by Judith Curry
Ted Cruz’s favorite temperature data set just got a lot hotter.
Sensitivity of satellite-derived tropospheric temperature trends to the diurnal cycle adjustment
Carl Mears and Frank Wentz
Abstract. Temperature sounding microwave radiometers flown on polar-orbiting weather satellites provide a long-term, global-scale record of upper-atmosphere temperatures, beginning in late 1978 and continuing to the present. The focus of this paper is the middle tropospheric measurements made by the Microwave Sounding Unit (MSU) channel 2, and the Advanced Microwave Sounding Unit (AMSU) channel 5. Previous versions of the RSS dataset have used a diurnal climatology derived from general circulation model output to remove the effects of drifting local measurement time. In this paper, we present evidence that this previous method is not sufficiently accurate, and present several alternative methods to optimize these adjustments using information from the satellite measurements themselves. These are used to construct a number of candidate climate data records using measurements from 15 MSU and AMSU satellites. The new methods result in improved agreement between measurements made by different satellites at the same time. We choose a method based on an optimized second harmonic adjustment to produce a new version of the RSS dataset, Version 4.0. The new dataset shows substantially increased global-scale warming relative to the previous version of the dataset, particularly after 1998. The new dataset shows more warming than most other middle tropospheric data records constructed from the same set of satellites. We also show that the new dataset is consistent with long-term changes in total column water vapor over the tropical oceans, lending support to its long-term accuracy.
Published in Journal of Climate [link to abstract]
Here is the main result, showing the difference between RSS v3.3 and v4
The political significance of this paper is summed up by the title of this Guardian article: Ted Cruz’s favorite temperature data just got a lot hotter.
Lets dig in and clarify the scientific importance (or not) of this dataset, and its implication for the pause.
Roy Spencer has a post Comments on New RSS Pause-Busting Global Temperature Data Set, with numerous figures. Excerpts:
While the title of their article implies that their new diurnal drift adjustment to the satellite data has caused the large increase in the global warming trend, it is actually their inclusion of what the evidence will suggest is a spurious warming (calibration drift) in the NOAA-14 MSU instrument that leads to most (maybe 2/3) of the change.
Here I have included their Fig. 7 as an inset to show that they know there is a substantial trend difference between the old NOAA-14 MSU and the newer NOAA-15 AMSU measurements. That trend difference amounts to +0.20 C/decade…a large discrepancy.
Importantly, Mears and Wentz choose to leave this calibration drift in without adjustment for it. In effect they are saying, ‘we don’t know which of the two satellites is at fault, so we will leave both satellites in without adjustment’.
Here are the reasons why we believe we can blame the calibration drift on the NOAA-14 MSU instrument, and why we remove that spurious warming from the NOAA-14 data in our v6 LT and MT products:
- the old MSU instruments’ calibration did not have near the sophistication of the newer AMSU instruments (NASA AMSU design engineer Jim Shiue once told me the AMSUs had “Cadillac”-quality calibration)
- the NOAA-14 satellite orbit was drifting far beyond any of the other dozen satellites in the record, leading to warming of the instrument itself (which is why we cut the record short after 6 yrs, RSS uses all 10 years), while the NOAA-15 satellite had very little orbital drift during its overlap with NOAA-14.
We find it curious (to say the least) that RSS would treat these two satellites as equally accurate.
About a third of the trend difference appears to be due to a change in the RSS method for diurnal drift adjustment, as indicated by the dashed ovals in the second plot, above. (Diurnal drift is the result of the satellite overpass time changing over the years, so that measurements are made at a different times of day; over land in particular this causes a drift in measured temperature due to the day-night cycle, not climate). Their new adjustment appears to provide a stronger correction for the diurnal cooling of the NOAA-11 satellite (first oval) and the NOAA-18 satellite (second oval). RSS uses the diurnal cycle from a climate model (CCM3), with empirical adjustments. We (UAH) use a pure empirical adjustment based of the the observed drift between NOAA-18 and NOAA-19 (for the “1:30” satellites) and NOAA-15 and Aqua (for the “7:30” satellites).
