Upper-air temperature is a basic variable measuring the atmospheric state and its change, and thus an important indicator for the climate change. Temperature data are generally obtained from in-situ and satellite remote sensing. In-situ data are from surface instruments and upper air radiosonde balloons. Instrumental records of surface air temperatures dated back to the 1850s while radiosonde upper air temperatures only extended back to the 1950s. Satellite observations from the Microwave Sounding Unit (MSU) and their follow-on Advanced Microwave Sounding Unit (AMSU) have been used to monitor temperature changes from the lower troposphere to the upper stratosphere since the late 1970s (Spencer and Christy, 1990; Zou et al., 2006; Zou and Wang, 2011; Zou and Qian, 2016). In-situ and satellite remote sensing are two distinct systems with their own advantages and limitations. In-situ data have longer time series with higher vertical resolution but are limited by sparse spatial sampling and discontinuities due to changes in instruments and processing methods. Satellite remote sensing data have a global coverage with the higher sampling rate but are limited with the coarse vertical resolution and shorter period of data records.

The satellite MSU/AMSU data have been inter-calibrated and homogenized to generate the atmospheric la-yer mean temperature time series from the late 1978 until present. Three groups developed such climate data records (CDRs), including the NOAA Center for Satellite Applications and Research (STAR; Zou et al., 2006; Zou and Wang, 2011), Remote Sensing Systems (RSS; Mears et al., 2003; Mears and Wentz, 2009a, b, 2016, 2017), and University of Alabama Huntsville (UAH; Christy et al., 2003; Spencer et al., 2017). The University of Washington (UW) group also developed the mid-tropospheric temperature data record from MSU/AMSU observations, but its data are limited to the deep tropics (Po-Chedley et al., 2015). In addition to satellite temperature CDRs, radiosonde data were homogenized to provide an independent source for assessing upper air temperature changes. Global homogenized radiosonde datasets were derived by different research groups, which include HadRT and HadAT (Hadley Centre; Parker et al., 1997; Thorne et al., 2005b), RATPAC (The Radiosonde Atmospheric Temperature Products for Assessing Climate, NOAA; Free et al., 2005); RAOBCORE (The Radiosonde Observation Correction using Reanalysis; Haimberger, 2007) and RICH (The Radiosonde Innovation Composite Homogenization at University of Vienna; Haimberger et al., 2012), and IUK (The Iterative Universal Kriging Radiosonde Analysis Project; Sherwood and Nishant, 2015).

Over the past two decades, there have been many studies on inter-comparisons of both statistical characteristics and atmospheric temperature trends between the MSU and radiosonde observations (Hurrell et al., 2000; Santer et al., 2000; Seidel et al., 2004; Christy and Norris, 2006, 2009; Randel and Wu, 2006; Haimberger et al., 2008; Randall and Herman, 2008; Thorne et al., 2011; Mears et al., 2012; Po-Chedley and Fu, 2012). Common global-scale features from radiosonde and satellite temperature observations are warming in the troposphere and cooling in the stratosphere during the satellite era. However, differences in atmospheric temperature trends existed between the MSU and homogenized radiosonde records (Seidel et al., 2009). Seidel et al. (2011) and Thorne et al. (2011) provided a comprehensive review on these trend differences in the troposphere and stratosphere, respectively. It was unclear, however, whether these trend differences were caused by errors in radiosonde or satellite homogenization procedures. Construction methodologies for both homogenized MSU and radiosonde datasets may contain biases that lead to trend differences (Thorne et al., 2005a; Christy and Norris, 2006, 2009; Mears et al., 2012; IPCC, 2013). It would be a privilege if benchmark references exist that could constrain temperature trends observed by different types of instruments. Recent studies by Santer et al. (2017a, b) suggested that the mid-tropospheric temperature trends derived by two of the three aforementioned research groups in their latest versions of satellite data records agreed with each other very well. Zou et al. (2018) indicated that the new generation of satellite microwave sounders have achieved high radiometric stability performance for use as references for adjusting radiosonde observations. In fact, Christy et al. (2018) have already used satellite data as references to adjust the original radiosonde observations on the global scale by using a shift-point approach.

Although studies on atmospheric temperature trends from satellite and radiosonde data are rich at the global scale and over the tropics, few studies have focused on the regional scales, especially over China with its own unique radiosonde network. The in-situ radiosonde network in China has been established since the 1950s which includes about 120 stations, ranging from tropical and subtropical to temperate and sub-frigid zones. Homogenizations of radiosonde data over China started from the mid-1990s (Zhai and Eskridge, 1996; Zhai, 1997; Guo et al., 2008; Guo and Ding, 2009, 2011). A new version of quality controlled, twice-daily upper-air sounding data and homogenized monthly upper-air temperatures over China at nine mandatory levels were released by the National Meteorological Information Center/China Meteorological Administration (NMIC/CMA) at the end of 2013. This version is more complete than previous radiosonde temperature datasets over China. Metadata including instruments and position changes at individual stations over China were applied in homogenizations (Chen and Yang, 2014; Ruan et al., 2015).

