Satellite temperature data vs ground based data

[Kevin McHugh]

I've read in the popular media that there is a difference between the two. Is this true, and if so, why?

 

[berkeman]

Can you post some links?

 

[Bandersnatch]

Have a read through this:
https://skepticalscience.com/Response-Data-or-Dogma-hearing.html
Here's a paper in Nature covering the same thing:
https://www.nature.com/ngeo/journal/v10/n7/full/ngeo2973.html

The blog post is by the authors of the paper, and includes extensive sources.

 

[Ron G]

From the point of view of a non-scientist who was fortune to work a short time with the folks at NCDC, Asheville, NC, I was able to gradually come to the unavioble truth of climate change and its looming prospects. My work began as a liaison for the NC Arboretum and after retirement I continued as a consulted for a short time to NCDC for managing their expanding mission responsibilities. I came to highly respect the skilled and devoted members of the NOAA team, the supporting universities, and the commercial companies as they worked on the data input end of the national climate response. My simple understanding is:

1. Data sources for long-range predictions must be accurate and set to a common baseline. For prehistorical times, NCDC gets data estimates from inferences like tree rings and soil composition along with carbon dating and other forms of dating sampling. For historical periods, we have among other source, naval and commercial ship logs well back into the age of sailing that record temperature at specific times during each day even when in port. We have some local weather station data going back into the dark ages. We have weather satellite data going back to NASA's original research efforts and forward to today. It's easy to see why older data might not be using the same yard stick for comparison purposes; but, even weather satellite data must be corrected to coincide with a standard for comparison. Weather data are collected with a very short term objective; so, accuracy of historical data, meaning from about two weeks ago, just are not important for analysis purposes. Furthermore, weather data need only reference itself for historical comparisons.

2. Satellite Climate Data must be Ground-Proofed to be Useful. A climate satellite records temperatures as well as other related climate data, and these raw data are kept in tact; but, in order to make it useful, it does need to match as closely as possible the ground-collected data. So, a major portion of NCDC work is to "Ground-Proof" these raw data by looking at the factors that cause differences, such as atmospheric interference, altitude variations, or instrument calibration loss over time. They then collect ground recordings from local weather stations and a network of specialized climate data collection sites scattered all over the globe. With these inputs, they can adjust the non-standard ground data and satellite data to a common baseline for comparison. The raw data and the "ground-proofed" versions are then stored permanently in a network of data centers all over the US to ensure integrity and safety of the record.

3. Turning the Data into Information Requires Continual Adjustment. The data in the various storage centers are available for analysis by persons and institutions capable of understanding it. They perform peer-reviewed studies on all manner of things like algorithms for adjusting ship's logs, using tree rings to estimate historical temperatures, looking back at old weather data and adjusting it to be compatible to the climate baseline standards. The currently "certified acceptable" data base is then available for public use by academia and participating commercial firms to make long-range predictions and consumer relevant interface applications . All these activities are closely scrutinized and duplicated by other climate centers around the nation and other countries to ensure the best possible data. As we learn more, the baseline standards can be changed and the certified data for public use is updated.

4. Making the Information Available is the Last Important Task. When the US realized the value of weather predictions for the public safety, we did for weather prediction what we are now doing for climate. We created the National Weather Service and a Collaborative Center with academia and commercial vendors with the sole mission of collecting, analyzing and presenting weather predictions to the public. This center is located in Norman Oklahoma, and their end products include what you see or hear during the evening news. There are now several climate collaboration centers in the US, one of which is collocated with NCDC in Asheville with the single purpose of producing information usable for the government and the general public.

5. IPCC is Real Scientist doing Real Science. All industrialized nations, except three, accept the fundamental conclusions of the scientific community. But every person should make up their own mind. So, which is easier to believe: thousands of scientists from all over the globe in essential unanimous agreement that the climate is changing and that we face dire consequences if we do not act concertedly soon; or, the message of deniers that see a global conspiracy to make up stories to thwart the ambitions of trans-national corporations?

I have worked closely with the honorable and skilled scientists and engineers of NOAA. I have see in detail their process for adjusting data for comparability. I have seen how terribly seriously they take the threats implied by climate change. I have seen politicians like Ted Cruz who will say absolutely anything to get reelected or to serve the deep pockets of special interests. The PhD's that won Nobel prizes for their climate work, many of them here in Asheville, didn't get a pay raise, and they won't make personal millions from their efforts. They are simply trying to do right by our only spaceship through time -- earth.

