Construction of homogenized daily surface air temperature for Tianjin city during 1887&ndash;2019

Si, Peng; Li, Qingxiang; Jones, Phil

doi:https://doi.org/10.5194/essd-2020-343

Preprints

https://doi.org/10.5194/essd-2020-343

Preprints

22 Jan 2021

Review status: a revised version of this preprint was accepted for the journal ESSD and is expected to appear here in due course.

Construction of homogenized daily surface air temperature for Tianjin city during 1887–2019

Peng Si¹, Qingxiang Li^2,4, and Phil Jones³ Peng Si et al.

Show author details

¹Tianjin Meteorological Information Center, Tianjin Meteorological Bureau, Tianjin, China
²School of Atmospheric Sciences, Sun Yat-sen University, Guangzhou, China
³Climatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich, UK
⁴Key Laboratory of Tropical Atmosphere-Ocean System (Sun Yat-sen University), Ministry of Education, and Southern Laboratory of Ocean Science and Engineering (Guangdong Zhuhai), Zhuhai, China

Received: 11 Nov 2020 – Accepted for review: 11 Jan 2021 – Discussion started: 22 Jan 2021

Abstract. The century-long continuous daily observations from some stations are important for the study of long-term trends and extreme climate events in the past. In this paper, three daily data sources: (1) Department of Industry Agency of British Concession in Tianjin covering Sep 1 1890–Dec 31 1931 (2) Water Conservancy Commission of North China covering Jan 1 1932–Dec 31 1950 and (3) monthly journal sheets for Tianjin surface meteorological observation records covering Jan 1 1951–Dec 31 2019 have been collected from the Tianjin Meteorological Archive. The completed daily maximum and minimum temperature series for Tianjin from Jan 1 1887 (Sep 1 1890 for minimum) to Dec 31 2019 has been constructed and assessed for quality control and an early extension from 1890 to 1887. Several significant breakpoints are detected by the Penalized Maximal T-test (PMT) for the daily maximum and minimum time series using multiple reference series around Tianjin from monthly Berkeley Earth, CRUTS4.03 and GHCNV3 data. Using neighboring daily series the record has been homogenized with Quantile Matching (QM) adjustments. Based on the homogenized dataset, the warming trend in annual mean temperature in Tianjin averaged from the newly constructed daily maximum and minimum temperature is evaluated as 0.154 ± 0.013 °C decade-1 during the last 130 years. Trends of temperature extremes in Tianjin are all significant at the 5 % level, and have much more coincident change than those from the raw, with amplitudes of −1.454 d decade⁻¹, 1.196 d decade⁻¹, −0.140 d decade⁻¹ and 0.975 d decade⁻¹ for cold nights (TN10p), warm nights (TN90p), cold days (TX10p) and warm days (TX90p) at the annual scale. The adjusted daily maximum, minimum and mean surface air temperature dataset for Tianjin city presented here is publicly available at https://doi.pangaea.de/10.1594/PANGAEA.924561 (Si and Li, 2020).

Download & links

Preprint (PDF, 1099 KB)

Supplement (507 KB)

Peng Si et al.

Status: closed

RC1:
'Comment on essd-2020-343', Anonymous Referee #1, 31 Jan 2021
The century-long homogenized temperature data is typically lacking and further hampers the robust detection of climate change. Therefore, it is imperative to construct this dataset, laying foundation for better understanding of climate change and variability. To this end, the manuscript by Si et al. confronted this challenge by attempting to construct more than 100-y homogenized temperature dataset at Tianjin, China. This work made significant contribution to the community of climate and climate change. Overall, the homogeneity procedures and test methods are technically robust and scientifically sound. The trends of Tmax, Tmin and Tmean are basically consistent with the well-established trend in midlatitudes, suggesting the findings obtained by the authors are meaningful. Nevertheless, there are still room for improving the quality of this manuscript by adequately addressing following comments:

Major comments:

Figure 6: I noticed that the significant breakpoint occurred in 1955 for Tmax but in 1992 for Tmin. Why did these points differ greatly? is this due to non-climate factor such as station relocation? If so, the influence is supposed to be the same. Explanation or discussion for this interesting phenomenon will benefit the improved understanding of the newly-constructed homogenized dataset.

L393-396: it seems a parodox to argue that “annual trend change in mean temperature based on newly constructed series in Tianjin is similar to that for China (Li et al., 2020c)..”. On the contrary, the trends derived from the other two dataset (Berkeley Earth and CRUTS4.03) are more similar the national warminig rate as shown in Table 5. The authors are suggested to clarify this point.

Minor comments:

L55: “representativeness” -> “better representativeness”

L286: “surface observation station” -> “surface weather station”

L291: “quantile matching” has been widely used in recent research associated with precipitation (Lv et al. 2020, doi:1016/j.atmosres.2019.104671), and PM2.5 (Bai et al. 2020, doi:10.5194/essd-12-3067-2020), which is suggested to be mentioned.

L384: “indicates”-> “shows”

L392: “internationally authoritative data calculations”? are there any references to support this argument? Further, this sentence is not logically connected with the following sentence “so they are more consistent”. For instance, what does the “they” refer to? Therefore, it is suggested to be rewritten.

L404: “indiates”-> “shows” or “illustrates”

L409: grammar errors in ”all passed”

L410: “trends of TN10p and TX10p in spring are the largest. They are” -> “the negative trends of TN10p and TX10p in spring are the largest, reaching up to be”

L422: it seems a little strong tone to argue “These same procedures could and should be use”, which can be softened, since there are large room to improve the procedures for data homogeity. More importantly, it remains unknown whether the procedure developed here can be genalized or applied to other regions, which merits further investigation.
Citation: https://doi.org/10.5194/essd-2020-343-RC1
- CC1:
  'Reply on RC1', Peng Si, 03 Feb 2021
  Thank you very much for the valuable time devoted to this paper by the referees and responsible editor of ESSD, as well as for the opportunity to reply to these comments. A point-by-point response to Anonymous Referee #1' comments is as followed.
  Major comments:
  Figure 6: I noticed that the significant breakpoint occurred in 1955 for Tmax but in 1992 for Tmin. Why did these points differ greatly? Is this due to non-climate factor such as station relocation? If so, the influence is supposed to be the same. Explanation or discussion for this interesting phenomenon will benefit the improved understanding of the newly-constructed homogenized dataset.
  