The purpose of this study is to assess the consistency of upper-air temperature trends from MSU and radiosonde observations over China. Layer mean temperature anomalies during 1979–2015 from the lower-troposphere to lower-stratosphere from MSU and in-situ observations over China are compared at individual stations and for averages over China. The advantage in focusing on Chinese radiosonde records is that changes and updates in radiosonde instruments occurred in similar time periods over China with good metadata records; as a result, trend differences between MSU and radiosonde data could be better understood by examining details of radiosonde metadata. This understanding could help the reconciliation of radiosonde and satellite observations in terms of accurate trend determination. In addition, with the sufficient number of radiosonde stations over China, trend differences between radiosonde and MSU satellite data could be derived with good statistical confidence, providing clues on their differences for regions other than China.

Section 2 describes the details of data and outlines the comparison method used in our study. Section 3 describes the statistical characteristics of differences between MSU and radiosonde data, and identifies challenges to integrate data over the Tibetan Plateau. Section 4 analyzes the temperature differences between MSU and radiosonde observations, explains these differences based on available radiosonde metadata and satellite merging approaches, and re-adjusts the radiosonde observations by using satellite data as a reference. Section 5 provides the comparison of upper-air temperature trends over China from MSU and radiosonde observations. Section 6 contains the conclusions and discussion.

The in-situ data applied in this work include the original and homogenized monthly air temperatures at the surface and nine mandatory levels (850, 700, 500, 400, 300, 200, 100, 50, and 30 hPa) at 114 stations over China during 1979–2015. Stations with the minimum data availability above 70% (Guo and Ding, 2009) are selected as the research objective in this study (Fig. 1).

Figure 1 Distribution of the radiosonde observation network in China. Black circles represent 114 stations in the study, green crosses are stations with surface elevation above 2000 m, and pentacles are the selected representative stations.

Figure options

Raw and homogenized temperatures over China are denoted as RAW and ADJ respectively for simplification. RAW data at the surface are from China National Stations Fundamental Elements Datasets V3.0 (Ren et al., 2012). RAW data at nine mandatory levels are derived from the twice daily radiosonde observations (0000 and 1200 UTC) in China (Ruan et al., 2015), which are quality-controlled to remove erroneous records incongruent with synoptic meteorology. ADJ data are from RHtest (V3) (Wang et al., 2007) homogenized by PMTred algorithm to detect multiple changing points. Surface air temperatures are adjusted by the reference from neighbor stations (Xu et al., 2013; Cao et al., 2016), while radiosonde pressure temperatures are adjusted by the reference from nighttime observations, ERA-40, and ERA-Interim (Chen and Yang, 2014). All the detected changing points are confirmed by metadata records at stations over China.

Considering the large uncertainty in homogenization of radiosonde temperatures with different techniques, it is necessary to include the other quality controlled and homogenized radiosonde temperatures for a comparison in this study. Among the homogenized datasets (RAOBCORE, RICH, HadAT, IUK, and RATPAC), only HadAT2 contains stations similar to but with different homogenization algorithms from ADJ over China. In addition, the station numbers over China from IUK or RATPAC are much lower than those from ADJ, while the RAOBCORE and RICH apply ERA-40 and ERA-Interim to determine the break points which are similar to ADJ in reference selections. For these reasons, only HadAT2 is selected as a contrast to ADJ in comparison to satellite data. HadAT2 consists of monthly radiosonde temperatures at nine pressure levels (850, 700, 500, 300, 200, 150, 100, 50, and 30 hPa) and at 676 stations over the globe (92 stations over China) during 1958–2012. Surface temperatures from ADJ and temperatures at pressure levels from HadAT2 over China are combined to form another set of homogenized radiosonde dataset (hereinafter referred to as HAD) for the mutual validation with ADJ. The difference between ADJ and RAW is used to evaluate the impact of discontinuity on homogenizations, while the difference between ADJ and HAD is used to assess uncertainties of homogenizations resulting from different references or algorithms.

The satellites MSU/AMSU data used in this study are monthly mean layer temperatures from three groups: STAR, RSS, and UAH. The STAR data (Zou et al., 2006; Zou and Wang, 2011) include version 4.0 of Temperature Middle-Troposphere (TMT) and Temperature Lower-Stratosphere (TLS) from 1979 to 2015 and Temperature Upper-Troposphere (TUT) during 1981–2015. The RSS data (Mears and Wentz, 2009a, b, 2016) include Version 4.0 of Temperature of Lower-Troposphere (TLT), TMT, and TLS during 1979–2015, and TUT (also known as Temperature Troposphere Stratosphere) during 1987–2015. The UAH data include TLT, TMT, TLS, and TUT (also known as Temperature Tropopause) during 1979–2015 in the latest version 6.0 (Spencer et al., 2017).