 

[FrankMlinar]

I am in a discussion with someone concerning AGW, and he likes to use "the last two decades!" to prove there is no AGW or AGW is in the "noise". I generated a 20 year moving trend plot for HADCRUT (surface) and three troposphere (satellite) data sets. I think this clearly shows the strong similarities between the data sets. (The center of the moving trend is the plot point)

 

[berkeman]

I am in a discussion with someone concerning AGW, and he likes to use "the last two decades!" to prove there is no AGW or AGW is in the "noise". I generated a 20 year moving trend plot for HADCRUT (surface) and three troposphere (satellite) data sets. I think this clearly shows the strong similarities between the data sets. (The center of the moving trend is the plot point)

Could you supply a link to the source of the data you plotted? The rules for discussion of Global Warning in the Earth forum are fairly restrictive -- we can discuss the science behind the measurements, but not the politics involved. (See the stickie thread at the top of the Earth forum for details) Thank you.

 

[FrankMlinar]

Certainly!
The raw temperature data is from RSS for satellites:
http://images.remss.com/msu/msu_time_series.html

And HADCRUT for surface:
https://crudata.uea.ac.uk/cru/data/temperature/HadCRUT4-gl.dat

I used a program that I wrote to calculate and plot the 20 year trends. I like this representation of the data because it very clearly shows the similarities and differences whereas just plotting the temperature data does not.

Ignore the politics!

 

[olivermsun]

Is the OP asking whether there is a difference in the climate trends between the methods or whether the two methods retrieve the same temperature from the same height or what?

Some of the posts above are also bringing in the differences between model vs. satellite trends, which is different from questions about measurement techniques.

 

[Kevin McHugh]

http://tqe.quaker.org/2007/aux/GWSatTemps.html

These graphs were published in an article I read on Yahoo. These are the differences I'm referring to. Not interested in politics, just what these differences mean.

 

[FrankMlinar]

First, I would not be at all concerned with the absolute values because that is due to how the baseline is calculated. For example, Figure 2 appears to use the peak of the elNino effect as the baseline. Also, separating the data sets can make the chart more readable. Having said that, the trend (slope of the line) is what is important.

The second point is, because of the relatively large variations in the data, it is good to have a lot of data points when determining a trend. Climate scientists like to have 30 or more years of data before declaring a trend. Figure 1 has 18 years of data, and the trend has a lot of uncertainty associated with it. A larger database, 30 years in Figure 2) reduces the uncertainty in the trend and increases the confidence in the trend value. As a result, the trends are more similar between satellite and ground meaning both are tracking each other.

I did the fancy moving trend analysis using 20 years (political reasons) because I thought it would more clearly show the tracking and similarities.

Finally, MSU/RSS very recently updated the TLT corrections and further reduced the differences.
I hope this answers your question.

 

[olivermsun]

...because of the relatively large variations in the data, it is good to have a lot of data points when determining a trend. Climate scientists like to have 30 or more years of data before declaring a trend. Figure 1 has 18 years of data, and the trend has a lot of uncertainty associated with it. A larger database, 30 years in Figure 2) reduces the uncertainty in the trend and increases the confidence in the trend value. As a result, the trends are more similar between satellite and ground meaning both are tracking each other.

"Smoothing over weather noise" when comparing observed and model-output trends is well-justified because they look at different "weather noise," but aren't two given observation methods supposed to be measuring the same (real) "weather noise"?

 

[FrankMlinar]

"Smoothing over weather noise"... Actually this is not "noise", but high frequency information. That is, real, measured data with variations that happen over short time periods. Smoothing is done to look at the long term variations and to determine their underlying causes. Plus, I don't believe I ever said "weather noise".

Now, "two given observation methods", by the very definition, will result in two different measurements simply because of uncertainties in the measurement methods; the instruments themselves, the locations of the measurements, the time of measurement, etc. For this discussion, we are looking at satellite based measurements and surface based measurements, that is, different locations. For satellite measurements, we are looking at different heights in the atmosphere (TLT, TTT, TMT), and differences will appear. The surface based measurement is a different location from the satellites, using different instruments at different times, and will show differences in the measurements.

 

[olivermsun]

"Smoothing over weather noise"... Actually this is not "noise", but high frequency information. That is, real, measured data with variations that happen over short time periods. Smoothing is done to look at the long term variations and to determine their underlying causes. Plus, I don't believe I ever said "weather noise".

The high-frequency part that gets filtered out when people look for climate trends is often called "weather noise" because it's assumed to be due to "weather" along with other higher frequency stuff superposed on the underlying long-term trend.

Now, "two given observation methods", by the very definition, will result in two different measurements simply because of uncertainties in the measurement methods; the instruments themselves, the locations of the measurements, the time of measurement, etc. or this discussion, we are looking at satellite based measurements and surface based measurements, that is, different locations. For satellite measurements, we are looking at different heights in the atmosphere (TLT, TTT, TMT), and differences will appear. The surface based measurement is a different location from the satellites, using different instruments at different times, and will show differences in the measurements.

My point is that time-filtering to remove differences between observations obtained by different instruments, sampling at different places and times, etc., isn't as trivial to justify as time-filtering to remove "weather noise."