  Reply：Thank you for your valuable comments.
  It may be related with the physical characteristics of Tmax and Tmin themselves. The Tmin generally occurs near sunrise when calm and stable atmospheric boundary layer conditions are prevalent. Under these conditions, near surface temperature fields are strongly coupled to the local surface characteristics. On the other hand, during daylight hours (like Tmax), the boundary layer is commonly well mixed, and microclimate differences between nearby sites may be less evident. So the difference of the discontinuities between Tmax and Tmin temperature is very normal (Tmin always has more discontinuities in total, see the details in Li and Dong, 2009).
  As to the discontinuities, the breakpoints in Jan 1 1955 and Jan 1 1992 are both caused by station relocation, which have an effect on homogeneity of different temperature elements. In Jan 1 1955 Tianjin meteorological station relocated from No. 22 Ziyou Road to Zunyi Road (Table 1), where is 5km north of the original site. However, the relocation of Tianjin meteorological station in 1992 was from Qixiangtai Road, Hexi District to Xidawa, Xiqing District. In addition, there are another two discontinuities in 1897 and in 1907 for Tmin, and both of these two are not for Tmax.
  We have added this paragraph on L330—L333 in revised paper * - Feb 2, 2021.pdf.
  Ref：
  Li, Q. X., and Dong, W. J.: Detection and adjustment of undocumented discontinuities in Chinese temperature series using a composite approach, Adv., Atmos., Sci.，26(1): 143-153, https://doi.org/ 10.1007/s00376-009-0143-8, 2009.
  2. L393-396: it seems a parodox to argue that “annual trend change in mean temperature based on newly constructed series in Tianjin is similar to that for China (Li et al., 2020c)”. On the contrary, the trends derived from the other two dataset (Berkeley Earth and CRUTS4.03) are more similar the national warminig rate as shown in Table 5. The authors are suggested to clarify this point.
  Reply：Thank you for your valuable comments.
  The annual trend in Tmean from newly constructed series in Tianjin is a little larger than that over the whole China. We think this conforms that the result in this manuscript is reasonable. Because the trend in northern region is more prominent than that of other regions in mainland China (Li et al., 2004; Zhai et al., 2004).
  Therefore, according to the comments, we have rewritten the sentence into ‘Moreover, annual trend change in mean temperature based on newly constructed series in Tianjin is also a little larger than that over the whole China (Li et al., 2020c), which are 0.130±0.009°C decade^-1, 0.114±0.009°C decade^-1 and 0.121±0.009°C decade^-1 respectively from CRUTEM4, GHCNV3 and C-LSAT (during 1900 - 2017).’ on L398—L401 in revised paper * - Feb 2, 2021.pdf.
  And we also added the sentence ‘It conforms that the result is reasonable since the air temperature trends in northern China is more prominent than those in other regions in mainland China (Li et al., 2004; Zhai et al., 2004).’ on L401—L403, and added two references (Li et al., 2004; Zhai et al., 2004) in References section on L527—L529, L618—L619 in revised paper * - Feb 2, 2021.pdf.
  Ref：
  Li., Q. X., Zhang., H. Z., Liu., X. N., and Huang., J. Y.: Urban heat island effect on annual mean temperature during the last 50 years in China, Theor. Appl. Climatol., 79, 165-174, https://doi.org/10.1007/s00704-004-0065-4, 2004.
  Zhai., P. M., Chao., Q. C., and Zou., X. K.: Progress in China’s climate change study in the 20th century, J. Geograph. Sci., 14(1): 3-11, https://doi.org/10.1007/BF02841101, 2004.
  Minor comments:
  L55: “representativeness” -> “better representativeness”
  
  Done.
  2. L286: “surface observation station” -> “surface weather station”
  Done.
  L291: “quantile matching” has been widely used in recent research associated with precipitation (Lv et al. 2020, doi:1016/j.atmosres.2019.104671), and PM2.5 (Bai et al. 2020, doi:10.5194/essd-12-3067-2020), which is suggested to be mentioned.
  
  Done.
  4. L384: “indicates”-> “shows”
  Done.
  5. L392: “internationally authoritative data calculations”? are there any references to support this argument? Further, this sentence is not logically connected with the following sentence “so they are more consistent”. For instance, what does the “they” refer to? Therefore, it is suggested to be rewritten.
  According to the Major Comments 2, we have deleted this sentence ‘The average temperature trend changes from the newly constructed series are much closer to internationally authoritative data calculations, so they are more consistent.’ on L396—L398, and rewritten the second half of the paragraph into ‘Moreover, annual trend change in mean temperature based on newly constructed series in Tianjin is also a little larger than that over the whole China (Li et al., 2020c), which are 0.130±0.009°C decade^-1, 0.114±0.009°C decade^-1 and 0.121±0.009°C decade^-1 respectively from CRUTEM4, GHCNV3 and C-LSAT (during 1900 - 2017). It conforms that the result is reasonable since the air temperature trends in northern China is more prominent than those in other regions in mainland China (Li et al., 2004; Zhai et al., 2004).’ on L398—L403 in revised paper * - Feb 2, 2021.pdf.
  L404: “indiates”-> “shows” or “illustrates”
  
  Done.
  7. L409: grammar errors in ”all passed”
  Done.
  8. L410: “trends of TN10p and TX10p in spring are the largest. They are” -> “the negative trends of TN10p and TX10p in spring are the largest, reaching up to be”
  Done.
  9. L422: it seems a little strong tone to argue “These same procedures could and should be use”, which can be softened, since there are large room to improve the procedures for data homogeity. More importantly, it remains unknown whether the procedure developed here can be genalized or applied to other regions, which merits further investigation.
  The word ‘procedure’ refers to the steps of constructing a long and complete climate time series, rather than the techniques and methods. To avoid the confusion, we have modified this sentence into ‘These similar procedures could and should be used for other sufficiently long and complete series across the world.’ on L429 in revised paper * - Feb 2, 2021.pdf.
  
  Citation: https://doi.org/10.5194/essd-2020-343-CC1
- AC1:
  'Reply on RC1', Qingxiang Li, 12 Mar 2021
  Thank you very much for the valuable time devoted to this paper by the referees and the responsible editor of ESSD, as well as for our opportunity to reply to these comments. A point-by-point response to Anonymous Referee #1' comments is as follows.
  Major comments:
  Figure 6: I noticed that the significant breakpoint occurred in 1955 for Tmax but in 1992 for Tmin. Why did these points differ greatly? Is this due to non-climate factor such as station relocation? If so, the influence is supposed to be the same. Explanation or discussion for this interesting phenomenon will benefit the improved understanding of the newly-constructed homogenized dataset.
  
  Reply：Thank you for your valuable comments.
  It may be related with the physical characteristics of Tmax and Tmin themselves. The Tmin generally occurs near sunrise when calm and stable atmospheric boundary layer conditions are prevalent. Under these conditions, near surface temperature fields are strongly coupled to the local surface characteristics. On the other hand, during daylight hours (like Tmax), the boundary layer is commonly well mixed, and microclimate differences between nearby sites may be less evident. So the difference of the discontinuities between Tmax and Tmin temperature is not that uncommon (e.g. Tmin always has more discontinuities in total, see the details in Li and Dong, 2009).
  As to the discontinuities, the breakpoints in Jan 1 1955 and Jan 1 1992 are both caused by station relocation, which affects the homogeneity of the temperature elements differently. In Jan 1 1955 Tianjin meteorological station relocated from No. 22 Ziyou Road to Zunyi Road (Table 1), which is 5km north of the original site. However, the relocation of Tianjin meteorological station in 1992 was from Qixiangtai Road, Hexi district to Xidawa, Xiqing District. In addition, there are another two discontinuities in 1897 and in 1907 for Tmin (and both of these two did not affect Tmax).
  We have added this paragraph on L330—L333 in revised manuscript ’Si et al -20210202_resub.pdf’.
  Ref：
  Li Q. and Dong W., 2009, Detection and Adjustment of Undocumented Discontinuities in Chinese Temperature Series Using a Composite Approach, Adv., Atmos., Sci.，26（1）：143-153，doi: 10.1007/s00376-009-0143-8
  L393-396: it seems a parodox to argue that “annual trend change in mean temperature based on newly constructed series in Tianjin is similar to that for China (Li et al., 2020c)”. On the contrary, the trends derived from the other two dataset (Berkeley Earth and CRUTS4.03) are more similar the national warminig rate as shown in Table 5. The authors are suggested to clarify this point.
  
