CFSR Global Temperature January 2016 Preliminary

One nice thing about the Climate Forecast System Reanalysis (CFSR) data is that it is available at least in preliminary summary form every day from the University of Maine (UM) Climate Change Institute (CCI).  They have now posted preliminary data for all of January 2016 which has been used to produce the graphs presented here.  For those new to this data, it is derived from the US National Center for Environmental Prediction (NCEP) Global Forecast System (GFS) global weather model run inputs four times each day and is currently Version 2.  The CFSR data are also processed and reported by WxBell and their output is usually very close to that provided by UM CCI:

The UM CCI CFSR summary maps and graphs are here:

The UM CCI January daily CFSR show a continuation of the large high spikes in global temperature anomaly estimates that began in October 2015 and may be associated with the El Niño pattern.  The highest spikes in daily global temperature anomalies have coincided largely with the highest spikes in Arctic temperature anomalies which may be at least partially a result of warm El Niño air being advected into the Arctic for dissipation.  Figure 1 shows the daily CFSR global temperature anomaly estimates since 2014 (click on the graph for more detail).  The January estimates are preliminary and are based on the UM CCI reported values adjusted using a linear fit derived from the October through December preliminary versus final estimates.

UM CCI CFSR daily global temperature anomaly estimates 2014-2016 Jan

Figure 1. UM CCI CFSR daily global temperature anomaly estimates for 2014 through January 2016.

Figure 2 shows the monthly CFSR trend in global temperature anomaly estimates for the current century so far, since 2001.  The October through December 2015 monthly estimates were calculated from the UM CCI final daily values.  They have not yet reported the final October through December monthly values, which may be slightly different.  The January 2016 value is preliminary and is based on the adjusted UM CCI preliminary daily data described above.

UM CCI CFSR monthly global temperature anomaly estimates 2001-2016 Jan

Figure 2. UM CCI CFSR monthly global temperature anomaly estimates and trend for 2001 through January 2016.

The 21st Century CFSR global monthly temperature anomaly estimates continues to show a substantial downward trend, despite the recent upward spike since October 2015.  The current trend is -0.001 degrees Celsius (C) per month, which corresponds to -0.18C over the 15 year and one month period and to -1.2C per century if maintained.  Considering a likely uncertainty on the order of plus or minus 0.3C to 0.5C, the current trend for this century so far is well within the uncertainty range, and thus a real trend cannot yet be confidently established.  Regardless however, there does not appear to be a large upward trend in temperature so far this century.  A century trend of +2C would correspond to a 15 year trend of +0.3C, which would barely be detectable with low confidence, so we can say with confidence that we are not seeing such a high trend as predicted by most long-range climate models.  That in turn indicates a poor performance by these models.

The preliminary January CFSR global temperature anomaly estimate of 0.51C is slightly lower than the 0.55C for December and may indicate the beginning of a downward trend in coming months as the El Niño slowly fades away.  At face value, the implied January global temperature estimate of 13.0C ties 2005 for the second highest January in the CFSR record since 1979.  The highest January was 13.1C in 2007.  Within an uncertainty of plus or minus 0.3C to 0.5C, we cannot really determine which of these January global temperature estimates was the highest.

The CFSR data begin in 1979 and the trend for the entire period through January 2016 is shown in Figure 3.  The trend for this period of 37 years and one month is +0.0011C per month, which corresponds to +0.48C for the period and to +1.32C for 100 years if maintained.

UM CCI CFSR monthly global temperature anomaly estimates and trend

Figure 3. UM CCI CFSR monthly global temperature anomaly estimates and trend for 1979 through January 2016.

Thus considering an uncertainty of about plus or minus 0.3C to 0.5C, there is only marginal confidence in this small upward trend for the period.  Even if this trend continues for a full 100 years, it will still be well below the 2C per century and higher projections from long-range global climate models.

For for the latest CFSR daily and monthly updates to key figures, see the Daily Updates and Monthly Trends pages accessible from the menu bar at the top of this page.

I recently found annual average temperatures from the weather station at the top of Mount Washington in New Hampshire in the northeastern US at an elevation of 1,912 meters (6,274 feet).  Measurements from this site may be fairly representative of temperature changes over time in the lower troposphere for higher latitudes in the Northern Hemisphere.  Figure 4 shows the Mount Washington annual temperature trend for 1949 through 2015.

Mt Washington NH annual temperature trend 1949-2015.

Figure 4. Mount Washington, New Hampshire, annual temperature trend for 1949 through 2015.

The temperature measurements at Mount Washington show a slight downward trend of -0.0115C per year, which corresponds to -0.77C over the 67 year period and to -1.15C over a 100 year period if maintained.  There is certainly no indication of warming at this location.

