scispace - formally typeset

Posted ContentDOI

A Comprehensive Evaluation of Surface Air Temperature Reanalyses over China against Urbanization Bias–Adjusted Observations

12 Feb 2021-Advances in Climate Change Research (Elsevier BV)-

Abstract: The reanalysis datasets (CRA40) have now been developed in China Meteorological Administration. We aim to compare the differences in surface air temperature (SAT) between observational that has been adjusted for urbanization bias and reanalysis data (NCEPV1, NCEPV2, ERA5, CFSR, MERRA, JRA55, 20CRV3 and CRA40) over mainland China during 1961–2015. The main results are presented as follows. The correlation and standard deviation between the reanalysis data and observations exhibit highly consistent interannual variability and dispersion, with the interannual SAT variability in JRA55 being closest to the observations for 1961–2015 and that of ERA5 for 1979–2015; the dispersions of 20CRV3 is most consistent with the observations for 1961–2015 and that of NCEPV1 for 1979–2015. Although annual mean SAT of the reanalysis is generally 0–2.0 °C lower than the observations, the bias in the SAT climatology of 20CRV3 is the least for 1961–2015 in all reanalysis datasets and that of CRA40 is the least for 1979–2015. The trends of NCEPV1 is closer to the observations than other reanalysis for 1961–2015 and that of 20CRV3 for 1979–2015. The biases in terms of interannual variability, dispersion, climatology, and linear trend are increase with altitude. Overall, in terms of the similarity of multiple measures to the urbanization bias-adjusted observations, JRA55 and CRA40 show the best performances for the periods 1961–2015 and 1979–2015 respectively in reproducing various aspects of climatological and climate change features in mainland China.

Content maybe subject to copyright    Report

A Comprehensive Evaluation of Surface Air
Temperature Reanalyses over China against
Urbanization Bias–Adjusted Observations
Siqi Zhang ( 444902334@qq.com )
National Climate Center
Guoyu Ren
China University of Geosciences (CUG) https://orcid.org/0000-0002-9351-4179
Yuyu Ren
National Climate Center, China Meteorological Administration (CMA)
Yingxian Zhang
National Climate Center, China Meteorological Administration (CMA)
Xiaoying Xue
China University of Geosciences (CUG),
Original Paper
Keywords: temperature, reanalysis data, urbanization bias, adjusted observations, comparison,
evaluation, China
Posted Date: February 12th, 2021
DOI: https://doi.org/10.21203/rs.3.rs-182529/v1
License: This work is licensed under a Creative Commons Attribution 4.0 International License. 
Read Full License

1 / 36
A Comprehensive Evaluation of Surface Air Temperature 1
Reanalyses over China against Urbanizationbiasadjusted 2
Observations 3
4
Siqi Zhang
1.2.3
Guoyu Ren
1.2
Yuyu Ren
2
Yingxian Zhang
2
Xiaoying Xue
1
5
6
1. Department of Atmospheric Science, School of Environmental Studies, China University of 7
Geosciences (CUG), Wuhan 430074, China 8
2. Laboratory for Climate Studies, National Climate Center, China Meteorological 9
Administration (CMA), China 10
3 Key laboratory for Cloud Physics of China Meteorological Administration, Beijing 100081, 11
China 12
*Corresponding author: Guoyu Ren 13
(Email: guoyoo@cma.gov.cn; Tel: +86 010-68406408) 14
15
Submitted to Theoretical and Applied Climatology for possible publication 16
17