  Reply：Thank you for your valuable comments.
  The annual trend in Tmean from the newly constructed series in Tianjin is a little larger than that over the whole of China. We think this conforms that the result in this manuscript is reasonable. Because the trend in northern regions is more prominent than for other regions in mainland China (Li et al., 2004; Zhai et al., 2004). Therefore, according to the comments, the authors have rewritten the sentence into ‘Moreover, annual trend change in mean temperature based on the newly constructed series at Tianjin is also a little larger than that over the whole China (Li et al., 2020c), which are 0.130±0.009°C decade-1, 0.114±0.009°C decade-1 and 0.121±0.009°C decade-1 respectively from CRUTEM4, GHCNV3 and C-LSAT (during 1900 - 2017).’ on L398—L402 in revised manuscript ’Si et al -20210202_resub.pdf’.
  And the authors also added the sentence ‘It conforms to the underlying changes across China. Increasing trends in northern China are more prominent than those from other regions in mainland China (Li et al., 2004; Zhai et al., 2004).’ on L402—L403, and added two references (Li et al., 2004; Zhai et al., 2004) in References section on L527—L529, L618—L619 in revised manuscript ’Si et al -20210202_resub.pdf’.
  Ref：
  Li., Q. X., Zhang., H. Z., Liu., X. N., and Huang., J. Y.: Urban heat island effect on annual mean temperature during the last 50 years in China, Theor. Appl. Climatol., 79, 165-174, https://doi.org/10.1007/s00704-004-0065-4, 2004.
  Zhai., P. M., Chao., Q. C., and Zou., X. K.: Progress in China’s climate change study in the 20th century, J. Geograph. Sci., 14(1): 3-11, https://doi.org/10.1007/BF02841101, 2004.
  Minor comments:
  L55: “representativeness” -> “better representativeness”
  
  Done
  L286: “surface observation station” -> “surface weather station”
  
  Done
  L291: “quantile matching” has been widely used in recent research associated with precipitation (Lv et al. 2020, doi:1016/j.atmosres.2019.104671), and PM2.5 (Bai et al. 2020, doi:10.5194/essd-12-3067-2020), which is suggested to be mentioned.
  
  Done
  L384: “indicates”-> “shows”
  
  Done
  L392: “internationally authoritative data calculations”? are there any references to support this argument? Further, this sentence is not logically connected with the following sentence “so they are more consistent”. For instance, what does the “they” refer to? Therefore, it is suggested to be rewritten.
  
  According to the Major Comments 2, we have deleted this sentence ‘The average temperature trend changes from the newly constructed series are much closer to internationally authoritative data calculations, so they are more consistent.’ on L396—L398, and rewritten the second half of the paragraph into ‘Moreover, annual trend change in mean temperature based on the newly constructed series at Tianjin is also a little larger than that over the whole China (Li et al., 2020c), which are 0.130±0.009°C decade-1, 0.114±0.009°C decade-1 and 0.121±0.009°C decade-1 respectively from CRUTEM4, GHCNV3 and C-LSAT (during 1900 - 2017). It conforms to the underlying changes across China. Increasing trends in northern China are more prominent than those from other regions in mainland China (Li et al., 2004; Zhai et al., 2004).’ on L398—L403 in revised manuscript ’Si et al -20210202_resub.pdf’.
  L404: “indiates”-> “shows” or “illustrates”
  
  Done
  L409: grammar errors in ”all passed”
  
  Done
  L410: “trends of TN10p and TX10p in spring are the largest. They are” -> “the negative trends of TN10p and TX10p in spring are the largest, reaching up to be”
  
  Done
  L422: it seems a little strong tone to argue “These same procedures could and should be use”, which can be softened, since there are large room to improve the procedures for data homogeity. More importantly, it remains unknown whether the procedure developed here can be genalized or applied to other regions, which merits further investigation.
  
  The word ‘procedure’ refers to the steps of constructing a long and complete climate time series, rather than the techniques and methods. To avoid the confusion, we have modified this sentence into ‘These similar procedures could and should be used for other sufficiently long and complete series across the world.’ on L429 in revised manuscript ’Si et al -20210202_resub.pdf’.
  
  Citation: https://doi.org/10.5194/essd-2020-343-AC1
RC2:
'Comment on essd-2020-343', Anonymous Referee #2, 07 Mar 2021

The authors present a methodology and analysis for the homogenization of in situ century-long daily observations. The paper addresses an interesting and original problem, since very few studies have carried out the homogenization of daily extreme temperature in China. However, some detailed analyses are needed to further convince audience.

Specific comments:

Firstly, if the authors are keen to use SAT for the detection of the changepoint: 1. Please provide the information about SAT stations (location, surroundings, rural, towns or city, as well as altitude); 2. Provide a correlation analysis and other measure to convince the use these data; 3. Justify why 300km is reasonable. 4. Please explore why use monthly and daily reference series respectively for the extreme temperature series? The extreme temperature series is monthly or daily series?

Minor comments

L248 For Table 2, why there are four data sources shown here, the Table 1 is puzzle?

L284 “Berkeley-daily data” should be “Berkeley Earth-monthly data”

Citation: https://doi.org/10.5194/essd-2020-343-RC2
- CC2:
  'Reply on RC2', Peng Si, 11 Mar 2021
  Thank you very much for the valuable time devoted to this paper by the referees and responsible editor of ESSD, as well as for our opportunity to reply to these comments. A point-by-point response to Anonymous Referee #2' comments is as follows.
  Specific comments:
  Please provide the information about SAT stations (location, surroundings, rural, towns or city, as well as altitude).
  
  Reply：Thank you for your suggestions.
  We have provided the information about SAT stations in the new Table 3, and added some description sentence on L279 in revised manuscript.
  Provide a correlation analysis and other measure to convince the use these data.
  
  Reply：Thank you for your suggestions.
  It is difficult to find much observational data surrounding Tianjin before the 1950s, so in the establishing the reference series, we employed the station series or the interpolated temperature series using neighboring grid boxes from three global land surface temperature observation series (CRUTS4.03, GHCNV3 and Berkeley Earth). For monthly reference series, we established two types using a weighted average method. One was based on the combination of the interpolated temperature series from Berkeley Earth and CRUTS4.03 and station series from GHCNV3 data for Tianjin site and the other was based on the interpolated temperature series (they are the selected 9 stations) from Berkeley Earth data only. It is worth mentioning that the reference series established by the 3 datasets are very consistent, which reflects the consistency of air temperature variation in this region among different datasets. All the interpolated station series used in this current study have much higher correlations with each other, with the correlation coefficients all higher than 0.95.
  Justify why 300km is reasonable.
  
  Reply：Thank you for your comments.
  In the process of data homogenization, the standard practice is to select 3-5 surrounding stations as the reference stations (the correlation coefficient between these stations and candidate stations is greater than 0.8, and the distance between them is 300-500km) for discontinuous point detection and adjustment, establish a relatively homogenous reference series that can represent the local climate change characteristics for the candidate stations, and detect the homogeneity of the difference series between the candidate and the reference series. However, this ideal depends on the density of the network. In the current study, we found 9 stations (including 54527 itself) within 300 km as the basis of reference series construction, which shows that the reference series establishment in this area is relatively reasonable, and using as many as possible stations can offset some inhomogeneity of a station that might affect the reasonableness of the reference series as far as possible.
  Please explore why use monthly and daily reference series respectively for the extreme temperature series? The extreme temperature series is monthly or daily series?
  
  Reply：Thank you for your comments.
  Because daily temperatures vary on relatively small spatial scales and are influenced by local processes that are complex and nonlinear, homogenization detection for daily data is difficult. The monthly reference series were used to detect breakpoints for the extreme temperature series in the process of PMT, and the daily reference series was used to adjust breakpoints in the process of QM-adjustment. The extreme temperature series are daily series. What we have done in this current study is the same as in Xu et al (2013).
  Ref:
  Xu, W. Q., Li, Q. X., Wang, X. L., Yang, S., Cao, L. J., and Feng, Y.: Homogenization of Chinese daily surface air temperatures and analysis of trends in the extreme temperature indices, J. Geophys. Res. Atmos., 118, http:// doi.org/10.1002/jgrd.50791, 2013.
  Minor comments:
  L248 For Table 2, why there are four data sources shown here, the Table 1 is puzzle?
  
