Portal de Periódicos da CAPES

Comment on “Trends in the temperature and water vapor content of the tropical lower stratosphere: Sea surface connection” by Karen H. Rosenlof and George C. Reid

2009; American Geophysical Union; Volume: 114; Issue: D12 Linguagem: Inglês

10.1029/2008jd010542

2156-2202

John R. Lanzante,

Atmospheric and Environmental Gas Dynamics

[1] A recent article by Rosenlof and Reid [2008] (hereinafter referred to as RR08) examines the relationship in the tropics between long-term cooling in the stratosphere and warming of the sea surface, with special emphasis in the western Pacific. In the stratosphere, radiosonde observations seem to show that most of the cooling took place since the mid-1990s, apparently coincident to a considerable extent with an increase in the rate of warming of sea surface temperatures (SSTs). A link was proposed by RR08 whereby the increase in SSTs leads to enhanced convection, which is responsible, dynamically, for cooling in the stratosphere. [2] The stratospheric analyses of RR08 rely considerably on tropical radiosonde observations, particularly those from several island stations in the western tropical Pacific. There is extensive evidence that the long-term variation in worldwide radiosonde temperatures is affected critically by inhomogeneities introduced through changes in instruments and measurement practices [e.g., Gaffen, 1994; Gaffen et al., 2000; Lanzante et al., 2003a, 2003b; Climate Change Science Program (CCSP), 2006], a fact which is acknowledged by RR08. However, they minimize the importance of any such inhomogeneities in the six stations located in the “warm pool” region which they employ [RR08, paragraph 10]: “However, for our analysis, and in particular for changes noted post-1990, these discontinuities are not sufficient to preclude the use of these stations.” The purpose of this comment is to demonstrate that in fact such discontinuities are present, post-1990, and critically affect the evaluation of stratospheric temperatures made by RR08. [3] In their Figures 2, 3, and 5, RR08 display temperature time series for several levels in the lower stratosphere for the four stations located in the heart of the western Pacific warm pool. They note that while most of the earlier part of the records show little long-term trend, a pronounced cooling is seen for about the last 10 years of the record (∼1995–2005). Furthermore, they note an abrupt rise in temperature in early 1999 that lasts for about 2 years. As further evidence to bolster their findings, they note [RR08, paragraph 12] the similarity of the time series from the different stations: “… the overall similarity of the time series in Figure 2 shows that the assumption of homogeneity over the warm pool region is reasonable.” Unfortunately, there is considerable evidence to dispute the veracity of these claims. [4] The apparent substantial cooling during the final 10 years of the record can be explained largely by a change in radiosonde instrument type, specifically the transition from Viz to Vaisala, hereafter referred to as the VVT. Abrupt cooling upon introduction of Vaisala radiosondes was noted by Parker et al. [1997] for stations in Australia and New Zealand in the mid to late 1980s, and for U.S. controlled tropical Pacific stations beginning in 1995 [Stendel et al., 2000]. More generally, there has been a systematic worldwide change to Vaisala radiosondes since the 1980s with an expected artificial cooling, especially in the stratosphere [Lanzante et al., 2003a; Christy et al., 2003; CCSP, 2006] (see also Integrated Global Radiosonde Archive (IGRA) station history metadata available at http://www1.ncdc.noaa.gov/pub/data/igra/igra-metadata.txt). [5] The IGRA metadata indicate that the VVT occurs in November or December 1995 for five of the six stations shown in Figure 1 of RR08 (Lihue, Truk, Koror, Yap, and Pago Pago). The literature cited above suggests that the major part of the drop in stratospheric temperature after 1995 (as seen quite clearly in Figure 5 of RR08) is almost certainly artificial. To further illustrate this point, Figure 1a shows several time series of stratospheric temperatures at Koror. The thick black curve is based on the IGRA data [Durre et al., 2006], which were used by RR08. Note that the IGRA data, although of very high quality (i.e., they have been carefully checked for consistency in time and in the vertical, random errors and outliers, as well as duplicate soundings, etc.), have not been adjusted to account for artificial inhomogeneities. A number of teams have produced adjusted radiosonde temperature time series, each using a different homogenization method. Also shown in Figure 1a are corresponding adjusted time series (HadAT2) [Thorne et al., 2005], RATPAC [Free et al., 2005], IUK [Sherwood et al., 2008], RAOBCORE 1.4 [Haimberger et al., 2008], and RICH [Haimberger et al., 2008] produced by these teams. In comparing the IGRA with the various adjusted time series in Figure 1a it is seen that the adjusted versions all show much less long-term cooling, as indicated by the accompanying trend lines. A closer look at Figure 1a reveals that most of the difference is attributable to the transition across 1995. The adjusted time series shows