In order to directly compare in-situ temperatures over stations with the MSU gridded layer mean temperatures, the former has been converted to satellite layer mean temperatures for the four layers (TLT, TMT, TUT, and TLS) by vertically integrating surface and atmospheric temperatures at the nine pressure levels (Spencer and Christy, 1992) with MSU weighting functions downloaded from the RSS website (http://www.remss.com/measurements/upper-air-temperature; Fig. 2). These equivalent layer-mean temperatures are simply referred to as the RAW and ADJ for simplicity in the following discussions. The radiosonde equivalence at station 45004 is converted by temperatures at pressure levels from IGRA (Integrated Global Radiosonde Archive; Durre et al., 2006).

Figure 2 Weighting functions for MSU channels of TLT, TMT, TUT, and TLS. Dots represent in-situ measurements at 10 levels in this study.

Figure options

Since data availability varies with pressure levels and stations (grids) and thus affecting the calculation of MSU equivalent, the valid pressure levels in each month at individual stations are fixed to ensure the temporal and spatial compatibility. For the MSU equivalent of ADJ and RAW, temperatures at the surface and nine pressure levels from 850–30 hPa must be all available. As anot-her note, the data availability in HAD (ended in 2012) is generally less than ADJ for stations over China with exceptions over the stations 52652 (Zhangye), 56137 (Changdu), and 59981 (Xisha). This is because ADJ is derived from all available digital sources over China. To ensure that the MSU equivalent from HAD has enough valid stations to compare with ADJ, the requirement of data availability for HAD has been relaxed that allows one or two levels to be missing.

Figure 3a shows comparisons of the missing rates (%) for TLS at 114 stations between ADJ and HAD. As shown, HAD has more stations with the higher missing rate (> 30%) while fewer stations from ADJ with missing rates > 30%. Three such stations (Changdu 56137, Qingyuan 59280, and Xisha 59981) are selected (Figs. 3b–d) to understand how the missing rate changes at pressure levels. In general, higher missing values in ADJ occurred mainly over the Tibetan Plateau and southern China, as represented by the three stations (Figs. 1, 3a). Missing data at individual pressure levels for the three stations occur mainly above 200 hPa and especially at 30 hPa (Figs. 3b–d), although they change with the season for different stations. The highest missing rates are found at 52652 (Zhangye) and 52533 (Jiuquan) (Fig. 3a), which are caused by artificially setting temperatures of ADJ at 850 hPa as missing. Obviously, this setting is not suitable when surface pressures are higher than 850 hPa.

Figure 3 Missing data rates (%) of TLS at 114 stations over China from radiosonde equivalents and monthly missing data rates from ADJ at surface and pressure levels at three representative stations during 1979–2015. (a) TLS (114 stations), (b) station 56137, (c) station 59280, and (d) station 59981.

Figure options

In the following discussion, the mean temperature over China is an area-weighted average at the 114 stations or corresponding grids. The climatology is based on a common period of 1981–2010 with the exception that TUT of RSS started from 1987. Unless stated otherwise, comparisons are made between collocated grid cells from MSU and stations from the in-situ.

China not only has a vast land area covering tropical, temperate, and sub-frigid zones, but also has a complicated topography with the average elevation of the Tibetan Plateau more than 4000 m above the sea level. Among the 114 radiosonde stations over China, 16 of them have a surface elevation greater than 2000 m. Of these 16 stations, most are located in the Tibetan Plateau (Fig. 1). It is of interest to understand how the MSU and radiosonde data are correlated in such a complicated topography. Note that for the high mountain and plateau areas, it is ideal to use MSU weighting functions specifically calculated for these areas to convert radiosonde data to MSU equivalents. However, such weighting functions are unavailable and thus we used the standard MSU weighting functions that are based on the standard atmosphere for conversions from radiosonde to satellite equivalents for all stations over China. Alternatively, using radiative transfer models to calculate MSU equivalent temperatures is a viable approach for both plain and plateau areas and worth carrying out in future studies.

Figures 4 and 5 show correlations and root mean square deviation (RMSD), respectively, between the MSU datasets from different groups (STAR, RSS, and UAH) and ADJ for different layer temperatures. Statistically significant positive correlations greater than 0.9 and consistent RMSD are found for TLS over the entire China for all the different MSU datasets. This suggests that the weighting function for TLS, which peaks near 70 hPa, is high enough so that the high-altitude Tibetan Plateau does not have an effect on the radiosonde conversion. For TUT, the relatively lower correlation and larger RMSD are found over southern China and Tibetan Plateau for all the satellite datasets. This is related to radiosonde data issues near the tropopause in these areas as discussed in the previous section. Given that its weighting function peaks (near 250 hPa) higher than the Tibetan Plateau, the topography has a negligible effect on the radiosonde equivalent for TUT.