  There are three data sources in Table 2, which are CRUTS4.03, Berkeley Earth and GHCNV3, respectively.
  The CRUTS4.03 and GHCNV3 data meet the requirements of this current study and both are on monthly timescales, but the Berkeley Earth data are on monthly and daily timescales. Table 1 only introduced the metadata information of Tianjin meteorological observation station.
  We made some revisions to show this clearer in revised manuscript.
  L284 “Berkeley-daily data” should be “Berkeley Earth-monthly data”.
  
  It should be “BE-daily data”. We have changed on L286 in revised manuscript.
  
  Citation: https://doi.org/10.5194/essd-2020-343-CC2
- CC3:
  'Reply on RC2', Peng Si, 11 Mar 2021
  Thank you very much for the valuable time devoted to this paper by the referees and responsible editor of ESSD, as well as for our opportunity to reply to these comments. A point-by-point response to Anonymous Referee #2' comments is as follows.
  Specific comments:
  Please provide the information about SAT stations (location, surroundings, rural, towns or city, as well as altitude).
  
  Reply：Thank you for your suggestions.
  We have provided the information about SAT stations in the new Table 3, and added some description sentence on L279 in revised manuscript.
  Provide a correlation analysis and other measure to convince the use these data.
  
  Reply：Thank you for your suggestions.
  It is difficult to find much observational data surrounding Tianjin before the 1950s, so in the establishing the reference series, we employed the station series or the interpolated temperature series using neighboring grid boxes from three global land surface temperature observation series (CRUTS4.03, GHCNV3 and Berkeley Earth). For monthly reference series, we established two types using a weighted average method. One was based on the combination of the interpolated temperature series from Berkeley Earth and CRUTS4.03 and station series from GHCNV3 data for Tianjin site and the other was based on the interpolated temperature series (they are the selected 9 stations) from Berkeley Earth data only. It is worth mentioning that the reference series established by the 3 datasets are very consistent, which reflects the consistency of air temperature variation in this region among different datasets. All the interpolated station series used in this current study have much higher correlations with each other, with the correlation coefficients all higher than 0.95.
  Justify why 300km is reasonable.
  
  Reply：Thank you for your comments.
  In the process of data homogenization, the standard practice is to select 3-5 surrounding stations as the reference stations (the correlation coefficient between these stations and candidate stations is greater than 0.8, and the distance between them is 300-500km) for discontinuous point detection and adjustment, establish a relatively homogenous reference series that can represent the local climate change characteristics for the candidate stations, and detect the homogeneity of the difference series between the candidate and the reference series. However, this ideal depends on the density of the network. In the current study, we found 9 stations (including 54527 itself) within 300 km as the basis of reference series construction, which shows that the reference series establishment in this area is relatively reasonable, and using as many as possible stations can offset some inhomogeneity of a station that might affect the reasonableness of the reference series as far as possible.
  Please explore why use monthly and daily reference series respectively for the extreme temperature series? The extreme temperature series is monthly or daily series?
  
  Reply：Thank you for your comments.
  Because daily temperatures vary on relatively small spatial scales and are influenced by local processes that are complex and nonlinear, homogenization detection for daily data is difficult. The monthly reference series were used to detect breakpoints for the extreme temperature series in the process of PMT, and the daily reference series was used to adjust breakpoints in the process of QM-adjustment. The extreme temperature series are daily series. What we have done in this current study is the same as in Xu et al (2013).
  Ref:
  Xu, W. Q., Li, Q. X., Wang, X. L., Yang, S., Cao, L. J., and Feng, Y.: Homogenization of Chinese daily surface air temperatures and analysis of trends in the extreme temperature indices, J. Geophys. Res. Atmos., 118, http:// doi.org/10.1002/jgrd.50791, 2013.
  Minor comments:
  L248 For Table 2, why there are four data sources shown here, the Table 1 is puzzle?
  
  There are three data sources in Table 2, which are CRUTS4.03, Berkeley Earth and GHCNV3, respectively.
  The CRUTS4.03 and GHCNV3 data meet the requirements of this current study and both are on monthly timescales, but the Berkeley Earth data are on monthly and daily timescales. Table 1 only introduced the metadata information of Tianjin meteorological observation station.
  We made some revisions to show this clearer in revised manuscript.
  L284 “Berkeley-daily data” should be “Berkeley Earth-monthly data”.
  
  It should be “BE-daily data”. We have changed on L286 in revised manuscript.
  
  Citation: https://doi.org/10.5194/essd-2020-343-CC3
- AC2:
  'Reply on RC2', Qingxiang Li, 12 Mar 2021
  Thank you very much for the valuable time devoted to this paper by the referees and responsible editor of ESSD, as well as for our opportunity to reply to these comments. A point-by-point response to Anonymous Referee #2' comments is as follows.
  Specific comments:
  Please provide the information about SAT stations (location, surroundings, rural, towns or city, as well as altitude).
  
  Reply：Thank you for your suggestions.
  We have provided the information about SAT stations in the new Table 3, and added some description sentence on L279 in revised manuscript ‘Si et al -20210311_resub.pdf’.
  Provide a correlation analysis and other measure to convince the use these data.
  
  Reply：Thank you for your suggestions.
  It is difficult to find much observational data surrounding Tianjin before the 1950s, so in the establishing the reference series, we employed the station series or the interpolated temperature series using neighboring grid boxes from three global land surface temperature observation series (CRUTS4.03, GHCNV3 and Berkeley Earth). For monthly reference series, we established two types using a weighted average method. One was based on the combination of the interpolated temperature series from Berkeley Earth and CRUTS4.03 and station series from GHCNV3 data for Tianjin site and the other was based on the interpolated temperature series (they are the selected 9 stations) from Berkeley Earth data only. It is worth mentioning that the reference series established by the 3 datasets are very consistent, which reflects the consistency of air temperature variation in this region among different datasets. All the interpolated station series used in this current study have much higher correlations with each other, with the correlation coefficients all higher than 0.95.
  Justify why 300km is reasonable.
  
  Reply：Thank you for your comments.
  In the process of data homogenization, the standard practice is to select 3-5 surrounding stations as the reference stations (the correlation coefficient between these stations and candidate stations is greater than 0.8, and the distance between them is 300-500km) for discontinuous point detection and adjustment, establish a relatively homogenous reference series that can represent the local climate change characteristics for the candidate stations, and detect the homogeneity of the difference series between the candidate and the reference series. However, this ideal depends on the density of the network. In the current study, we found 9 stations (including 54527 itself) within 300 km as the basis of reference series construction, which shows that the reference series establishment in this area is relatively reasonable, and using as many as possible stations can offset some inhomogeneity of a station that might affect the reasonableness of the reference series as far as possible.
  Please explore why use monthly and daily reference series respectively for the extreme temperature series? The extreme temperature series is monthly or daily series?
  
  Reply：Thank you for your comments.
  Because daily temperatures vary on relatively small spatial scales and are influenced by local processes that are complex and nonlinear, homogenization detection for daily data is difficult. The monthly reference series were used to detect breakpoints for the extreme temperature series in the process of PMT, and the daily reference series was used to adjust breakpoints in the process of QM-adjustment. The extreme temperature series are daily series. What we have done in this current study is the same as in Xu et al (2013).
  Ref:
  Xu, W. Q., Li, Q. X., Wang, X. L., Yang, S., Cao, L. J., and Feng, Y.: Homogenization of Chinese daily surface air temperatures and analysis of trends in the extreme temperature indices, J. Geophys. Res. Atmos., 118, http:// doi.org/10.1002/jgrd.50791, 2013.
  Minor comments:
  L248 For Table 2, why there are four data sources shown here, the Table 1 is puzzle?
  
  There are three data sources in Table 2, which are CRUTS4.03, Berkeley Earth and GHCNV3, respectively.
  The CRUTS4.03 and GHCNV3 data meet the requirements of this current study and both are on monthly timescales, but the Berkeley Earth data are on monthly and daily timescales. Table 1 only introduced the metadata information of Tianjin meteorological observation station.
  We made some revisions to show this clearer in revised manuscript ‘Si et al -20210311_resub.pdf’.
  L284 “Berkeley-daily data” should be “Berkeley Earth-monthly data”.
  
