference between observed global surface temperatures and the IPCC climate model-based projections of them.

There are a number of ways to present how poorly climate models simulate global surface temperatures. Normally they are compared in a time-series graph.  See the example in Figure 10. In that example, the UKMO HadCRUT4 land+ocean surface temperature reconstruction is compared to the multi-model mean of the climate models stored in the CMIP5 archive, which was used by the IPCC for their 5th Assessment Report. The reconstruction and model outputs have been smoothed with 61-month running-mean filters to reduce the monthly variations.  The climate science community commonly uses a 5-year running-mean filter (basically the same as a 61-month filter) to minimize the impacts of El Niño and La Niña events, as shown on the GISS webpage here. Using a 5-year running mean filter has been commonplace in global temperature-related studies for decades. (See Figure 13 here from Hansen and Lebedeff 1987 Global Trends of Measured Surface Air Temperature.) Also, the anomalies for the reconstruction and model outputs have been referenced to the period of 1880 to 2013 so not to bias the results.  That is, by using the almost the full term of the data, no one with the slightest bit of common sense can claim I’ve cherry picked the base years for anomalies with this comparison.


Figure 10

It’s very hard to overlook the fact that, over the past decade, climate models are simulating way too much warming…even with the small recent El Niño-related uptick in the data.

Another way to show how poorly climate models perform is to subtract the observations-based reconstruction from the average of the model outputs (model mean). We first presented and discussed this method using global surface temperatures in absolute form. (See the post On the Elusive Absolute Global Mean Surface Temperature – A Model-Data Comparison.) The graph below shows a model-data difference using anomalies, where the data are represented by the UKMO HadCRUT4 land+ocean surface temperature product and the model simulations of global surface temperature are represented by the multi-model mean of the models stored in the CMIP5 archive. Like Figure 10, to assure that the base years used for anomalies did not bias the graph, the full term of the graph (1880 to 2013) was used as the reference period.

In this example, we’re illustrating the model-data differences smoothed with a 61-month running mean filter. (You’ll notice I’ve eliminated the monthly data from Figure 11. Example here. Alarmists can’t seem to grasp the purpose of the widely used 5-year (61-month) filtering, which as noted above is to minimize the variations due to El Niño and La Niña events and those associated with catastrophic volcanic eruptions.)


Figure 11

The difference now between models and data is almost worst-case, comparable to the difference at about 1910. 

There was also a major difference, but of the opposite sign, in the late 1880s. That difference decreases drastically from the 1880s and switches signs by the 1910s.  The reason:  the models do not properly simulate the observed cooling that takes place at that time.  Because the models failed to properly simulate the cooling from the 1880s to the 1910s, they also failed to properly simulate the warming that took place from the 1910s until the 1940s. (See Figure 12 for confirmation.) That explains the long-term decrease in the difference during that period and the switching of signs in the difference once again.  The difference cycles back and forth, nearing a zero difference in the 1980s and 90s, indicating the models are tracking observations better (relatively) during that period. And from the 1990s to present, because of the slowdown in warming, the difference has increased to greatest value ever…where the difference indicates the models are showing too much warming.

It’s very easy to see the recent record-high global surface temperatures have had a tiny impact on the difference between models and observations.

See the post On the Use of the Multi-Model Mean for a discussion of its use in model-data comparisons.


Yet another way to show how poorly climate models simulate surface temperatures is to compare 30-year running trends of global surface temperature data and the model-mean of the climate model simulations of it. See Figure 12. In this case, we’re using the global GISS Land-Ocean Temperature Index for the data. For the models, once again we’re using the model-mean of the climate models stored in the CMIP5 archive with historic forcings to 2005 and worst case RCP8.5 forcings since then.


Figure 12

There are numerous things to note in the trend comparison. First, there is a growing divergence between models and data starting in the early 2000s. The continued rise in the model trends indicates global surface warming is supposed to be accelerating, but the data indicate little to no acceleration since then. Second, the plateau in the data warming rates begins in the early 1990s, indicating that there has been very little acceleration of global warming for more than 2 decades. This suggests that there MAY BE a maximum rate at which surface temperatures can warm. Third, note that the observed 30-year trend ending in the mid-1940s is comparable to the recent 30-year trends. (That, of course, is a function of the new NOAA ERSST.v4 data used by GISS.)  Fourth, yet that high 30-year warming ending about 1945 occurred without being caused by the forcings that drive the climate models.  That is, the climate models indicate that global surface temperatures should have warmed at about a third that fast if global surface temperatures were dictated by the forcings used to drive the models. In other words, if the models can’t explain the observed 30-year warming ending around 1945, then the warming must have occurred naturally. And that, in turns, generates the question: how much of the current warming occurred naturally? Fifth, the agreement between model and data trends for the 30-year periods ending in the 1960s to about 2000 suggests the models were tuned to that period or at least part of it. Sixth, going back further in time, the models can’t explain the cooling seen during the 30-year periods before the 1920s, which is why they fail to properly simulate the warming in the early 20th Century.

One last note, the monumental difference in modeled and observed warming rates at about 1945 confirms my earlier statement that the models can’t simulate the warming that occurred during the early warming period of the 20th Century.