Figure 4 Correlation of monthly temperature anomalies between MSU (STAR, RSS, and UAH) and ADJ over China during 1979–2015. So-lid circles indicate the st atistical significance of correlation (> 0.35). Panels from top to bottom are for TLS, TUT, TMT, and TLT; panels from left to right are for STAR, RSS, and UAH.

Figure options

Figure 5 RMSD of monthly temperature anomalies between MSU (STAR, RSS, and UAH) and ADJ over China during 1979–2015. Panels from top to bottom are for TLS, TUT, TMT, and TLT; panels from left to right are for STAR, RSS, and UAH.

Figure options

For TMT, the large correlation greater than 0.9 and consistent RMSD are found over most regions in China for all the three MSU datasets except over the Tibetan Plateau. This occurred because the MSU TMT weighting function peaks at about 500 hPa, close to the surface elevation of the Tibetan Plateau. This weighting function is not expected to work well for converting the radiosonde data to TMT equivalent over the Tibetan Plateau. Larger differences, as represented by RMSD over most Tibetan Plateau regions (Fig. 5), between the MSU and radiosonde equivalent based on such a weighting function inevitably occur. For TLT, high correlations above 0.9 are found for both RSS and UAH, but RSS RMSD is apparently larger than those of UAH for most regions. Relatively smaller correlations and larger RMSD are found over the Tibetan Plateau for both UAH and RSS. Similar to TMT, this is expected as the MSU TLT weighting function peaks below the surface of the Tibetan Plateau where it is inappropriate for conversions from radiosonde level temperatures to satellite layer temperatures. Hence, it is better to remove these data for more accurate trend estimates over China. It is seen that beyond 2000 m the correlation decreases and RMSD deceases more rapidly with the surface elevation; as a result, the 2000-m elevation is a reasonable criterion to separate stations for surface elevation effects. In the following discussion, regional mean anomalies and trends over China are generally calculated over areas where surface elevations are less than 2000 m (hereinafter being referred to as Cn-notb in Tables 1, 3), although tests have been conducted to include stations with elevations above 2000 m (Cn, Cn-sta, and Cn-nosta in Table 3).

Table 1 summarizes the comparison statistics associated with findings in the previous subsection in terms of the correlation and RMSD between MSU and the radiosonde datasets before (RAW) and after homogenizations (ADJ and HAD) during 1979–2015. Statistics are for stations in China without the station elevation above 2000 m (Cn-notb). The summary emphasizes common features in the differences and agreement between the satellite and radiosonde data before and after the homogenization. The STAR data are used to represent satellite MSU data for TMT, TUT, and TLS, while the RSS data are used for TLT. Statistics for the same layer temperatures but from UAH and RSS datasets show similar results, although their correlations with the radiosonde data are slightly higher for certain layer temperatures. This will not affect discussion on differences between the radiosonde and MSU data.

The results revealed the excellent agreement between MSU and homogenized radiosonde observations with statistical significant positive correlations greater than 0.9 and similar RMSD for TLT, TMT, and TLS. Correlations for TUT are lower than those of the other layer temperatures. The radiosonde homogenization and readjustment obviously improve the agreement of ADJ and ReADJ (the re-adjusted ADJ time series, details in Section 4.2) with MSU data. Both ADJ and HAD show a better agreement with the MSU dataset than RAW by the remarkable higher correlation coefficient and smaller RMSD for TMT, TUT, and TLS. For TMT, MSU correlations go from 0.86 with the RAW up to 0.92 with the ADJ (HAD), 0.96 with ReADJ. Similar improvements are also found for TUT. As discussed in Fig. 6, this resulted from reduction of the discontinuity in raw radiosonde temperatures. On the other hand, correlations between MSU and radiosonde data remain the same for TLT before (RAW) and after (ADJ and HAD) homogenizations. The reason for this is that the impact of homogenizations for TLT is generally much smaller than the other three layers (Chen and Yang, 2014). The reason why HAD agrees better than ADJ with MSU is the impact of different references used in homogenizations (Guo and Ding, 2011). Since radiosonde observations are an input source for reanalysis, the independence between the RAW and reference causes the inhomogeneity in nighttime temperature time series hard to be detected or adjusted thoroughly with the reference from ERA-Interim at individual stations. While the reference of HAD is based on the larger scale network and thus effectively minimizing the large systematic biases in raw data (Thorne et al., 2005b). This approach complements the deficiency in ADJ reference and, as a result, improves the consistency with MSU.