Update 2016 February 11

The Mount Washington data graphed above was obtained from the Weather Warehouse at the following link:

I thought it was a bit odd how much the annual mean temperatures varied from year to year and I was a bit surprised how cold the annual temperatures were.  So I decided to try and corroborate the data using the daily data available on the Weather Underground (WU).  I had to retrieve a year at a time of daily data, but managed to get daily temperature average, minimum, and maximum data for 1973 to current and then calculated annual average temperatures.  The results do not match the Weather Warehouse data.  The average temperature over the entire period from 1973 through 2015 for the Weather Underground data was -2.5C as compared to -14.8C for the Weather Warehouse data.  The WU lowest annual average was -3.8C for 1980 and highest was -0.9C for 2012, showing much less range than the Weather Warehouse data which ranged from -21.4C in 2004 to -8.9C in 2006.  The long-term average temperature from the WU data are more in line with the -2.6C mean temperature for 1981-2010 reported in Wikipedia.  Therefore, I am now very doubtful that what was obtained from Weather Warehouse is correct.

I went to the US National Center for Environmental Information (NCEI) and retrieved daily data by month to spot check the WU data.  It appears that the WU data were derived from hourly observations and do not fully reflect the maximum and minimum temperatures on most days.  Therefore, I plan to download the NCEI data to compile annual averages since 1948 and will make a post about the results.


10 responses to “CFSR Global Temperature January 2016 Preliminary

  1. Hi. Do you have a link to where the data presented above in figures 2 and 3 can be downloaded from? It is bizzare that your data, while showing the current ElNino, shows, in contrast to all the other temperature records, very little impact due to past major elNino events. In particular 1998 looks stunted and 2010 seems to be just part of the “noise”.

    • The general link to UM CCI was provided at the beginning of this post. It includes lots of graphical outputs, including maps and time series plots for CFSR and other data sources. When you graph the monthly CFSR time series, there is an option below the graph to download the data, including a CSV option that provides the text data. I posted this link to the CFSR monthly text data at the end of my first post on the CFSR series: Satellite Era Global Temperature Anomaly Comparisons posted October 28, 2015. Here is the link again so you don’t have to look it up, although you might want to read through some of the comments on earlier CFSR related posts:

      UM CCI reports the monthly data as global temperature estimates. You have to calculate your own reference period averages and associated anomalies, which is what I do. They also report daily global temperature anomalies on their Daily Reanalysis Maps link. There you have to view one day at a time and copy the anomalies for each day, which is tedious, but is what I have done for recent daily data.

  2. Why do you prefer CFSR to ERA interim?
    ECMWF is apparently the best reanalysis service in the world for medium range weather forecast.

    • Javier, I’m not surprised that the ECMWF outperforms the GFS overall, although I doubt that it is consistently better in every case. I look at the CFSRv2 since UM CCI provides near real-time daily output. I have not run across similar near real-time ERAI output. I have looked at the ERAI data provided by UM CCI but it is not current and only runs through 2014. The ERAI seems to match HadCRUT4 fairly well through 2014, unlike the CFSRv2. The CFSRv2 seems to match the UAH TLT better.

      I am not sure why there are differences and I have not researched the details of the processing that is done to the huge amounts of weather data input into the reanalyses. To me, the differences help to confirm the large uncertainties in trying to estimate global temperature anomalies. I believe most people tend to greatly underestimate those uncertainties.

  3. As I posted on this site a couple of months ago, CFS changed to CFS2 in early 2011. There is a cooling bias in the combined CFS/CFS2 record in the 2010/11 period. That is why CFS is not quite at record levels while all other re-analysis products are far above pre-2015/6 records. Below is a link to the most recent ECMWF re-analysis report for January 2016. The differences between CFS/CFS2 and the ECWMF prroduct in the 2010/11 period are clear.

  4. Hi Chubbs. Your point is well taken. However, biases are all relative. Perhaps the ERAI has high bias? The CFSR seems to match the satellite patterns well and does not show a bias in comparison. As I have said before, these differences are simply further indication of the large uncertainties in these estimates that is probably at least 0.3C to 0.5C and all of these estimates are within that uncertainty envelope.

  5. Fortunately CFS2 is now stable, so will track the other re-analysis products going forward, as it has since early 2011. Since then, CFS2 has provided a good indicator of short-term temperature trends. The combined CFS/CFS2 series will always lag the warming in the other re-analysis products due to the 2010/11 period but the differences will diminish with time..

    • Chubbs, the UM CCI says the GFS has a bias compared to CFSV2 but claims that CFSV2 is “apples to apples” compatible with CFSR. Below are quotes from their web page:

      “GFS has a model bias compared to CFSR/CFSV2 in which very warm and very cold temperatures tend to be greater, leading to more pronounced temperature anomalies. In an effort to diminish the effect of this bias, a simple correction factor (CF) is applied to each regional temperature anomaly.”

      “Note that bias-corrected GFS-CFSR temperature anomalies will differ slightly from those calculated from CFSV2-CFSR. CFSV2-CFSR is apples-to-apples; therefore, difference maps within the Daily Reanalysis Maps image archive should be considered reliable, and take precedent over the GFS-CFSR anomaly maps shown here as part of the current 7-day forecast.”

      Here they show comparisons for 2015 June 30:

  6. Very useful and interesting, thank 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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s