2 / 36
ABSTRACT 18
The goal of this study is to compare the differences in surface air temperature (SAT) between 19
observational and reanalysis data in mainland China from 1961–2015 for evaluating the reliability 20
and applicability of the reanalysis datasets, based on an observational dataset of 763 stations which 21
has been adjusted for urbanization bias, and 8 reanalysis datasets. The time series, anomaly 22
correlations, standard deviations, climate state, and linear trends of the reanalysis data are evaluated 23
against the observations. The reanalysis data are consistent with the observational climate 24
characteristics to a large extent. The correlation and standard deviation ratio between the reanalysis 25
data and observations exhibited highly consistent interannual variability and dispersion, with the 26
interannual SAT variability of JRA55 and ERA5 the closest to the observations for the periods 27
1961–2015 and 1979–2015, and the dispersions of 20CRV3 and NCEPV1 the most consistent with 28
the observations for the two periods. The annual mean SAT of the reanalyses is generally 0–2.0°C 29
lower than the observations, while the linear trends of all datasets exhibited clear warming. The 30
biases in the SAT climatology of 20CRV3 and CRA40 are lower than other reanalysis datasets, and 31
the linear trends of NCEPV1 and 20CRV3 are closer to the observations. With increasing elevation, 32
the biases of the reanalysis data in terms of correlation, standard deviation, climate state, and linear 33
trend all increased. Overall, in terms of the similarity of multiple measures to the urbanization bias–34
adjusted observations, CRA40 and JRA55 show the best performance of the products in 35
reproducing various aspects of climatological and climate change features in mainland China for 36
the period 1979–2015 and 1961–2015 respectively. 37
KEY WORDS: temperature; reanalysis data; urbanization bias; adjusted observations; comparison; 38
evaluation; China 39
40
1. INTRODUCTION 41
Reanalysis datasets have been widely used in climate change and climatological research. Over 42
the past 30 years, the United States, the European Union, and Japan have produced a series of 43
reanalysis datasets. The fourth generation reanalysis datasets have now been developed, which 44
include NCEP–NCAR (Kalnay et al., 1996), NCEP–DOE (Kanamitsu et al., 2002), CFSR (Saha 45
et al., 2010), NASA–MERRA (Rienecker et al., 2011), 20CR (Compo et al., 2011), ERAinterim 46
(Dee et al., 2011), and JRA55 (Kobayashi et al., 2015). 47
The ERA5 and CRA40 were developed by the European Centre for MediumRange Weather 48
Forecasts (ECMWF) (Hersbach et al., 2020) and China Meteorological Administration (CMA) 49
(Liang et al., 2020), respectively, as the new datasets in 2019. In addition, the NOAACIRES50
DOE Twentieth Century Reanalysis (20CR) project has generated an updated fourdimensional 51
global atmospheric dataset spanning the period 1836–2015 to replace the current 20CRV2 and 52
20CRV2C datasets. Since these reanalyses have the disadvantages of incorporating the errors from 53
the numerical prediction models, the assimilation processes, and the observation systems 54
(Bengtsson et al., 2004a, 2004b; Zhao et al., 2010; Zhao et al., 2015), the reliability of the reanalysis 55
data needs to be proven on global and regional scales. 56
Kistler et al. (2001) showed that the surface air temperatures belonging to class B were 57
generally not as reliable as air pressure. Liu et al. (2008) compared NCEPV2 and ERA–40 to buoy 58
observations and found that the reanalysis had warm biases and underestimated the observed inter–59
annual variability of summer surface air temperature from 1979–1999. Lindsay et al. (2014) 60
showed that the third generation models (CFSR, MERRA, and ERA–Interim) stood out as being 61