Figure 6 Annual temperature anomalies (left panels; K) averaged over China during 1979–2015 from MSU (STAR, RSS, and UAH) and radiosonde (RAW, ADJ, and HAD), and their differences (right panels; K) from RAW. (a1, a2) TLT, (b1, b2) TMT, (c1, c2) TUT, and (d1, d2) TLS.

Figure options

Figure 6 presents annual temperature anomalies averaged over China during 1979–2015 from radiosonde and MSU datasets (Figs. 6a1–d1) and their differences from the RAW (Figs. 6a2–d2). Note that the use of RAW as a common reference for differencing here is to help assess the impact of homogenizations from HAD and ADJ. Annual anomalies show a good agreement between the MSU datasets and radiosondes in terms of similar variability. The time series of differences shows more diversity in TUT and TLS than that in TLT and TMT. Comparing with MSU anomalies, RAW anomalies show an apparent sharp decrease nearly 0.5 K in 2001, causing the entire time series of differences, which are divided into negative and positive phases across 2001. This large shift is closely related to the radiation correction and radiosonde system update in China during 1999–2001 as well as the replacement of GZZ2 with GTS1 (L-band) during 2002–10 over China reported in radiosonde metadata records (Chen and Yang, 2014; Guo et al., 2016). Both HAD and ADJ have detected the shift around 2000 and adjusted the RAW to a certain extent. Meanwhile, MSU anomalies are smoother around radiosonde jumping points, indicating that MSU data could be used as potential references to validate original radiosonde temperatures. Comparing with ADJ, HAD adjusts the RAW more substantially (larger difference with RAW) and has better agreements with MSU datasets. This suggests that ADJ may have residual discontinuities and HAD is more reliable especially during the 2000s.

The discontinuity problem shown in regional mean anomalies can be better illustrated by anomalies at individual stations. This is because the station metadata, with exact information on the station location and instrument changes, are more accurate for quantifying the artificial influence caused by observing system changes. As an example, monthly temperature anomalies from ADJ, HAD, and the three MSU datasets at station 52681 (which has the most complete record) are shown in Fig. 7. Although such a time series is noisier than that of the annual and regional averages, stepwise changes and timing in the anomalies match very well with the radiation correction and radiosonde system updating during 1999–2001 and instrument change from GZZ2 to GTS1 in January 2006 at this station. Both ADJ and HAD have detected changing points and adjusted RAW with different magnitudes. Similar to Fig. 6, HAD adjusts RAW with the larger magnitude than ADJ since the former has larger differences from RAW in 2001. Generally, HAD has a better agreement with MSU datasets than ADJ in TLT, TMT, and TUT (Table 1), due to different homogenization procedures. In contrast to the obvious discrepancy between HAD and ADJ, MSU datasets show remarkable high consistency in their differences from RAW with the similar phase and magnitude in variability. This suggests that MSU data from different groups either all represent the reality very well or have common issues in their merging, especially during the period for radiosonde system updating or instrument change in the 2000s.

Figure 7 Monthly temperature differences (K) with respect to RAW at station 52681 (Minqin) from ADJ, HAD, and the three MSU datasets (STAR, RSS, and UAH) during 1979–2015.

Figure options

Five-year moving trends of the monthly differences at individual radiosonde stations can reveal more information on timing when differences between the MSU and radiosonde occur (Randel and Wu, 2006; Randall and Herman, 2008; Mears et al., 2012). Figure 8 shows 5-yr moving trends of differences at four different stations for TUT between the data pairs. These four stations are representatives of the northwestern, northeastern, northern, and eastern China (52681 Minqin, 50953 Harbin, 54511 Beijing, and 58362 Baoshan; Fig. 1). These 5-yr moving trends are composed of the maximum and minimum at different times, representing bias jumps between the data pairs at the specific time. Roughly speaking, maximum represents the time when ADJ, HAD or MSU data start to become warmer than RAW while the minimum is opposite. As a result, the maximum and minimum in Fig. 8 represent the most significant changing points between the data pairs. Table 2 summarizes the times when large changes occur at these stations between MSU and RAW data pairs. The maximum around 1999–2001 and mini-mum around 2004–06 occurred for all the four stations and they matched well with the two major changes in the radiosonde observations: system update over China during 1999–2001 and instrument mode replacement at station 52681 in January 2006, at station 50953 in January 2005, at station 54511 in January 2002, and at station 58362 in August 2003. Four major changes are identified in 1983, 1986, 1988, and 1991 for at least two of the four stations (Fig. 8). Metadata for these stations indicate that there are no major radiosonde changes during this period. This is why ADJ is close to RAW during this period for all the four stations. Consequently, these changes are most likely related to satellite changes. Indeed, the years for these changes respectively coincide well with the launch time of NOAA-8, NOAA-10, NOAA-11, and NOAA-12 satellites when MSU data onboard these satellites become available to merge with previous satellites for development of the MSU time series. These consistencies are also found for differences between the other MSU layer temperatures and RAW data. Such information suggests that radiosonde data at individual stations can help to identify bias jumps in the merged satellite time series, which will in turn help the satellite data developer to improve the merging accuracy of satellite CDRs.