The most recent sea surface temperature update can be found here. The satellite-enhanced sea surface temperature composite (Reynolds OI.2) are presented in global, hemispheric and ocean-basin bases.


We discussed the recent record-high global sea surface temperatures for 2014 and 2015 and the reasons for them in General Discussions 2 and 3 of my most recent free ebook On Global Warming and the Illusion of Control (25MB). (And, of course, the record highs in 2016 are lagged responses to the 2015/16 El Niño.) The book was introduced in the post here (cross post at WattsUpWithThat is here).

About Bob Tisdale

Research interest: the long-term aftereffects of El Niño and La Nina events on global sea surface temperature and ocean heat content. Author of the ebook Who Turned on the Heat? and regular contributor at WattsUpWithThat.
This entry was posted in TLT and LOST Updates. Bookmark the permalink.

12 Responses to December 2016 Global Surface (Land+Ocean) and Lower Troposphere Temperature Anomaly Update – With a Look at the Year-End Annual Results

  1. ngard2016 says:

    Bob do you have a trend map link like this one from RSS that shows the world since 1998? I’m particularly interested to show Australian trend since say 1998.1. I can’t seem to find anything at Roy Spencer’s that covers that period for UAH V 6. Can you help with a trend for Australia since 1998.1 from UAH V 6? Thanks Bob.

  2. ngard2016 says:

    Sorry I have a comment in moderation. Why?

  3. Bob Tisdale says:

    ngard2016, I moderate ALL comments.

    You can create trend maps at the KNMI Climate Explorer.

    For some reason unknown to me, it’s not including 2016 UAH TLT v6.0 data. However, the trend is so small from 1998 to 2015, it shows nothing for Australia.

  4. ngard2016 says:

    Thanks Bob , but I think that site is beyond my capability. I know that the OZ data is available from Roy Spencer’s site every month but I haven’t the skills to generate a trend from it. Thanks again.

  5. ngard2016 says:

    Bob, Roy Spencer showed a UAH V 6 trend for Australia from Dec 1978 to Mar 2015 of 2.4 c a century. UAH V 5.6 actua;lly had a lower trend of about 1.6 c a century.

    What I’d like to know is the Aust trend since August 1995 to 2017. And if that trend is flat can you then consider that to be a cooling trend or not? Thanks.

  6. Bob Tisdale says:

    ngard2016, for the period of the December 1978 to Dec 2016, the UAH TLT v6.0 for Australia is 0.16 Deg C/decade:
    For the period you requested (Aug 1995 to Dec 2016), the trend is 0.00 deg C/decade, no warming, no cooling.

  7. ngard2016 says:

    Thanks for your help Bob.

  8. ngard2016 says:

    Bob, I posted this at Don Aitkin’s blog and wonder if you’d care to comment? Why are GISS and NOAA trends so low for Jones’s period 1910 to 1940?

    There have been some strange things going on with the word’s temp data-sets since Phil Jones’s 2010 interview with the BBC. The warming trend from 1910 to 1940 was 0.150 c/ decade, but that has now changed to 0.129 c/ decade first HAD 4 column and 0.139 HAD 4 Krig data.

    But here is the interesting part, GISS data shows just 0.090c/ decade and the Berkeley data is 0.143 c/ decade, plus NOAA is just 0.090 c /decade. But even NOAA land trend is now below what Jones quoted for L&O in 2010. In fact NOAA Land trend is just 0.134 c / decade today. This has be nonsense when even the much faster land trend can’t reach Jones’s L&O trend in 2010. Yet Berkeley land trend is 0.172 c/ decade for that period.

    But move forward and look at the trend from 1910 to 1945 ( 36 years) and we find more strange results. Had 4 trend is 0.140 c/dec, Had 4 krig is 0.151/ dec, NOAA is 0.134c /dec, GISS is 0.133 c/ dec and Berkeley jumps to 0.160 c/ dec.

    NOAA land trend is now 0.145 c/ dec , while Berkeley land trend is now 0.173 c/ dec. Clearly there have been adjustments made for HAD data since 2010 and the GISS and NOAA temp trends are ridiculously low for the period 1910 to 1940. But they do help to raise the later overall trend I suppose. Please note that Jones chose the trend period 1910 to 1940 and they are not a cherry pick.

    Here’s the York uni tool.

    And here’s Jones’s 2010 BBC interview.

  9. Bob Tisdale says:

    Thanks, ngard2016.

  10. Alec aka Daffy Duck says:

    Hi Bob…. OT of course
    Re: broad general cooling off at 400m subsurface

    Twice a week the Aussies post “Global Ocean 400m depth-averaged Temperature” graphic, with anomaly

    Do you know if there are public archives of those?
    Last Spring I started saving copies of the images now and then to my iPhone …looking back there appears to be a broad general cooling, but with out images from before the El Niño it hard to tell if it is meaningful ; Is 400m warmer or cooler than before the El Niño?

    Here’s 4 of the images I saved:


    sept 12

    nov 3

    feb 2

  11. Bob Tisdale says:

    Hi Alec. I don’t know if there are public archives of them.


  12. Hi Bob keen to connect over email, please. we are based in India need some help from you

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s