3 / 36
more consistent with independent observations for period 1981–2010 in the Arctic region. Fan and 62
Liu (2013) indicated the consistency of SAT climatology in the Southern Hemisphere for the data 63
derived from 20CR and observations from 1979–2010. Zhou et al. (2016) evaluated the SATs in 64
reanalysis data (NCEPV1, NCEPV2, JRA55, ERA–interim, and MERRA) over global deserts, 65
revealing annual mean values ranging from 20.8–24.5°C and long–term trends varying from 0.10–66
0.14°C/decade. 67
With the newly–developed reanalysis datasets, many researchers have examined the reliability 68
and applicability of reanalysis surface air temperatures in China (Su et al., 1999; Xu et al., 2001; 69
Zhao et al., 2006; Gao et al., 2010; Li et al., 2012; He et al., 2013; Zhu et al., 2015). The biases in 70
inter–annual variability, time series, and spatial distribution characteristics between reanalysis and 71
observational data have been analyzed. The results revealed that the application of reanalysis air 72
temperature in eastern China was more consistent with observation than the application in western 73
China (Zhao et al., 2006; Zhu et al., 2015). The climate characteristics displayed by the datasets 74
had a higher agreement with observations after 1979 than before 1979, reflecting the fact that the 75
new generation reanalysis data were generally better than the old products. 76
Many researchers have confirmed the above conclusions and found that the bias of SAT 77
between reanalysis and observation was affected by elevation (Ma et al., 2008; Zhao et al., 2006; 78
Zhou et al., 2018). The biases in the temperature trends were derived jointly from those in 79
atmospheric downward long–wave radiation and precipitation frequency. For the linear trend bias 80
of temperature from 1909–2010, Zhang et al. (2019) analyzed the linear trend of temperature in 81
eastern China based on the homogenized data of 16 observation stations and two sets of 20th 82
century reanalysis data (20CR and ERA20C), and found that, although ERA20C was generally 83
closer to the observational temperatures than 20CR from 1909–2010, their consistency did not 84
indicate that ERA20C and 20CRV2 were suitable for climate change research due to the systematic 85
bias related to the observational data from the stations. 86
The aforementioned studies revealed the uncertainties of reanalysis datasets in different regions 87
and different periods. These may be due to limited assimilation sources, errors from the numerical 88
prediction models, and observation system changes, which produced different results for different 89
periods and regions (Xu et al., 2001; Thorne and Vose, 2010; Parker, 2016; Lahoz and Schneider, 90
2014; Dee et al., 2014; Zhou et al., 2017). In addition, the previous comparisons of SAT data 91
between reanalysis and observation generally considered the inhomogeneity of the reference data, 92
but ignored the bias of the linear trend caused by the urbanization that has occurred in recent 93
decades. Thus, the question as to how to evaluate the potential of reanalysis data for climate change 94
research has not been given sufficient attention, since the effects of urbanization have not been 95
considered. 96
There have been many recent studies demonstrating that the SAT observation records of urban 97
stations and national meteorological station networks contain a significant urbanization bias, 98
regardless of whether homogenization corrections were applied, especially since the middle of the 99
20th century in China (Ren et al., 2007, 2008; Hua et al., 2008; Li et al., 2010; Ren and Ren, 2011; 100
Wang and Ge, 2012; He et al., 2013; Ren and Zhou, 2014; Shi et al., 2015; Li et al., 2016). Due to 101
the differences in research regions, time periods, and data processing methods, and, in particular, 102
to the differences in the reference stations networks that were applied, estimating of the 103
urbanization effect and contribution varied among the different research groups. It was therefore 104
necessary to develop an actual rural (reference) national station network for assessing and adjusting 105