a relatively small drop while that for IGRA shows a much larger one. [6] The similarity of the outstanding features of the time series between stations bolstered RR08's confidence. However, this was ill advised because these stations also tend to have similar and coincident changes in instruments as indicated by station histories. To further illustrate the interconnected nature of these problems, Figure 2 shows comparisons between unadjusted (IGRA) and adjusted versions of the 50 hPa time series made for each of the stations used by RR08. The metric used in the comparisons is the difference in 10-year means before and after the VVT, as an approximate measure of the magnitude of the instrument-induced step function change in temperature. Except for Majuro the estimated difference is reduced substantially when the data are adjusted. Even at Majuro 4 of the 5 differences for adjusted data are less than for IGRA. On average, adjustment reduces the difference by nearly a factor of three. Calculations using the linear trend (not shown) instead of the difference of 10-year means produce qualitatively similar results. [7] Note that methods used to identify and adjust for artificial discontinuities carry with them some degree of uncertainty. While evidence presented here suggests that the major part of the stratospheric cooling at the end of the record is artificial, there is reason to believe that estimates of the magnitude of the discontinuities and their effects on the trend are likely lower bounds. For example, the homogenization method used to produce the HadAT2 data set tends to underestimate the magnitude of discontinuities [McCarthy et al., 2008; Titchner et al., 2009] when the biases at different stations tend to be of the same sign (negative for cooling biases in this case); RICH and HadAT2 share the same adjustment algorithm, although they differ in the way in which discontinuities are identified. Evidence of similar biases has been found in other adjusted data (RATPAC [Randel and Wu, 2006] and IUK [Sherwood et al., 2008]). Thus, the stratospheric cooling of the 1990s is likely less than shown for the adjusted data in Figure 1a. [8] Another feature considered by RR08 at their warm pool stations is the abrupt rise in temperature of about 4 K in early 1999, followed by an abrupt drop of about 6 K about 2 years later. While it is not as easy to attribute this feature to artificial causes, there is good reason to be suspicious. The station history metadata indicate a computer system change at all six of their select stations in early 1999. Data from the Upper Air Research Satellite/UK Meteorological Office (UARS/UKMO) assimilation [Swinbank and O'Neill, 1994] is used by RR08 to lend further credence to this feature. It should be noted that this product blends observations from radiosondes and satellites together. As depicted in Figure 4 of RR08, this feature is unlike all other short-term warming or cooling events in the record. While all other events of similar amplitude have a deep vertical extent, this feature is limited to a shallow layer ∼ 100–50 hPa. It is well established that radiosonde discontinuities are often limited to one or several adjacent vertical levels [Lanzante et al., 2003a] and procedural changes, rather than changes in measuring devices are often to blame. Unfortunately, the available metadata are not as up-to-date as would be desirable as they are based largely on an earlier source [Gaffen, 1996]. While attempts to keep these metadata updated to present are ongoing, efforts initiated for another project [Free et al., 2005] demonstrated that results are incomplete. Therefore it is not clear to what extent measurement-related changes made since the turn of the century have been documented. [9] Further analyses conducted by RR08 are aimed at demonstrating the larger-scale nature of the aforementioned features. In this pursuit they utilize the UARS/UKMO product along with NCAR/NCEP reanalyses and a larger collection of 52 tropical radiosonde stations from the IGRA collection. Because all three products utilize radiosonde data, these cannot be used to independently verify any outstanding features. In examining Figure 7 of RR08, which shows the temperature difference between the time periods 2001–2003 and 1995–1997, it is curious that by far the largest differences in the tropical zone occur in the region dominated by the radiosonde stations having suspect continuity at this time. In conjunction with their Figure 8, RR08 note that stations remote to the warm pool region did not exhibit the temperature drop at the end of 2000, except for Niamey in Africa. This finding raises further doubts since Lanzante et al. [2003a, 2003b] examined in detail several African stations, including Niamey, and concluded that these were the least reliable in terms of continuity from the global collection. [10] The larger collection of 52 tropical stations used by RR08 to construct the time series shown in their Figure 6 consists of six whose station histories indicates the VVT in 1995 (Ponape in addition to the five cited earlier). An eyeball estimate of RR08's Figure 6 suggests that the drop in temperature from say 1986–1995 to 1996–2005 is ∼0.5 K. Given the earlier estimates herein from