Reposted from Dr Roy Spencer’s Weblog
November seventh, 2019 by Roy W. Spencer, Ph. D.
A follower of our UAH international decrease tropospheric temperature (LT) dataset named “JJ” emailed me asking about what is likely to be thought-about a spurious function within the dataset.
The function is most simply seen in the event you plot the month-to-month international time collection of Land-minus-Ocean (hereafter “L-O”) temperature anomalies. The end result appears to indicate a step-up of about zero.16 deg. C in Might of 1998.
Fig. 1. Distinction between the UAH decrease tropospheric (LT) land and ocean temperature anomalies between January 1979 and August 2019, displaying an obvious step-up within the distinction occurring in Might 1998. The dashed strains present the typical values earlier than and after that date, whereas the curve is a fifth order polynomial match to the information.
The 12 months 1998 is vital for our dataset as a result of that’s when the primary (NOAA-15) Superior Microwave Sounding Unit (AMSU) got here on-line, which initiated the transition from the older Microwave Sounding Items (MSU, the final of which was on the NOAA-14 satellite tv for pc).
AMSU didn’t have precisely the identical channel frequency choice because the MSU, so the nominal layers of the environment sensed have been barely totally different. Most significantly, the AMSU channel 5 has a weighting perform that senses considerably extra of the floor and decrease troposphere than MSU channel 2. If one didn’t account for this reality, the AMSU’s larger floor sensitivity would produce greater temperatures over land and decrease temperatures over the ocean (after a global-average intercalibration between MSU and AMSU was carried out). [The reason why is that these channel frequencies are not sensitive to changes in sea surface temperature, because the microwave emissivity decreases as SST increases. The effect is small, but measurable.]
However since these are through-nadir scanners, every view angle relative to the native vertical measures a barely totally different layer anyway, which permits us to match the AMSU and MSU measurements. Once we developed Model 6 of the dataset, we discovered that the 50-60 GHz oxygen absorption idea used to seek out the view angle from AMSU5 that greatest matches MSU2, the ensuing temperature anomalies over land have been nonetheless too heat relative to the oceans. This meant that we needed to carry out an empirical (data-dependent) reasonably than theoretical matching of the AMSU and MSU view angles.
The best way we gauged the match between MSU and AMSU is how the temperature anomaly patterns transition throughout coastlines: we required that there needs to be little discernible change in that sample. Earlier than our optimized matching, the land anomalies have been noticeable hotter than the ocean anomalies as options crossed coastlines. However after optimization in our Model 6 dataset, right here’s the LT anomaly map for final month (October 2019), which reveals no proof for land-vs-ocean artifacts.
Fig. 2. October 2019 LT temperature anomalies relative to the 1981-2010 common annual cycle. Observe the anomalies have a easy transition between land and ocean, as could be anticipated for deep-layer tropospheric temperatures (however not essentially floor temperatures).
Nonetheless, changes like these are by no means excellent. So, the query stays: Is there a spurious change within the L-O temperature distinction occurring in 1998?
Proof that the L-O change in 1998 is actual
There are a couple of strains of proof that recommend the Might 1998 step-up in L-O temperatures is actual.
First, if the impact was as a result of introduction of AMSU in 1998, it will have occurred in August, not in Might (Three months earler). Additionally, the impact ought to have been gradual since for nearly four years after August 1998 the LT dataset is half MSU (NOAA-14) and half AMSU (NOAA-15), after which it becaume 100% AMSU.
However a extra vital piece of proof is the impact of El Nino and La Nina on L-O. Throughout El Nino, the ocean airmasses heat greater than the land airmasses (particularly within the tropics), in order that L-O tends to be extra unfavorable. Up till the 1997-98 tremendous El Nino a interval of larger El Nino exercise existed, after which a shift to extra La Nina exercise occurred. (That is in all probability additionally what triggered the prolonged international warming ‘hiatus’ after that El Nino occasion.)
I statistically regressed the L-O values in Fig. 1 towards Three-month working averages of the Multivariate ENSO Index (MEI), and eliminated that estimate of the ENSO affect from the information. The ensuing ENSO-adjusted time collection in proven in Fig. Three.
Fig. Three. As in Fig. 1, however with the typical affect of El Nino and La Nina (ENSO) subtracted out. Observe the proof for a “break” in 1998 is way weaker.
Observe the step-up in mid-1998 is way much less evident, and the fifth order polynomial match to the information is smoother with a extra gradual transition in L-O over the 41-year satellite tv for pc report.
However that’s not the one factor happening throughout this era that impacts the L-O values. There have been two main volcanic eruptions (El Chichon in early 1982, and particularly Pinatubo in mid-1991) that triggered extra cooling over land than ocean, inflicting quickly enhanced unfavorable values in L-O. Since these occasions usually are not as simply correlated with an index like MEI is with ENSO, I merely eliminated the information from 1982-83 and 1992-93 in Fig. Three and replotted the ends in Fig. four.
Fig. four. As in Fig. Three, however with the information influenced by main volcanoes El Chichon and Pinatubo eliminated.
Now we see that the fifth order polynomial match to the information comes fairly near the linear pattern (dashed grey line), which means that the step-up in 1998 in L-O was actual, and associated principally to a change in ENSO exercise earlier than versus after the 1997-98 tremendous El Nino, and with the most important volcanic eruptions in 1982 and 1991 contributing to the seemingly spurious function.
The remaining upward pattern in L-O is just the land airmasses warming sooner than the ocean, as could be anticipated for any warming pattern, whether or not pure or human-caused.
There stays what is likely to be a spurious function throughout 1980-81 in Fig. four, which might most probably be associated to our advert hoc correction for MSU channel Three drift throughout that point. This, nevertheless, ought to have little affect on the land and ocean traits as evidenced by the pattern line match (dashed grey line) in Fig. four.