Figure 8 Five-year moving trends for differences (K decade–1) of TUT with respect to RAW from ADJ, HAD, and the three MSU datasets (STAR, RSS, and UAH) during 1979–2015 at the four stations: (a) station 52681 (Minqin), (b) station 50953 (Harbin), (c) station 54511 (Beijing), and (d) station 58362 (Baoshan).

Figure options

Similar breakpoints are also found for differences between ADJ and RAW during 1999–2001 and 2002–10 for individual stations (Figs. 7, 8) and regional averages (Fig. 6), although magnitudes of the breakpoints vary. Given the fact that differences between MSU and RAW are much larger during the radiosonde transition period of 1999–2001 and differences between ADJ and RAW are much smaller during the same period (Fig. 6), it is likely that the adjustment in ADJ is not large enough to remove biases caused by the radiosonde system update. In other words, residual biases may still remain in ADJ after 2001. In order to reduce its impact on trends in the homogenized radiosonde data, we perform further re-adjustment on ADJ by using MSU averages over the three groups (abbreviated as MSUav) as a reference. We use a shift-point adjustment approach similar to those used in Christy et al. (2018) for re-adjustment. However, instead of conducting re-adjustment at each individual station, we only re-adjust the mean time series over China as a demonstration of the adjusting concept. In this approach, the maximum magnitude of difference shift between RAW and MSU datasets in Fig. 6 is firstly selected as shifting points, being 2001 for events of the 1999–2001 radiosonde system update and 2005 for events of 2002–10 radiosonde instrument changes. These shift points actually correspond to the maximum and mini-mum points for relevant events in the 5-yr moving trends as seen in Fig. 8, except that the 5-yr moving trends here should be for the averaged time series over China (not shown). After these shift points are determined, the regional mean ADJ is adjusted to match with MSUav at and after the shift points by the difference between two segments of 36-month in length on either side of the shift point from ADJ and MSUav. The re-adjusted ADJ time series, abbreviated as ReADJ, is also shown in Fig. 6. Comparison statistics (Table 1) show that ReADJ agrees much better with the MSU time series and its trends are discussed in Section 5.

Note that breakpoints during 1980–91, which are attributed to satellite launches in the MSU merging (Table 2 and Fig. 8), are not adjusted in this study. This is because these breakpoints caused the differences between MSU and radiosonde to jump up and down at different times (Figs. 6, 7). To the first order approximation, they did not appear to cause major trend differences between the two observational types because biases due to these up-and-down jumps cancelled out in the long-term trend estimates. This is different from radiosonde-related breakpoints, which cause stepwise changes in the time series of differences between satellite and radiosonde data. Such stepwise changes are the main reason causing the long-term trends from radiosonde data to be cooler than the satellite for TMT, TUT, and TLS and thus needing to be adjusted. The satellite adjustment will be conducted in future investigations when radiosonde is adjusted at individual stations.

By the inter-comparison between radiosonde equivalence and MSU layer temperatures, the significant difference in the 2000s over China has been found, which is related with residual discontinuities in ADJ even after homogenizations. For temperatures at neighboring stations are commonly used as the reference in homogenizations, we selected Hongkong (45004, 22.3°N, 114.2°E) as the reference of Qingyuan (59280, 23.7°N, 113.1°E) (Fig. 1) to check the discontinuity in the 2000s and MSU applicability as the reference. For station 45004 is not contained in Chinese radiosonde network, nationwide radiosonde system changes over China around 2000 have no impact on the temperature time series at station 45004. The layer mean temperatures from three MSU datasets and radiosonde equivalence from ADJ at station 59280 have been compared with the common reference respect to radiosonde equivalence at station 45004. TUT anomalies from MSU at station 59280 and the equivalence at station 45004 have similar variation, and the difference has no significant changes in the 2000s, while the difference between stations 59280 and 45004 shows two significant downward jumps around 2000 and 2010, which are related with the remained discontinuities in ADJ caused by the radiosonde system change in the 2000s (Fig. 9). The comparison demonstrates that the MSU temperature can be used as potential references to the validate homogenization, which is highly valuable for the radiosonde homogenization over China. For the systematic change in Chinese radionde network usually occurs simultaneously, it is hard to find applicative neighbor stations within Chinese network, especially in central China. MSU temperatures with better coverage can provide more useful information on possible breakpoints over China and are helpful to verify or improve the homogenization. Furthermore, the radiosonde equivalence and MSU layer temperatures magnify the difference at individual pressure levels. Temperature differences at individual pressure levels between stations 59280 and 45004 are not so significant as those of the layer mean from MSU and radiosonde equivalence. Therefore, MSU temperatures have better covering and amplifying effects, comparing with the reference at pressure levels from neighbor stations, which contribute to detecting the discontinuity in radiosonde temperature time series. Even so, further research is needed in assigning adjustment from the layer mean to multiple pressure levels.