After Carl made the new RSS data available to us, John Christy computed the level of agreement (explained variance) that three satellite datasets (RSSv4, UAHv6, NOAAv3.0) have with the corresponding values from various radiosonde and reanalysis datasets. The results indicate that, with the exception of one reanalysis dataset (MERRA-2, which has by far the warmest trend), the UAH anomalies have better agreement with other data sources than does the RSS (or NOAA) dataset
The evidence suggests that the new RSS v4 MT dataset has spurious warming due to a lack of correction for calibration drift in the NOAA-14 MSU instrument. Somewhat smaller increases in their warming trend are due to their use of a climate model for diurnal drift adjustment, compared to our use of an empirical approach that relies upon observed diurnal drift from the satellite data themselves. While the difference in diurnal drift correction methodolgy is a more legitimate point of contention, in the final analysis independent validation with radiosonde data and most reanalysis datasets suggest better agreement with the UAH product than the RSS product.
Chip Knappenberger has pointed out that, while the warming in RSS v4 versus UAH v6 might be as described above, when RSS v4 is compared to RSS v3.3, the increase in warming might be mostly due to their new diurnal cycle adjustment. In other words, the NOAA-14 calibration issue was also in their v3.3, but maybe it was obscured more by diurnal drift adjustment issues.
WUWT has a post with a brief interview of Spencer:
The paper is for MT, not LT…but I think we can assume that changes in one will be reflected in the other when Mears completes their analysis.
From what little we have looked at so far, it appears that they did not correct for spurious warming in NOAA-14 MSU relative to NOAA-15 AMSU…see their Fig. 7c. They just leave it in.
Since this spurious warming is near the middle of the whole time period, this shifts the second half of the satellite record warmer when NOAA-14 MSU (the last in the MSU series) is handed off to NOAA-15 AMSU (the first in the AMSU series).
Why do we think NOAA-14 MSU is at fault?
1) AMSU is supposed to have a “Cadillac” calibration design (that’s the term a NASA engineer, Jim Shiue, used when describing to me the AMSU design, which he was involved in).
2) NOAA-14 MSU requires a large correction for the calibrated TB increasing with instrument temperature as the satellite drifts into a different orbit. The NOAA-15 AMSU requires no such correction…and it wasn’t drifting during the period in question anyway.
So, it looks like they decided to force good data to match bad data. Sound familiar?
UPDATE1: Given this sort of work has only two groups doing it, it is a very narrow field of scientific specialty, I asked Dr. Spencer this question:
I assume neither you or Christy were asked to review this paper?
There aren’t many satellite temperature data experts in the world.
John reviewed their original paper submission to JGR, in detail, asking for additional evidence — but not advocating rejection of the paper. The JGR editor ended up rejecting it anyway.
Mears & Wentz then revised the paper, submitted it to J. Climate instead, and likely asked that we be excluded as reviewers.
By a statistically significant amount, February 2016 was the warmest month in the satellite temperature record, according to Dr. John Christy, director of the Earth System Science Center at The University of Alabama in Huntsville. Interestingly, however, that record might have as much to do with an extraordinarily warm month in the Arctic as it does with warming caused by the El Niño Pacific Ocean warming event. Temperatures in the tropics and the Southern Hemisphere were not at record levels in February.
While the Arctic temperature anomaly is large, big temperature swings in the Arctic region aren’t unusual, especially during the winter months. Those swings are also normally somewhat transient, so the extra heat represented in February could dissipate over the next few weeks. If that happens, it doesn’t appear the heat from the El Niño by itself will be enough to continue pushing temperatures to new records later in the year, in which case this February anomaly might stand out as a singular spike in the dataset rather than part of an ongoing trend.