4 / 36
the current temperature biases in the national historical SAT datasets widely used in studies of 106
climate change. This national station network has been conducted for mainland China over the last 107
decade by the CMA (Ren et al., 2008, 2015; Zhang et al., 2010; Ren and Zhou, 2014; Wen et al., 108
2019). Wen et al. (2019) recently adjusted the monthly 1961–2015 temperature data of 763 national 109
stations in China for urbanization bias based on previous research. Their study concluded that the 110
trend of the original temperature was 0.048—0.049/10year higher than that of the urbanization 111
bias adjusted temperature series. Therefore, the urbanization bias adjustment removed systematic 112
bias in the quality-controlled and homogenized data, which had caused an overestimate of the 113
annual warming rate of more than 19.6%. In addition, the spatial patterns of the annual mean SAT 114
linear trends also exhibit an obvious difference from those of the previous analyses. Wen et al. 115
(2019) concluded that a weak warming area appear in the small part of southwestern North China, 116
the northwestern Central China and the eastern part of Southwest China. The annual mean warming 117
trends in Northeast and North China obviously decreased, which caused a relatively more 118
significant cooling in Northeast China after 1998. This indicates that the influence of urbanization–119
bias on the estimate of temperature trend is not ignorable. This adjusted dataset is thus considered 120
to be the most suitable one for use in monitoring and studying regional climate change in China 121
than the only homogenized temperature data. 122
It remains necessary, therefore, to evaluate climatology and the long–term trends of the 2 m air 123
temperature of the reanalysis data against the new observational data exempt from urbanization 124
bias. In this study, we evaluate the applicability of the reanalysis data by using the urbanization 125
bias–adjusted observational data from 1961–2015 across mainland China. The results of this work 126
can serve as a reference for both developers of reanalysis data and researchers of climate change. 127
128
2. MATERIALS AND METHODS 129
The monthly mean surface air temperature (referred to as SAT) observational dataset (referred 130
to as OBS) of 763 national meteorological stations across China for the period 1961–2015 was 131
developed by adjusting for urbanization bias (Wen et al. 2019) after it had been adjusted for 132
inhomogeneities mainly caused by station and instrumentation relocation (Cao et al., 2016). The 133
distribution of the stations is shown in Figure 1. The method of urbanization bias–adjustment 134
consisted of 3 steps: (1) The SAT reference station networks developed in previous studies (Ren et 135
al., 2010; Ren et al., 2015) were applied, and the reference stations chosen for any target stations 136
were stipulated to be within 500 km (Janis et al., 2004; Ren et al., 2012); (2) The weighted averages 137
of monthly mean temperature for all reference stations were calculated using the correlation 138
coefficients of the detrended monthly mean temperature series between the reference stations and 139
the target station as weights; (3) The linear trend bias of temperature between the target station and 140
the reference series was used as the total correction in order to adjust for urbanization bias at each 141
target station (Zhang et al., 2009). Given the lack of high–quality observational data before 1961, 142
we selected 1961–2015 as the research period for comparing the observational and reanalysis 143
temperatures. 144
145
Figure 1. Distribution of the elevation distribution and 763 observation stations across 146
mainland China. (The insert shows the number of stations at various elevations, the asl stands for 147
elevation (units: m), the red bar stands for the number of the elevation distribution) 148
149

Figures (18)
Citations
More filters

01 Jan 2012-
Abstract: Since the 1980s, China has undergone rapid urbanization. Meanwhile, the climate has been warming substantially. In this paper, the urbanization effect on observed temperatures from 1980 to 2009 in China is estimated, based on analysis of urban land use from satellite observation. Urban land-use expansion (ΔU) during 1980–2005 is applied as an urbanization index. According to these ΔU values, stations are divided into three categories: (C1) intense urbanization around the stations; (C2) moderate urbanization around the stations; and (C3) minimal urbanization around the stations. Most C1 stations are in municipalities or provincial capitals, while C2 stations tend to be in prefecture-level cities. C3 stations are mostly in counties. The urban heat island (UHI) effect can be estimated if the urban effect on C3 is negligible. The warming of C1 or C2 relative to that of C3 represents their urbanization effects, assuming that the same larger-scale natural warming has affected each category. For C1, the local urbanization effect is 0.258°C/10 a over 1980–2009, accounting for 41% of the total warming; the trend at C2 is 0.099°C/10 a, or 21%. For all China, the urbanization effect is 0.09°C/10a, accounting for 20% of the total national warming. Winter urban warming is greater than in summer. The assumption of negligible urbanization effect on C3 is debatable, and so the true urbanization effect may equal or slightly exceed estimates. Further, the ΔU index may have some uncertainties, for it is only one of the urbanization indices. However, it provides a new and direct estimation of environmental change, in contrast to indirect indices.

58 citations


Journal Article
TL;DR: The results show that the difference of mean daily temperature between city and suburb is the largest on 24 December in 1995 during the last 4 0 years, and an obvious 12 years cycle of mean annual temperature is found.
Abstract: Variations of mean temperature in Beijing city are ana ly zed in this paper.The results show that the difference of mean daily temperature between city and suburb is the largest on 24 December in 1995 during the last 4 0 years,and the mean daily temperature in the city is 4.6℃ higher than that in the suburb;the difference of temperature between the city and the suburb is the utmost in winter season and the mean seasonal temperature in the city is 1.11℃ higher than that in the suburb,while the difference is the least in spring seaso n,and the temperature in the city is 0.26℃ higher than that in the suburb;for t he inter-annual variability of temperature,the difference of temperature is lit tle from 1961 to 1977,while the difference is great from 1978 to 2000,and the te mperature in the city is 0.62℃ higher than that in the suburb,so the heat islan d effect is reinforced from 1978 to 2000;the difference of temperature is the le ast in 1960's,and the temperature in the city is 0.13℃ higher than that in the suburb,whereas the difference is the maximal in 1990's,and the temperature in th e city is 0.78℃ higher than that in the suburb;the number of high temperature d ays(≥35℃)is going up obviously in recent years,but the highest temperature is not changed greatly,and the highest temperature is higher than 38℃ only in 1997 ,1999 and 2000. The mean annual temperature is increasing apparently during the last 40 years,and it is increasing by 0.43℃/10ɑ in the city and 0.21℃/10ɑ in the suburb,an obvious 12 years cycle of mean annual temperature is found.