Figure 2 of an artificial drop of ∼2 K, the six stations in question (out of 52) could account for nearly half of the drop shown in Figure 6 of RR08. It is quite conceivable that most or all of the remainder could be explained by spurious drops in temperature at other stations. This conjecture follows from the findings of Sherwood et al. [2005] (especially Figure 3) that spurious drops in tropical radiosonde temperatures were largest in the mid to late 1990s. [11] Although satellite temperature time series are plagued by inhomogeneities of their own, such measurements can be used to shed some light on the problem at hand. Christy et al. [2003] performed a direct comparison at Truk between radiosonde and Microwave Sounding Unit (MSU) satellite temperatures for the lower stratosphere. Their Figure 7 shows an abrupt drop in radiosonde temperature ∼ 2 K in late 1995 (right at the time of the VVT) as compared to that derived from the MSU. This comparison is complicated by the fact that a satellite transition occurred in 1995 [CCSP, 2006], although the transition took place nearly a year prior to the discontinuity reported by Christy et al. [2003]. Similar comparisons at Majuro reported by Randel and Wu [2006] indicate a cooling of the radiosonde time series derived from the data set of Lanzante et al. [2003a] compared to MSU during the late 1980s, even after homogeneity adjustment had been applied to the radiosonde measurements. [12] A limited examination of some satellite temperature data is made here to complement the analyses involving radiosonde data. Temperatures derived from the MSU produced by two different homogenization teams (RSS [Mears et al., 2003] and UAH [Christy et al., 2003]) are shown in Figure 1b, along with equivalent measures derived from IGRA and RATPAC radiosonde data. The measure used (TLS) represents a weighted, vertically averaged temperature in the lower stratosphere. The weighting function for TLS peaks just above 100 hPa, with most of the weight from 30 to 150 hPa. The time series in Figure 1b show satellite and adjusted radiosonde temperatures in accord, with much less of a downward trend and jump down across 1995 than for IGRA. In Figure 2, the decadal differences at Koror involving the satellite measures (K*) show the same relationship to IGRA as do the differences based on the various adjusted radiosonde data sets (K). Thus, the satellite data examined here are consistent with that from adjusted radiosonde temperature data sets. Finally, regarding the spike up in temperature starting ∼1999 lasting for about 2 years reported by RR08, the satellite data in Figure 1b show a smaller amplitude spike of less than half of that indicated by the radiosonde data. [13] In conclusion, serious doubts have been raised regarding the veracity of noteworthy temperature changes in the tropical stratosphere reported by RR08. Considerable confidence in this statement is supported by the consistency across 5 adjusted radiosonde data sets and 2 satellite data sets that provide measurements independent of any radiosonde data. Specifically, the apparent association between historically warm SSTs and cold stratospheric temperatures in the tropical warm pool region at the end of the record is questioned. While the outstanding features found by RR08 cannot be totally dismissed, they are at best exaggerated, and at worst nonexistent. [14] The remainder of this note responds to the reply by Rosenlof and Reid [2009] (hereinafter referred to as RR09), as it relates to this comment. An initial point that RR09 raise is that results shown here are for 50 hPa while RR08 examined the 70 and 100 hPa levels as well. The distinction is not important for the warm pool stations because examination of the time series indicate that all of the discontinuities found at 50 hPa are present at 70 and 100 hPa as well. [15] An attempt is made by RR09 (including their Figure 1 and Table 1) to dispute the artificial component of change in the warm pool stations, as illustrated here in Figure 1 for Koror, by examining the character of monthly time series. They note a sharp drop in temperature beginning in 1993 which continues over a period of a year. They further note that the documented VVT occurs nearly 2 years later. The fallacy in their argument lies in the fact that instrument changes and climate noise are superimposed, sometimes interfering and sometimes reinforcing one another, rendering the recorded time series itself often of limited value in identifying discontinuities. This is the reason that the teams that produced all five homogenized radiosonde data sets used various external reference series in this endeavor. Examination of difference curves (original minus homogenized) (not shown) clearly identify the discontinuities for the VTT in late 1995, which is in accord with Parker et al. [1997] and Christy et al. [2003]. [16] The timing of discontinuities is further questioned by RR09 “The time period of possible suspect continuity was previously placed in the mid-1990s but is now extended to the period around the turn of the century … one would have to assume that the radiosonde stations of the western tropical Pacific have been in a constant state of turmoil since the early 1990s, after having apparently