 

[technophile50]

Kevin - The satellites measure temperatures of a range of atmospheric heights - usually called lower and middle troposphere, and lower stratosphere. The TLT (Temperature of the Lower Trop.) gets most of its signal from the lower troposphere(microwave emissions from oxygen), with less amounts from near the surface, and at higher altitudes, even into the stratosphere; the amounts of signal vary smoothly with altitude. see http://www.drroyspencer.com/wp-content/uploads/MSU2-vs-LT23-vs-LT.gif The surface air temperature is measured near the ground(close enough that the lapse rate doesn't make any difference; But of course, the temperature at Denver Colorado, USA is from a different altitude than Perth, Australia. On the plot, the trends are given for the various mean satellite measurements, and a curve for the trend of the RAOB (Radiosonde Atmospheric OBsevation, e.g. weather balloons). The negative trend in the stratosphere and positive trend in the troposphere and at the surface is a characteristic of greenhouse gas warming. GCMs project a larger trend at ~8-10 km than is observed (the positive bulge shown in the RAOB, and the roughly corresponding sat. trends is smaller than expected). The "satellite" trends in the graphs you linked to probably refer to RSS satellite derived temperatures, but there are different versions of satellite processing used to correct for things such as orbital decay, time of observation, and contribution from the different altitudes of the atmosphere, and they only go through 2008. data and trends through 2016 can be seen here
http://www.woodfortrees.org/plot/gi...t:-0.10/trend/plot/uah5/trend/plot/uah6/trend

Some people have noted that UAH v6 from noted skeptics Spencer and Christie has an "interesting and unusual" lower trend than their UAH v5.6, but since I'm just a college dropout layman, I couldn't possibly comment on this.
One comment I can make is that the smaller observed bump in trend between ~8-10 km than projected by climate models means that the lapse rate feedback is actually smaller than modeled. Since this climate feedback is negative, the models would show a higher climate sensitivity if they more accurately matched the observations

 

[jim mcnamara]

@technophile50 - your links are not from any standard reference. While people are entitled to opinions and PF does not try to prevent discussion, you will notice that Roy Spencer does not currently publish any place where his point of view could be questioned by someone else in his field - this is called a referee. Just like in football.

As expected this thread has demoted itself to political debate. Moved to General Discussion.

 

[technophile50]

"you will notice that Roy Spencer does not currently publish any place where his point of view could be questioned by someone else in his field "
Actually, I have noticed that Spencer (and his oft-time collaborator John Christie) does currently publish, and has a long track record of publishing peer reviewed science. Dr. Spencer's religious and political beliefs, in combination with his understanding of climatology and his recognized expertise in satellite atmospheric temperature retrieval makes him, in my opinion, a skeptic but not a political hack the likes of Monckton, Morano, Watts, or even McIntyre and McKitrick.
If one is suspicious of the graph I linked to on Dr. Spencer's website, you can see the same satellite sensitivity data from Dr Richard Mears of Remote Sensing Systems at http://www.remss.com/measurements/upper-air-temperature#Uncertainty figure 1 - which shows the same continuously varying, overlapping weighting functions as the graph from Spencer. It has the disadvantage of being cluttered with data from less relevant channels, and lacking the RAOB trend graph for easy visual comparison. Woodfortrees just displays data from other original sources, to which he has links on the homepage. If you think he's politically motivated in the way his interface shows the data, it is trivial to get the original data and process them yourself. I will note for anyone who hasn't looked at the woodfortrees link that UAH v6 warming trend is closer to RSS (the other satellite data set) than the various surface data sets, whereas UAHv5.6 trends were closer to the surface data sets. This may tell us something about comparing apples, oranges, and pears.

http://link.springer.com/article/10.1007/s13143-017-0010-y
UAH Version 6 global satellite temperature products: Methodology and results
RW Spencer, JR Christy, WD Braswell

http://journals.ametsoc.org/doi/abs/10.1175/BAMS-86-9-1303
http://search.proquest.com/openview...bd49a14d091f/1?pq-origsite=gscholar&cbl=40569
http://onlinelibrary.wiley.com/doi/10.1029/2005JD006881/full
http://journals.ametsoc.org/doi/abs/10.1175/1520-0442(2004)017<2225:UISOLC>2.0.CO;2
http://science.sciencemag.org/content/301/5636/1046.short
http://journals.ametsoc.org/doi/abs/10.1175/1520-0442(1992)005<0847:PARVOS>2.0.CO;2
http://journals.ametsoc.org/doi/abs/10.1175/1520-0426(1989)006<0254:PROLAO>2.0.CO;2
http://journals.ametsoc.org/doi/abs/10.1175/1520-0477(1997)078<1097:HDITTF>2.0.CO;2