Figure 9 Monthly difference of TUT anomaly (K) between the radiosonde equivalence at station 45004 from MSU (STAR, RSS, and UAH) and the radiosonde equivalence at station 59280 during 1979–2015.

Figure options

Figure 10 shows linear trends of the monthly layer mean temperature anomalies from radiosonde (RAW, ADJ, ReADJ, and HAD) and MSU (STAR, RSS, UAH, and their average MSUav) datasets during 1979–2015 averaged over China for areas without the surface elevation greater than 2000 m (Cn-notb). As seen, trends from RAW are the coolest for TLS and TUT and least warming for TMT. As found earlier, this occurs because of anomaly jumps from radiosonde system updates and instrument changes in the 2000s. This jumps have large effects on TLS, TUT, and TMT but much smaller effects on TLT. In addition, for all the layers except for TLT, MSU has larger warming tropospheric and smaller cooling stratospheric trends than the homogenized ADJ and HAD radiosonde datasets. Trend differences between MSU and ADJ range from 0.07 K decade^–1 for TLT to 0.12 K decade^–1 for TLS, TUT, and TMT. The reason for this can be seen in Fig. 6 where MSU temperatures are lower during 1980–2000 and higher after 2001 than ADJ and HAD for respective layers. As found earlier, this occurs because adjustments in ADJ and HAD are not large enough to remove discontinuities caused by the radiosonde system update in China in the 2000s.

Figure 10 Linear trends (K decade–1) of monthly layer temperature anomalies from radiosonde (RAW, ADJ, ReADJ, and HAD) and MSU (STAR, RSS, UAH and their average MSUav) datasets during 1979–2015 averaged over China for areas without stations with the elevation above 2000 m. Superimposed lines on trend bars indicate uncertainties at a 95% confidence level.

Figure options

Re-adjustment to radiosonde datasets conducted in this study substantially reduces the trend differences between MSU and homogenized radiosonde temperatures for most layers. In Fig. 10 and Table 3, trend values for ReADJ are 0.203 ± 0.066, 0.128 ± 0.044, 0.034 ± 0.039, and –0.329 ± 0.135 K decade^–1 for TLT, TMT, TUT, and TLS, respectively, and corresponding values for MSUav are 0.149 ± 0.064, 0.134 ± 0.044, 0.040 ± 0.039 and –0.278 ± 0.135 K decade^–1, respectively. Trend differences between satellite and radiosonde data are within 0.01 K decade^–1 for TMT and TUT, suggesting that trend estimates from radiosonde and satellite observations can be greatly reconciled with the adequate application of metadata information for the radiosonde homogenization. In addition, ReADJ trends are 0.054 K decade^–1 warmer for TLT and 0.051 K decade^–1 cooler for TLS than satellite averages, indicating that further adjustments on either satellite data or radiosonde data are still needed.

Figure 11 shows the spatial distribution of trends from the individual MSU, ADJ and HAD. ReADJ is not included since it is done only for regional averages. HAD shows noisier trend patterns for all the layers than the other datasets, caused by the data missing issues as discussed in Section 2. Trend patterns show consistent TLS cooling and TLT warming trends at most stations in China, consistent with the results for regional averages. TLS cooling and TLT warming trends from MSU datasets are generally weaker than those from ADJ and HAD. STAR and RSS TUT show warming over the entire China with the former being more warming over western China. UAH TUT shows cooling over northeastern China and warming over the rest of China. On the other hand, TUT of ADJ shows cooling over the entire China with significant cooling over part of southern China. TMT has warming trends for both MSU and ADJ over the entire China with warming from the former being stronger over the Tibetan Plateau, especially for STAR and RSS data.

Figure 11 Spatial distributions of linear trends of monthly temperature anomalies from MSU datasets (STAR, RSS, and UAH), ADJ，and HAD equivalence over China for 1979–2015. Solid circle indicates the 95% confidence level.

Figure options

In order to study the regional feature of MSU datasets, anomalies and trends during 1979–2015 are compared for averages over the globe and China. Since different MSU datasets exhibit trends with the same signs and their time series are very similar (Figs.11, 6), only average of the three MSU datasets are shown in Fig. 12 and Table 3. In general, China (Cn, averaged over the entire China) is warmer for TLT, TMT, and TUT and cooler for TLS than the global mean. These trend differences are associated with the variability and larger magnitude over China than over the globe in the anomaly time series (Fig. 12).