  It should be “BE-daily data”. We have changed on L286 in revised manuscript ‘Si et al -20210311_resub.pdf’.
  
  Citation: https://doi.org/10.5194/essd-2020-343-AC2
- AC3: 'Reply on RC2', Qingxiang Li, 12 Mar 2021
  
  The comment was uploaded in the form of a supplement: https://essd.copernicus.org/preprints/essd-2020-343/essd-2020-343-AC3-supplement.pdf
  
  Citation: https://doi.org/10.5194/essd-2020-343-AC3

Status: closed

RC1:
'Comment on essd-2020-343', Anonymous Referee #1, 31 Jan 2021
The century-long homogenized temperature data is typically lacking and further hampers the robust detection of climate change. Therefore, it is imperative to construct this dataset, laying foundation for better understanding of climate change and variability. To this end, the manuscript by Si et al. confronted this challenge by attempting to construct more than 100-y homogenized temperature dataset at Tianjin, China. This work made significant contribution to the community of climate and climate change. Overall, the homogeneity procedures and test methods are technically robust and scientifically sound. The trends of Tmax, Tmin and Tmean are basically consistent with the well-established trend in midlatitudes, suggesting the findings obtained by the authors are meaningful. Nevertheless, there are still room for improving the quality of this manuscript by adequately addressing following comments:

Major comments:

Figure 6: I noticed that the significant breakpoint occurred in 1955 for Tmax but in 1992 for Tmin. Why did these points differ greatly? is this due to non-climate factor such as station relocation? If so, the influence is supposed to be the same. Explanation or discussion for this interesting phenomenon will benefit the improved understanding of the newly-constructed homogenized dataset.

L393-396: it seems a parodox to argue that “annual trend change in mean temperature based on newly constructed series in Tianjin is similar to that for China (Li et al., 2020c)..”. On the contrary, the trends derived from the other two dataset (Berkeley Earth and CRUTS4.03) are more similar the national warminig rate as shown in Table 5. The authors are suggested to clarify this point.

Minor comments:

L55: “representativeness” -> “better representativeness”

L286: “surface observation station” -> “surface weather station”

L291: “quantile matching” has been widely used in recent research associated with precipitation (Lv et al. 2020, doi:1016/j.atmosres.2019.104671), and PM2.5 (Bai et al. 2020, doi:10.5194/essd-12-3067-2020), which is suggested to be mentioned.

L384: “indicates”-> “shows”

L392: “internationally authoritative data calculations”? are there any references to support this argument? Further, this sentence is not logically connected with the following sentence “so they are more consistent”. For instance, what does the “they” refer to? Therefore, it is suggested to be rewritten.

L404: “indiates”-> “shows” or “illustrates”

L409: grammar errors in ”all passed”

L410: “trends of TN10p and TX10p in spring are the largest. They are” -> “the negative trends of TN10p and TX10p in spring are the largest, reaching up to be”

L422: it seems a little strong tone to argue “These same procedures could and should be use”, which can be softened, since there are large room to improve the procedures for data homogeity. More importantly, it remains unknown whether the procedure developed here can be genalized or applied to other regions, which merits further investigation.
Citation: https://doi.org/10.5194/essd-2020-343-RC1
- CC1:
  'Reply on RC1', Peng Si, 03 Feb 2021
  Thank you very much for the valuable time devoted to this paper by the referees and responsible editor of ESSD, as well as for the opportunity to reply to these comments. A point-by-point response to Anonymous Referee #1' comments is as followed.
  Major comments:
  Figure 6: I noticed that the significant breakpoint occurred in 1955 for Tmax but in 1992 for Tmin. Why did these points differ greatly? Is this due to non-climate factor such as station relocation? If so, the influence is supposed to be the same. Explanation or discussion for this interesting phenomenon will benefit the improved understanding of the newly-constructed homogenized dataset.
  
  Reply：Thank you for your valuable comments.
  It may be related with the physical characteristics of Tmax and Tmin themselves. The Tmin generally occurs near sunrise when calm and stable atmospheric boundary layer conditions are prevalent. Under these conditions, near surface temperature fields are strongly coupled to the local surface characteristics. On the other hand, during daylight hours (like Tmax), the boundary layer is commonly well mixed, and microclimate differences between nearby sites may be less evident. So the difference of the discontinuities between Tmax and Tmin temperature is very normal (Tmin always has more discontinuities in total, see the details in Li and Dong, 2009).
  As to the discontinuities, the breakpoints in Jan 1 1955 and Jan 1 1992 are both caused by station relocation, which have an effect on homogeneity of different temperature elements. In Jan 1 1955 Tianjin meteorological station relocated from No. 22 Ziyou Road to Zunyi Road (Table 1), where is 5km north of the original site. However, the relocation of Tianjin meteorological station in 1992 was from Qixiangtai Road, Hexi District to Xidawa, Xiqing District. In addition, there are another two discontinuities in 1897 and in 1907 for Tmin, and both of these two are not for Tmax.
  We have added this paragraph on L330—L333 in revised paper * - Feb 2, 2021.pdf.
  Ref：
  Li, Q. X., and Dong, W. J.: Detection and adjustment of undocumented discontinuities in Chinese temperature series using a composite approach, Adv., Atmos., Sci.，26(1): 143-153, https://doi.org/ 10.1007/s00376-009-0143-8, 2009.
  2. L393-396: it seems a parodox to argue that “annual trend change in mean temperature based on newly constructed series in Tianjin is similar to that for China (Li et al., 2020c)”. On the contrary, the trends derived from the other two dataset (Berkeley Earth and CRUTS4.03) are more similar the national warminig rate as shown in Table 5. The authors are suggested to clarify this point.
  Reply：Thank you for your valuable comments.
  The annual trend in Tmean from newly constructed series in Tianjin is a little larger than that over the whole China. We think this conforms that the result in this manuscript is reasonable. Because the trend in northern region is more prominent than that of other regions in mainland China (Li et al., 2004; Zhai et al., 2004).
  Therefore, according to the comments, we have rewritten the sentence into ‘Moreover, annual trend change in mean temperature based on newly constructed series in Tianjin is also a little larger than that over the whole China (Li et al., 2020c), which are 0.130±0.009°C decade^-1, 0.114±0.009°C decade^-1 and 0.121±0.009°C decade^-1 respectively from CRUTEM4, GHCNV3 and C-LSAT (during 1900 - 2017).’ on L398—L401 in revised paper * - Feb 2, 2021.pdf.
  And we also added the sentence ‘It conforms that the result is reasonable since the air temperature trends in northern China is more prominent than those in other regions in mainland China (Li et al., 2004; Zhai et al., 2004).’ on L401—L403, and added two references (Li et al., 2004; Zhai et al., 2004) in References section on L527—L529, L618—L619 in revised paper * - Feb 2, 2021.pdf.
  Ref：
  Li., Q. X., Zhang., H. Z., Liu., X. N., and Huang., J. Y.: Urban heat island effect on annual mean temperature during the last 50 years in China, Theor. Appl. Climatol., 79, 165-174, https://doi.org/10.1007/s00704-004-0065-4, 2004.
  Zhai., P. M., Chao., Q. C., and Zou., X. K.: Progress in China’s climate change study in the 20th century, J. Geograph. Sci., 14(1): 3-11, https://doi.org/10.1007/BF02841101, 2004.
  Minor comments:
  L55: “representativeness” -> “better representativeness”
  
  Done.
  2. L286: “surface observation station” -> “surface weather station”
  Done.
  L291: “quantile matching” has been widely used in recent research associated with precipitation (Lv et al. 2020, doi:1016/j.atmosres.2019.104671), and PM2.5 (Bai et al. 2020, doi:10.5194/essd-12-3067-2020), which is suggested to be mentioned.
  