been stable and well behaved for at least the preceding 30 years.” RR09 later state “When five independent radiosonde time series from five stations in the western tropical Pacific all show similar dramatic changes … one is forced to the unlikely conclusion that all stations changed instruments or data procedures at exactly the same time.” [17] First, there is no single “period of possible suspect continuity.” For those stations experiencing the VVT during the mid-1990s the discontinuity is large, highly influential in the stratosphere, and very confidently identified by the various radiosonde teams. But aside from this, vigilance is needed throughout the record since no time period is immune. Second, there is a good deal of truth in their characterization for the warm pool stations, whose instrumentation has followed that of the U.S. through their lifetimes. Historically U.S. stations used exclusively VIZ instrumentation until the last decade or so. Globally, modernization has taken the form of transitioning to Vaisala instrumentation, and has had a particularly striking effect in the tropics because instrumentation was of lower quality initially. Evidence of the acceleration of this effect during the late 80s and 90s is shown in Figure 3 of Sherwood et al. [2005]. Third, the fact that the stations all changed at the same time is not at all surprising since changes are often made simultaneously as dictated by country of control. The widespread VVT has been documented, for example, by Parker et al. [1997] and Christy et al. [2003]. [18] To their credit, RR09 bring to bear additional evidence to support the abrupt rise/fall of temperature near the turn of the century in the form of satellite measures of temperature and water vapor. To bolster their case, they would need to extend their analyses beyond the single station shown in their Figure 3 to at least the additional warm pool stations plus those shown in their Figure 2, particularly in light of the fact that the station they chose to highlight is the least problematic of the warm pool stations (see Figure 2). Note also that analyses (not shown) based on homogenized data and metadata very strongly suggests that the long-term continuity of Koror, Lihue, Ascension, Darwin, and to a lesser extent Singapore, as depicted in Figure 2 of RR09 have been compromised by the VVT. [19] A final but important point raised several times by RR09 is their claim that the magnitudes of the changes do not matter, rather the character of the time series: little cooling in the 1970s and 1980s with more substantial cooling during the 1990s. Furthermore, they claim any artificial changes are not large enough to affect the main conclusions of RR08. They conclude “…that the conclusions of RR08 are robust and valid and that the criticisms of L08 are largely irrelevant.” [20] In fact, the magnitudes of changes are important in two different contexts. First is the question as to whether a real (physically and statistically significant) change has occurred toward the end of the record. One gauges such a change by the magnitude of the change in comparison to the natural variability or climate noise. As a concrete example the reader is referred to Figure 1b. Taken at face value, for the IGRA data there is a dramatic and most surely a significant change in climate after about 1995. On the other hand, for the other data sets any such change is much more subtle if in fact it occurs at all. Similarly, in Figure 3 of RR09 the unadjusted radiosonde data suggest a much more dramatic change than do the satellite products, in which values at the end of the series are comparable to those toward the beginning. Furthermore, it is likely that homogenization does not remove all of the spurious cooling bias [McCarthy et al., 2008; Titchner et al., 2009] so that the apparent cooling even in the homogenized series is overstated. [21] Second, any major discontinuities affect the assessment of short-term variability. In their abstract, RR08 state that “… only weak cooling trends occurred before the 1990s, but a strong and rapid cooling of 4° to 6°C took place in the mid-1990s and has persisted since that time.” When homogenized data are considered (see Figure 2), the short-term cooling is reduced to 2° to 4°C and its persistence is then highly debatable. This undermines the connection between historically high SSTs in the warm pool and purportedly historically low overlying stratospheric temperatures which is the main focus of their paper. [22] The IGRA and RATPAC radiosonde data and IGRA station history metadata were obtained from the National Climatic Data Center's (NCDC) Web site (http://www.ncdc.noaa.gov/oa/ncdc.html), the HadAT2 radiosonde data from the Hadley Centre's website (http://hadobs.metoffice.com/), and the IUK radiosonde data from Steve Sherwood's homepage at Yale University (http://earth.geology.yale.edu/∼sherwood). Leo Haimberger (University of Vienna) provided the RAOBCORE 1.4 and RICH radiosonde data. The satellite MSU temperature data and radiosonde vertical weighting functions were provided by Carl Mears (Remote Sensing Systems). Gabriel Lau, Gabe Vecchi, Peter Thorne, and an anonymous reviewer kindly provided comments which helped to improve the manuscript.