10 citations


01 Dec 2015-
Abstract: Many multidecadal atmospheric reanalysis products are available now, but their consistencies and reliability are far from perfect. In this study, atmospheric precipitable water (PW) from the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR), NCEP/Department of Energy (DOE), Modern Era Retrospective-Analysis for Research and Applications (MERRA), Japanese 55 year Reanalysis (JRA-55), JRA-25, ERA-Interim, ERA-40, Climate Forecast System Reanalysis (CFSR), and 20th Century Reanalysis version 2 is evaluated against homogenized radiosonde observations over China during 1979–2012 (1979–2001 for ERA-40). Results suggest that the PW biases in the reanalyses are within ∼20% for most of northern and eastern China, but the reanalyses underestimate the observed PW by 20%–40% over western China and by ∼60% over the southwestern Tibetan Plateau. The newer-generation reanalyses (e.g., JRA25, JRA55, CFSR, and ERA-Interim) have smaller root-mean-square error than the older-generation ones (NCEP/NCAR, NCEP/DOE, and ERA-40). Most of the reanalyses reproduce well the observed PW climatology and interannual variations over China. However, few reanalyses capture the observed long-term PW changes, primarily because they show spurious wet biases before about 2002. This deficiency results mainly from the discontinuities contained in reanalysis relative humidity fields in the middle-lower troposphere due to the wet bias in older radiosonde records that are assimilated into the reanalyses. An empirical orthogonal function (EOF) analysis revealed two leading modes that represent the long-term PW changes and El Nino–Southern Oscillation-related interannual variations with robust spatial patterns. The reanalysis products, especially the MERRA and JRA-25, roughly capture these EOF modes, which account for over 50% of the total variance. The results show that even during the post-1979 satellite era, discontinuities in radiosonde data can still induce large spurious long-term changes in reanalysis PW and other related fields. Thus, more efforts are needed to remove spurious changes in input data for future long-term reanalyses.

3 citations


References
More filters

Journal ArticleDOI
Abstract: The NCEP and NCAR are cooperating in a project (denoted “reanalysis”) to produce a 40-year record of global analyses of atmospheric fields in support of the needs of the research and climate monitoring communities. This effort involves the recovery of land surface, ship, rawinsonde, pibal, aircraft, satellite, and other data; quality controlling and assimilating these data with a data assimilation system that is kept unchanged over the reanalysis period 1957–96. This eliminates perceived climate jumps associated with changes in the data assimilation system. The NCEP/NCAR 40-yr reanalysis uses a frozen state-of-the-art global data assimilation system and a database as complete as possible. The data assimilation and the model used are identical to the global system implemented operationally at the NCEP on 11 January 1995, except that the horizontal resolution is T62 (about 210 km). The database has been enhanced with many sources of observations not available in real time for operations, provided b...

26,349 citations


Journal ArticleDOI
Dick Dee1, S. Uppala1, Adrian Simmons1, Paul Berrisford1  +32 moreInstitutions (6)
Abstract: ERA-Interim is the latest global atmospheric reanalysis produced by the European Centre for Medium-Range Weather Forecasts (ECMWF). The ERA-Interim project was conducted in part to prepare for a new atmospheric reanalysis to replace ERA-40, which will extend back to the early part of the twentieth century. This article describes the forecast model, data assimilation method, and input datasets used to produce ERA-Interim, and discusses the performance of the system. Special emphasis is placed on various difficulties encountered in the production of ERA-40, including the representation of the hydrological cycle, the quality of the stratospheric circulation, and the consistency in time of the reanalysed fields. We provide evidence for substantial improvements in each of these aspects. We also identify areas where further work is needed and describe opportunities and objectives for future reanalysis projects at ECMWF. Copyright © 2011 Royal Meteorological Society