Regarding what we can expect for the rest of 2016, Bob Tisdale provides some speculations:
As shown in Figure 2, the responses of the lower troposphere temperature data to the 2015/2016 El Niño are similar to the responses to the 1997/98 event.
In response to an El Niño, only a portion of the temporary upticks in global surface temperatures are a direct result of the warming of the eastern tropical Pacific (caused by the warm subsurface waters of the western tropical Pacific being shifted to the surface of the eastern tropical Pacific). There are also long-lasting responses to strong El Niños (Trenberth “big jumps”), when the leftover warm waters from the El Niño are redistributed around the surfaces of the global oceans.
On the other hand, the rises in lower troposphere temperatures in response to an El Niño are caused in two ways: first, by the increases in surface temperatures in the tropical Pacific and around the globe. Second, an El Niño releases a monumental amount of heat from the tropical Pacific to the atmosphere, primarily through evaporation. That additional warm and moist air rises into the colder atmosphere, cooling as it rises higher, and when the moisture condenses and form clouds, that heat from the tropical Pacific is released to the atmosphere…thus the additional warming of the lower troposphere.
As noted in the opening, if global temperatures respond to the 2015/16 El Niño as they had for the 1997/98 El Niño, we might expect to see a couple of additional upticks this year in the lower troposphere and surface temperature anomalies.
Will the evolutions of global temperatures this year mimic 1998? One wild card is The Blob, which was the primary cause of the “record” surface temperatures in 2014. The Blob continues to dissipate rapidly, but it might reemerge in 2016. IF (big if) The Blob continues to dissipate and doesn’t reappear later in the year, and IF (big if) it doesn’t have any long-term effects on the North Pacific, then its disappearance could suppress some of the additional El Niño-related warming in 2016. We just have to grab some popcorn and watch…better stock up, this is gonna take a while.
The climate models project strong warming in the tropical mid troposphere, which have not been borne out by the observations. The new RSS data set reduces the discrepancies with the climate model simulations.
Roy Spencer’s comments substantially reduce the credibility of the new data set. Their dismissal of the calibration problems with the NOAA-14 MSU is just astonishing. Presumably Christy’s review of the original submission to JGR included this critique, so they are unlikely to be unaware of this issue. The AMS journals have one the best review processes out there; I am not sure why Christy/Spencer weren’t asked to review. I have in the past successfully argued at AMS not to have as reviewers individuals that have made negative public statements about me (not sure if this is the case with Mears/Wentz vs Spencer/Christy).
There is a legitimate debate on how to correct for the diurnal cycle, but based on my assessment, the UAH empirically based approach seems better.
With regards to the ‘pause.’ The ‘pause’ in warming has generally been assessed using the lower tropospheric temperatures, which aren’t yet available from the new dataset. So it is not yet clear what impact the new data set will have on our interpretation of the pause.
With regards to the Feb 2016 spike, I think Bob Tisdale gets it mostly right. While a spike from the El Nino is expected, the Feb 2016 seems anomalous and largely associated with a warm spike in the Arctic (of ‘weather’ origin). I would expect a few more months of anomalously warm temperatures before the El Nino fades. I’m not sure what to make of the ‘re-emergent blob’ scenario.
We won’t know what the 2016 El Nino spike looks like until the end of the year. Then we can compare the 1997/1998 temperatures with 2015/2016 temperatures in a (cherry-less) apples to apples comparison, to assess the underlying trend in temperatures from 1998-2016. The trend will undoubtedly be positive, but most likely it will remain substantially less than the trend predicted by the climate models.
And what of the years following 2016? Will we see cooling and then a continuation of flat temperatures? Or continued warming? I suspect that there will be some cooling and continued flatness. I’ve stated before that it will be another 5 years before we have the appropriate prospective on the current temperature fluctuations and whether or not the early 21st century pause is over.
We just have to grab some popcorn and watch…better stock up, this is gonna take a while.