  Done.
  4. L384: “indicates”-> “shows”
  Done.
  5. L392: “internationally authoritative data calculations”? are there any references to support this argument? Further, this sentence is not logically connected with the following sentence “so they are more consistent”. For instance, what does the “they” refer to? Therefore, it is suggested to be rewritten.
  According to the Major Comments 2, we have deleted this sentence ‘The average temperature trend changes from the newly constructed series are much closer to internationally authoritative data calculations, so they are more consistent.’ on L396—L398, and rewritten the second half of the paragraph into ‘Moreover, annual trend change in mean temperature based on newly constructed series in Tianjin is also a little larger than that over the whole China (Li et al., 2020c), which are 0.130±0.009°C decade^-1, 0.114±0.009°C decade^-1 and 0.121±0.009°C decade^-1 respectively from CRUTEM4, GHCNV3 and C-LSAT (during 1900 - 2017). It conforms that the result is reasonable since the air temperature trends in northern China is more prominent than those in other regions in mainland China (Li et al., 2004; Zhai et al., 2004).’ on L398—L403 in revised paper * - Feb 2, 2021.pdf.
  L404: “indiates”-> “shows” or “illustrates”
  
  Done.
  7. L409: grammar errors in ”all passed”
  Done.
  8. L410: “trends of TN10p and TX10p in spring are the largest. They are” -> “the negative trends of TN10p and TX10p in spring are the largest, reaching up to be”
  Done.
  9. L422: it seems a little strong tone to argue “These same procedures could and should be use”, which can be softened, since there are large room to improve the procedures for data homogeity. More importantly, it remains unknown whether the procedure developed here can be genalized or applied to other regions, which merits further investigation.
  The word ‘procedure’ refers to the steps of constructing a long and complete climate time series, rather than the techniques and methods. To avoid the confusion, we have modified this sentence into ‘These similar procedures could and should be used for other sufficiently long and complete series across the world.’ on L429 in revised paper * - Feb 2, 2021.pdf.
  
  Citation: https://doi.org/10.5194/essd-2020-343-CC1
- AC1:
  'Reply on RC1', Qingxiang Li, 12 Mar 2021
  Thank you very much for the valuable time devoted to this paper by the referees and the responsible editor of ESSD, as well as for our opportunity to reply to these comments. A point-by-point response to Anonymous Referee #1' comments is as follows.
  Major comments:
  Figure 6: I noticed that the significant breakpoint occurred in 1955 for Tmax but in 1992 for Tmin. Why did these points differ greatly? Is this due to non-climate factor such as station relocation? If so, the influence is supposed to be the same. Explanation or discussion for this interesting phenomenon will benefit the improved understanding of the newly-constructed homogenized dataset.
  
  Reply：Thank you for your valuable comments.
  It may be related with the physical characteristics of Tmax and Tmin themselves. The Tmin generally occurs near sunrise when calm and stable atmospheric boundary layer conditions are prevalent. Under these conditions, near surface temperature fields are strongly coupled to the local surface characteristics. On the other hand, during daylight hours (like Tmax), the boundary layer is commonly well mixed, and microclimate differences between nearby sites may be less evident. So the difference of the discontinuities between Tmax and Tmin temperature is not that uncommon (e.g. Tmin always has more discontinuities in total, see the details in Li and Dong, 2009).
  As to the discontinuities, the breakpoints in Jan 1 1955 and Jan 1 1992 are both caused by station relocation, which affects the homogeneity of the temperature elements differently. In Jan 1 1955 Tianjin meteorological station relocated from No. 22 Ziyou Road to Zunyi Road (Table 1), which is 5km north of the original site. However, the relocation of Tianjin meteorological station in 1992 was from Qixiangtai Road, Hexi district to Xidawa, Xiqing District. In addition, there are another two discontinuities in 1897 and in 1907 for Tmin (and both of these two did not affect Tmax).
  We have added this paragraph on L330—L333 in revised manuscript ’Si et al -20210202_resub.pdf’.
  Ref：
  Li Q. and Dong W., 2009, Detection and Adjustment of Undocumented Discontinuities in Chinese Temperature Series Using a Composite Approach, Adv., Atmos., Sci.，26（1）：143-153，doi: 10.1007/s00376-009-0143-8
  L393-396: it seems a parodox to argue that “annual trend change in mean temperature based on newly constructed series in Tianjin is similar to that for China (Li et al., 2020c)”. On the contrary, the trends derived from the other two dataset (Berkeley Earth and CRUTS4.03) are more similar the national warminig rate as shown in Table 5. The authors are suggested to clarify this point.
  
  Reply：Thank you for your valuable comments.
  The annual trend in Tmean from the newly constructed series in Tianjin is a little larger than that over the whole of China. We think this conforms that the result in this manuscript is reasonable. Because the trend in northern regions is more prominent than for other regions in mainland China (Li et al., 2004; Zhai et al., 2004). Therefore, according to the comments, the authors have rewritten the sentence into ‘Moreover, annual trend change in mean temperature based on the newly constructed series at Tianjin is also a little larger than that over the whole China (Li et al., 2020c), which are 0.130±0.009°C decade-1, 0.114±0.009°C decade-1 and 0.121±0.009°C decade-1 respectively from CRUTEM4, GHCNV3 and C-LSAT (during 1900 - 2017).’ on L398—L402 in revised manuscript ’Si et al -20210202_resub.pdf’.
  And the authors also added the sentence ‘It conforms to the underlying changes across China. Increasing trends in northern China are more prominent than those from other regions in mainland China (Li et al., 2004; Zhai et al., 2004).’ on L402—L403, and added two references (Li et al., 2004; Zhai et al., 2004) in References section on L527—L529, L618—L619 in revised manuscript ’Si et al -20210202_resub.pdf’.
  Ref：
  Li., Q. X., Zhang., H. Z., Liu., X. N., and Huang., J. Y.: Urban heat island effect on annual mean temperature during the last 50 years in China, Theor. Appl. Climatol., 79, 165-174, https://doi.org/10.1007/s00704-004-0065-4, 2004.
  Zhai., P. M., Chao., Q. C., and Zou., X. K.: Progress in China’s climate change study in the 20th century, J. Geograph. Sci., 14(1): 3-11, https://doi.org/10.1007/BF02841101, 2004.
  Minor comments:
  L55: “representativeness” -> “better representativeness”
  
  Done
  L286: “surface observation station” -> “surface weather station”
  
  Done
  L291: “quantile matching” has been widely used in recent research associated with precipitation (Lv et al. 2020, doi:1016/j.atmosres.2019.104671), and PM2.5 (Bai et al. 2020, doi:10.5194/essd-12-3067-2020), which is suggested to be mentioned.
  
  Done
  L384: “indicates”-> “shows”
  
  Done
  L392: “internationally authoritative data calculations”? are there any references to support this argument? Further, this sentence is not logically connected with the following sentence “so they are more consistent”. For instance, what does the “they” refer to? Therefore, it is suggested to be rewritten.
  