19,659 citations


Journal ArticleDOI
Abstract: The NCEP–DOE Atmospheric Model Intercomparison Project (AMIP-II) reanalysis is a follow-on project to the “50-year” (1948–present) NCEP–NCAR Reanalysis Project. NCEP–DOE AMIP-II re-analysis covers the “20-year” satellite period of 1979 to the present and uses an updated forecast model, updated data assimilation system, improved diagnostic outputs, and fixes for the known processing problems of the NCEP–NCAR reanalysis. Only minor differences are found in the primary analysis variables such as free atmospheric geopotential height and winds in the Northern Hemisphere extratropics, while significant improvements upon NCEP–NCAR reanalysis are made in land surface parameters and land–ocean fluxes. This analysis can be used as a supplement to the NCEP–NCAR reanalysis especially where the original analysis has problems. The differences between the two analyses also provide a measure of uncertainty in current analyses.

4,666 citations


Journal ArticleDOI
15 Jul 2011-Journal of Climate
Abstract: The Modern-Era Retrospective Analysis for Research and Applications (MERRA) was undertaken by NASA’s Global Modeling and Assimilation Office with two primary objectives: to place observations from NASA’s Earth Observing System satellites into a climate context and to improve upon the hydrologic cycle represented in earlier generations of reanalyses. Focusing on the satellite era, from 1979 to the present, MERRA has achieved its goals with significant improvements in precipitation and water vapor climatology. Here, a brief overview of the system and some aspects of its performance, including quality assessment diagnostics from innovation and residual statistics, is given.By comparing MERRA with other updated reanalyses [the interim version of the next ECMWF Re-Analysis (ERA-Interim) and the Climate Forecast System Reanalysis (CFSR)], advances made in this new generation of reanalyses, as well as remaining deficiencies, are identified. Although there is little difference between the new reanalyses i...

4,201 citations


Journal ArticleDOI
Abstract: The National Centers for Environmental Prediction (NCEP) and National Center for Atmospheric Research (NCAR) have cooperated in a project (denoted “reanalysis”) to produce a retroactive record of more than 50 years of global analyses of atmospheric fields in support of the needs of the research and climate monitoring communities. This effort involved the recovery of land surface, ship, rawinsonde, pibal, aircraft, satellite, and other data. These data were then quality controlled and assimilated with a data assimilation system kept unchanged over the reanalysis period. This eliminated perceived climate jumps associated with changes in the operational (real time) data assimilation system, although the reanalysis is still affected by changes in the observing systems. During the earliest decade (1948–57), there were fewer upper-air data observations and they were made 3 h later than the current main synoptic times (e.g., 0300 UTC), and primarily in the Northern Hemisphere, so that the reanalysis is less reliable than for th later 40 years. The reanalysis data assimilation system continues to be used with current data in real time (Climate Data Assimilation System or CDAS), so that its products are available from 1948 to the present. The products include, in addition to the gridded reanalysis fields, 8-day forecasts every 5 days, and the binary universal format representation (BUFR) archive of the atmospheric observations. The products can be obtained from NCAR, NCEP, and from the National Oceanic and Atmospheric Administration/ Climate Diagnostics Center (NOAA/CDC). (Their Web page addresses can be linked to from the Web page of the NCEP–NCAR reanalysis at http:// wesley.wwb.noaa.gov/Reanalysis.html.) This issue of the Bulletin includes a CD-ROM with a documentation of the NCEP–NCAR reanalysis (Kistler et al. 1999). In this paper we present a brief summary and some highlights of the documentation (also available on the Web at http://atmos.umd.edu/ ~ekalnay/). The CD-ROM, similar to the one issued with the March 1996 issue of the Bulletin, contains 41 yr (1958–97) of monthly means of many reanalysis variables and estimates of precipitation derived from satellite and in situ observations (see the appenThe NCEP–NCAR 50-Year Reanalysis: Monthly Means CD-ROM and Documentation

4,106 citations