Figure 12 Monthly temperature anomalies (thin line; K) and their 12-month moving averages (thick line; K) over China (red line) and globe (blue line) during 1979–2015 from the MSU averages of STAR, RSS, and UAH. (a)TLT, (b) TMT, (c) TUT, and (d) TLS.

Figure options

Trends over China are further separated into regions with Cn-sta and without Cn-nosta stations, and without the higher elevation stations (Cn-notb; Table 3). Trends in regions without the stations are stronger than regions with stations and over the entire China, while trends in regions without the higher-elevation stations are weaker than regions with stations and over the entire China. This is supported by the spatial distribution of trends where trends over the Tibetan Plateau and northern China are stronger where fewer stations exist. This difference suggests the importance to use satellite data to estimate trends over the regions where radiosonde data are sparse (especially over the Tibetan Plateau) and to provide a full picture on trend estimate over China.

We have investigated differences in the layer mean atmospheric temperature anomaly time series and long-term trends between MSU (STAR, RSS, and UAH) and radiosonde (RAW, ADJ, and HAD) datasets over China. Correlations between MSU and homogenized radiosonde time series are generally higher than 0.9 for most areas and atmospheric layers at individual stations and for regional averages over China. Exceptions are for TUT over southern China where radiosonde data have quality issues near the tropopause, and for TLT and TMT over the Tibetan Plateau where weighting functions are inadequate to convert radiosonde data to satellite equivalents for stations with the higher surface elevation. Overall, high correlations indicate similar variations between MSU and radiosonde equivalents and thus they could be used as references for each other in the homogenization procedure and for mutual validation. Using radiative transfer models instead of weighting functions for calculating MSU equivalent temperatures could improve correlations over the Tibetan Plateau and will be applied in future studies.

Breakpoints in the time series of differences between MSU and radiosondes were identified during 1980–91, 1999–2001, and 2002–10 for both averages over China and at individual stations. Analyses on the radiosonde metadata and satellite merging indicate that multiple breakpoints during 1980–91 are mainly associated with launch time of different satellites used in the merged MSU time series. On the other hand, differences during the 1999–2001 and 2002–10 periods are caused by radiosonde system updates and instrument changes, respectively, in China. These radiosonde-related breakpoints result in stepwise changes in the time series of differences between satellite and radiosonde observations. Such stepwise changes are the main reasons causing long-term trends from radiosonde data to be cooler than the averaged MSU time series from the three groups for most layer temperatures (TMT, TUT, and TLS) except for TLT.

To reduce the impact of radiosonde system and instrument changes, a shift-point adjusting approach is used to re-adjust radiosonde data of ADJ for the time series averaged over China, excluding the Tibetan Plateau. Adjustment is for the concept demonstration as it does not adjust radiosonde data at individual stations. The layer mean temperature trends during 1979–2015 from the re-adjusted radiosonde data over China were 0.203 ± 0.066, 0.128 ± 0.044, 0.034 ± 0.039, and –0.329 ± 0.135 K decade^–1 for TLT, TMT, TUT, and TLS equivalents. These trends were within 0.01 K decade^–1 for TMT and TUT, and 0.05 K decade^–1 for TLT and TLS of the averaged MSU trends from the three groups. Compared to the range of trend differences between MSU and radiosonde data without this adjustment, the re-adjustment largely improves the consistency of MSU, radiosonde time series, and trends.

For regional characteristics in trend estimates, it is found that long-term temperature trends over China are much stronger than global averages in TLT and TMT. Such differences are also found in surface temperatures (Shi et al., 2016). However, stronger cooling is also found in the lower-stratosphere (TLS) over China than global averages. Attribution studies are needed in the future to understand reasons for these differences.

In conclusion, trends from ReADJ and satellite data over China averages provide encouraging results for the radiosonde homogenization over the globe and at individual stations. As a next step, further adjustment at individual radiosonde stations is to be conducted over China by using the approaches proposed in this study. Such an effort will provide the improved homogenization for radiosonde data for climate change studies. The study also provides useful information on possible breakpoints in the satellite time series. This information will likely help satellite data developers to improve the satellite time series for reliable temperature trend detections.

Acknowledgments. We thank NOAA’s Satellite Applications and Research (STAR), University of Alabama Huntsville (UAH), Remote Sensing System (RSS), Met Office Hadley Centre (MOHC), and National Meteorological Information Center (NMIC) of the China Meteorological Administration (CMA) for providing MSU and radiosonde temperature data. Thanks go to Dr. Chen Zhe for providing homogenized radiosonde temperature data and Mr. Zhang Siqi for programming support. The views, opinions, and findings contained in this report are those of the authors and should not be construed as an official NOAA or U.S. government position, policy, or decision.