  According to the Major Comments 2, we have deleted this sentence ‘The average temperature trend changes from the newly constructed series are much closer to internationally authoritative data calculations, so they are more consistent.’ on L396—L398, and rewritten the second half of the paragraph into ‘Moreover, annual trend change in mean temperature based on the newly constructed series at Tianjin is also a little larger than that over the whole China (Li et al., 2020c), which are 0.130±0.009°C decade-1, 0.114±0.009°C decade-1 and 0.121±0.009°C decade-1 respectively from CRUTEM4, GHCNV3 and C-LSAT (during 1900 - 2017). It conforms to the underlying changes across China. Increasing trends in northern China are more prominent than those from other regions in mainland China (Li et al., 2004; Zhai et al., 2004).’ on L398—L403 in revised manuscript ’Si et al -20210202_resub.pdf’.
  L404: “indiates”-> “shows” or “illustrates”
  
  Done
  L409: grammar errors in ”all passed”
  
  Done
  L410: “trends of TN10p and TX10p in spring are the largest. They are” -> “the negative trends of TN10p and TX10p in spring are the largest, reaching up to be”
  
  Done
  L422: it seems a little strong tone to argue “These same procedures could and should be use”, which can be softened, since there are large room to improve the procedures for data homogeity. More importantly, it remains unknown whether the procedure developed here can be genalized or applied to other regions, which merits further investigation.
  
  The word ‘procedure’ refers to the steps of constructing a long and complete climate time series, rather than the techniques and methods. To avoid the confusion, we have modified this sentence into ‘These similar procedures could and should be used for other sufficiently long and complete series across the world.’ on L429 in revised manuscript ’Si et al -20210202_resub.pdf’.
  
  Citation: https://doi.org/10.5194/essd-2020-343-AC1
RC2:
'Comment on essd-2020-343', Anonymous Referee #2, 07 Mar 2021

The authors present a methodology and analysis for the homogenization of in situ century-long daily observations. The paper addresses an interesting and original problem, since very few studies have carried out the homogenization of daily extreme temperature in China. However, some detailed analyses are needed to further convince audience.

Specific comments:

Firstly, if the authors are keen to use SAT for the detection of the changepoint: 1. Please provide the information about SAT stations (location, surroundings, rural, towns or city, as well as altitude); 2. Provide a correlation analysis and other measure to convince the use these data; 3. Justify why 300km is reasonable. 4. Please explore why use monthly and daily reference series respectively for the extreme temperature series? The extreme temperature series is monthly or daily series?

Minor comments

L248 For Table 2, why there are four data sources shown here, the Table 1 is puzzle?

L284 “Berkeley-daily data” should be “Berkeley Earth-monthly data”

Citation: https://doi.org/10.5194/essd-2020-343-RC2
- CC2:
  'Reply on RC2', Peng Si, 11 Mar 2021
  Thank you very much for the valuable time devoted to this paper by the referees and responsible editor of ESSD, as well as for our opportunity to reply to these comments. A point-by-point response to Anonymous Referee #2' comments is as follows.
  Specific comments:
  Please provide the information about SAT stations (location, surroundings, rural, towns or city, as well as altitude).
  
  Reply：Thank you for your suggestions.
  We have provided the information about SAT stations in the new Table 3, and added some description sentence on L279 in revised manuscript.
  Provide a correlation analysis and other measure to convince the use these data.
  
  Reply：Thank you for your suggestions.
  It is difficult to find much observational data surrounding Tianjin before the 1950s, so in the establishing the reference series, we employed the station series or the interpolated temperature series using neighboring grid boxes from three global land surface temperature observation series (CRUTS4.03, GHCNV3 and Berkeley Earth). For monthly reference series, we established two types using a weighted average method. One was based on the combination of the interpolated temperature series from Berkeley Earth and CRUTS4.03 and station series from GHCNV3 data for Tianjin site and the other was based on the interpolated temperature series (they are the selected 9 stations) from Berkeley Earth data only. It is worth mentioning that the reference series established by the 3 datasets are very consistent, which reflects the consistency of air temperature variation in this region among different datasets. All the interpolated station series used in this current study have much higher correlations with each other, with the correlation coefficients all higher than 0.95.
  Justify why 300km is reasonable.
  
  Reply：Thank you for your comments.
  In the process of data homogenization, the standard practice is to select 3-5 surrounding stations as the reference stations (the correlation coefficient between these stations and candidate stations is greater than 0.8, and the distance between them is 300-500km) for discontinuous point detection and adjustment, establish a relatively homogenous reference series that can represent the local climate change characteristics for the candidate stations, and detect the homogeneity of the difference series between the candidate and the reference series. However, this ideal depends on the density of the network. In the current study, we found 9 stations (including 54527 itself) within 300 km as the basis of reference series construction, which shows that the reference series establishment in this area is relatively reasonable, and using as many as possible stations can offset some inhomogeneity of a station that might affect the reasonableness of the reference series as far as possible.
  Please explore why use monthly and daily reference series respectively for the extreme temperature series? The extreme temperature series is monthly or daily series?
  
  Reply：Thank you for your comments.
  Because daily temperatures vary on relatively small spatial scales and are influenced by local processes that are complex and nonlinear, homogenization detection for daily data is difficult. The monthly reference series were used to detect breakpoints for the extreme temperature series in the process of PMT, and the daily reference series was used to adjust breakpoints in the process of QM-adjustment. The extreme temperature series are daily series. What we have done in this current study is the same as in Xu et al (2013).
  Ref:
  Xu, W. Q., Li, Q. X., Wang, X. L., Yang, S., Cao, L. J., and Feng, Y.: Homogenization of Chinese daily surface air temperatures and analysis of trends in the extreme temperature indices, J. Geophys. Res. Atmos., 118, http:// doi.org/10.1002/jgrd.50791, 2013.
  Minor comments:
  L248 For Table 2, why there are four data sources shown here, the Table 1 is puzzle?
  
  There are three data sources in Table 2, which are CRUTS4.03, Berkeley Earth and GHCNV3, respectively.
  The CRUTS4.03 and GHCNV3 data meet the requirements of this current study and both are on monthly timescales, but the Berkeley Earth data are on monthly and daily timescales. Table 1 only introduced the metadata information of Tianjin meteorological observation station.
  We made some revisions to show this clearer in revised manuscript.
  L284 “Berkeley-daily data” should be “Berkeley Earth-monthly data”.
  
  It should be “BE-daily data”. We have changed on L286 in revised manuscript.
  
  Citation: https://doi.org/10.5194/essd-2020-343-CC2
- CC3:
  'Reply on RC2', Peng Si, 11 Mar 2021
  Thank you very much for the valuable time devoted to this paper by the referees and responsible editor of ESSD, as well as for our opportunity to reply to these comments. A point-by-point response to Anonymous Referee #2' comments is as follows.
  Specific comments:
  Please provide the information about SAT stations (location, surroundings, rural, towns or city, as well as altitude).
  
  Reply：Thank you for your suggestions.
  We have provided the information about SAT stations in the new Table 3, and added some description sentence on L279 in revised manuscript.
  Provide a correlation analysis and other measure to convince the use these data.
  
  Reply：Thank you for your suggestions.
  It is difficult to find much observational data surrounding Tianjin before the 1950s, so in the establishing the reference series, we employed the station series or the interpolated temperature series using neighboring grid boxes from three global land surface temperature observation series (CRUTS4.03, GHCNV3 and Berkeley Earth). For monthly reference series, we established two types using a weighted average method. One was based on the combination of the interpolated temperature series from Berkeley Earth and CRUTS4.03 and station series from GHCNV3 data for Tianjin site and the other was based on the interpolated temperature series (they are the selected 9 stations) from Berkeley Earth data only. It is worth mentioning that the reference series established by the 3 datasets are very consistent, which reflects the consistency of air temperature variation in this region among different datasets. All the interpolated station series used in this current study have much higher correlations with each other, with the correlation coefficients all higher than 0.95.
  Justify why 300km is reasonable.
  
  Reply：Thank you for your comments.
  In the process of data homogenization, the standard practice is to select 3-5 surrounding stations as the reference stations (the correlation coefficient between these stations and candidate stations is greater than 0.8, and the distance between them is 300-500km) for discontinuous point detection and adjustment, establish a relatively homogenous reference series that can represent the local climate change characteristics for the candidate stations, and detect the homogeneity of the difference series between the candidate and the reference series. However, this ideal depends on the density of the network. In the current study, we found 9 stations (including 54527 itself) within 300 km as the basis of reference series construction, which shows that the reference series establishment in this area is relatively reasonable, and using as many as possible stations can offset some inhomogeneity of a station that might affect the reasonableness of the reference series as far as possible.
  Please explore why use monthly and daily reference series respectively for the extreme temperature series? The extreme temperature series is monthly or daily series?
  
  Reply：Thank you for your comments.
  Because daily temperatures vary on relatively small spatial scales and are influenced by local processes that are complex and nonlinear, homogenization detection for daily data is difficult. The monthly reference series were used to detect breakpoints for the extreme temperature series in the process of PMT, and the daily reference series was used to adjust breakpoints in the process of QM-adjustment. The extreme temperature series are daily series. What we have done in this current study is the same as in Xu et al (2013).
  Ref:
  Xu, W. Q., Li, Q. X., Wang, X. L., Yang, S., Cao, L. J., and Feng, Y.: Homogenization of Chinese daily surface air temperatures and analysis of trends in the extreme temperature indices, J. Geophys. Res. Atmos., 118, http:// doi.org/10.1002/jgrd.50791, 2013.
  Minor comments:
  L248 For Table 2, why there are four data sources shown here, the Table 1 is puzzle?
  
  There are three data sources in Table 2, which are CRUTS4.03, Berkeley Earth and GHCNV3, respectively.
  The CRUTS4.03 and GHCNV3 data meet the requirements of this current study and both are on monthly timescales, but the Berkeley Earth data are on monthly and daily timescales. Table 1 only introduced the metadata information of Tianjin meteorological observation station.
  We made some revisions to show this clearer in revised manuscript.
  L284 “Berkeley-daily data” should be “Berkeley Earth-monthly data”.
  
  It should be “BE-daily data”. We have changed on L286 in revised manuscript.
  
  Citation: https://doi.org/10.5194/essd-2020-343-CC3
- AC2:
  'Reply on RC2', Qingxiang Li, 12 Mar 2021
  Thank you very much for the valuable time devoted to this paper by the referees and responsible editor of ESSD, as well as for our opportunity to reply to these comments. A point-by-point response to Anonymous Referee #2' comments is as follows.
  Specific comments:
  Please provide the information about SAT stations (location, surroundings, rural, towns or city, as well as altitude).
  
  Reply：Thank you for your suggestions.
  We have provided the information about SAT stations in the new Table 3, and added some description sentence on L279 in revised manuscript ‘Si et al -20210311_resub.pdf’.
  Provide a correlation analysis and other measure to convince the use these data.
  
  Reply：Thank you for your suggestions.
  It is difficult to find much observational data surrounding Tianjin before the 1950s, so in the establishing the reference series, we employed the station series or the interpolated temperature series using neighboring grid boxes from three global land surface temperature observation series (CRUTS4.03, GHCNV3 and Berkeley Earth). For monthly reference series, we established two types using a weighted average method. One was based on the combination of the interpolated temperature series from Berkeley Earth and CRUTS4.03 and station series from GHCNV3 data for Tianjin site and the other was based on the interpolated temperature series (they are the selected 9 stations) from Berkeley Earth data only. It is worth mentioning that the reference series established by the 3 datasets are very consistent, which reflects the consistency of air temperature variation in this region among different datasets. All the interpolated station series used in this current study have much higher correlations with each other, with the correlation coefficients all higher than 0.95.
  Justify why 300km is reasonable.
  
  Reply：Thank you for your comments.
  In the process of data homogenization, the standard practice is to select 3-5 surrounding stations as the reference stations (the correlation coefficient between these stations and candidate stations is greater than 0.8, and the distance between them is 300-500km) for discontinuous point detection and adjustment, establish a relatively homogenous reference series that can represent the local climate change characteristics for the candidate stations, and detect the homogeneity of the difference series between the candidate and the reference series. However, this ideal depends on the density of the network. In the current study, we found 9 stations (including 54527 itself) within 300 km as the basis of reference series construction, which shows that the reference series establishment in this area is relatively reasonable, and using as many as possible stations can offset some inhomogeneity of a station that might affect the reasonableness of the reference series as far as possible.
  Please explore why use monthly and daily reference series respectively for the extreme temperature series? The extreme temperature series is monthly or daily series?
  
  Reply：Thank you for your comments.
  Because daily temperatures vary on relatively small spatial scales and are influenced by local processes that are complex and nonlinear, homogenization detection for daily data is difficult. The monthly reference series were used to detect breakpoints for the extreme temperature series in the process of PMT, and the daily reference series was used to adjust breakpoints in the process of QM-adjustment. The extreme temperature series are daily series. What we have done in this current study is the same as in Xu et al (2013).
  Ref:
  Xu, W. Q., Li, Q. X., Wang, X. L., Yang, S., Cao, L. J., and Feng, Y.: Homogenization of Chinese daily surface air temperatures and analysis of trends in the extreme temperature indices, J. Geophys. Res. Atmos., 118, http:// doi.org/10.1002/jgrd.50791, 2013.
  Minor comments:
  L248 For Table 2, why there are four data sources shown here, the Table 1 is puzzle?
  
  There are three data sources in Table 2, which are CRUTS4.03, Berkeley Earth and GHCNV3, respectively.
  The CRUTS4.03 and GHCNV3 data meet the requirements of this current study and both are on monthly timescales, but the Berkeley Earth data are on monthly and daily timescales. Table 1 only introduced the metadata information of Tianjin meteorological observation station.
  We made some revisions to show this clearer in revised manuscript ‘Si et al -20210311_resub.pdf’.
  L284 “Berkeley-daily data” should be “Berkeley Earth-monthly data”.
  
  It should be “BE-daily data”. We have changed on L286 in revised manuscript ‘Si et al -20210311_resub.pdf’.
  
  Citation: https://doi.org/10.5194/essd-2020-343-AC2
- AC3: 'Reply on RC2', Qingxiang Li, 12 Mar 2021
  
  The comment was uploaded in the form of a supplement: https://essd.copernicus.org/preprints/essd-2020-343/essd-2020-343-AC3-supplement.pdf
  
  Citation: https://doi.org/10.5194/essd-2020-343-AC3

Peng Si et al.

Supplement

https://doi.org/10.5194/essd-2020-343-supplement

Data sets

Tianjin homogenized daily surface air temperature over century-long scale Peng Si and Qingxiang Li https://doi.pangaea.de/10.1594/PANGAEA.924561

Peng Si et al.

Viewed

Total article views: 578 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
445	113	20	578	28	7	6

HTML: 445
PDF: 113
XML: 20
Total: 578
Supplement: 28
BibTeX: 7
EndNote: 6

Views and downloads (calculated since 22 Jan 2021)

Month	HTML	PDF	XML	Total
Jan 2021	155	35	6	196
Feb 2021	92	23	2	117
Mar 2021	128	32	11	171
Apr 2021	56	15	1	72
May 2021	14	8	0	22

Cumulative views and downloads (calculated since 22 Jan 2021)

Month	HTML	PDF	XML	Total
Jan 2021	155	35	6	196
Feb 2021	92	23	2	117
Mar 2021	128	32	11	171
Apr 2021	56	15	1	72
May 2021	14	8	0	22

Viewed (geographical distribution)

Total article views: 468 (including HTML, PDF, and XML) Thereof 464 with geography defined and 4 with unknown origin.

Country	#	Views	%

Latest update: 16 May 2021

Short summary

This paper documents the various procedures necessary to construct a homogenized daily maximum and minimum temperature series since 1887 for Tianjin. The newly constructed temperature series provides a set of new baseline data for the field of extreme climate change over the century-long scale and a reference for construction of other long-term reliable daily time series in the region.


Total:	0
HTML:	0
PDF:	0
XML:	0