Performance of the CRA-40/Land, CMFD, and ERA-Interim Datasets in Reflecting Changes in Surface Air Temperature over the Tibetan Plateau

CRA-40/Land、CMFD和ERA-Interim数据集在反映青藏高原气温变化中的表现

+ Author Affiliations + Find other works by these authors
  • Corresponding author: Panmao ZHAI, pmzhai@cma.gov.cn
  • Funds:

    Supported by the Second Tibetan Plateau Scientific Expedition and Research Program (2019QZKK1001) and Science Funds from Beijing Meteorological Service (BMBKJ202003008)

  • doi: 10.1007/s13351-021-0196-x

PDF

  • We analyzed the spatiotemporal variations in surface air temperature and key climate change indicators over the Tibetan Plateau during a common valid period from 1979 to 2018 to evaluate the performance of different datasets on various timescales. We used observations from 22 in-situ observation sites, the CRA-40/Land (CRA) reanalysis dataset, the China Meteorological Forcing Dataset (CMFD), and the ERA-Interim (ERA) reanalysis dataset. The three datasets are spatially consistent with the in-situ observations, but slightly underestimate the annual mean surface air temperature. The daily mean surface air temperature estimated by the CRA, CMFD, and ERA datasets is closer to the in-situ observations after correction for elevation. The CMFD shows the best performance in simulating the annual mean surface air temperature over the Tibetan Plateau, followed by the CRA and ERA datasets with comparable performances. The CMFD is relatively accurate in simulating the daily mean surface air temperature over the Tibetan Plateau on an annual scale, whereas both the CRA and ERA datasets perform better in summer than in winter. The increasing trends in the annual mean surface air temperature over the Tibetan Plateau from 1979 to 2018 reflected by the CRA dataset and the CMFD are 0.5°C (10 yr)−1, similar to the in-situ observations, whereas the warming rate in the ERA dataset is only 0.3°C (10 yr)−1. The trends in the length of the growing season derived from the in-situ observations, the CRA, CMFD, and ERA datasets are 5.3, 4.8, 6.1, and 3.2 day (10 yr)−1, respectively. Our analyses suggest that both the CRA dataset and the CMFD perform better than the ERA dataset in modeling the changes in surface air temperature over the Tibetan Plateau.
    本文基于22个站点的实测数据资料,对比了CRA-40/Land(CRA)再分析数据集、中国气象强迫数据集(CMFD)和ERA-Interim再分析数据集,分析了1979–2018年青藏高原气温和关键气候变化指标的时空变化特征,并评价了不同数据集在反映不同时间尺度气温变化上的表现。3个数据集在空间上与实测资料基本一致,但对年平均气温有轻微低估。经海拔高度插值后,CRA、CMFD和ERA数据集估算的日平均气温与实测数据更接近。CMFD对青藏高原年平均气温的模拟效果最好,CRA和ERA-Interim数据次之。CMFD在年尺度上对青藏高原日平均气温的模拟较为准确,而CRA和ERA-Interim数据在夏季的模拟效果优于冬季。1979–2018年,CRA数据集和CMFD数据集反映的青藏高原年平均气温上升趋势为0.5℃/10a,与实测数据一致,而ERA-Interim数据集的增温速率仅为0.3℃/10a。另外,实测数据、CRA、CMFD和ERA数据的生长季长度变化趋势分别为5.3、4.8、6.1和3.2 days/10a。总之,CRA数据集和CMFD数据集对青藏高原气温变化的表现效果均优于ERA数据集。
  • 加载中
  • Fig. 1.  Location of the national reference climate stations in China.

    Fig. 2.  Spatial patterns of the annual mean surface air temperature over the Tibetan Plateau averaged from 1979 to 2018 for (a) the in-situ observation sites, (b) CRA dataset, (c) CMFD dataset, and (d) ERA dataset.

    Fig. 3.  (a) MBE and (b) RMSE of the daily mean surface air temperature between the in-situ observations and the three gridded datasets for each site. Dataset names without (with) brackets (0.65) indicate results before (after) correction for elevation.

    Fig. 4.  Taylor diagrams comparing the CRA, CMFA and ERA datasets with the in-situ observations of the daily mean surface air temperature (a) on an annual scale, (b) during summer (June–August), and (c) during winter (December–the following February).

    Fig. 5.  Changes in the daily mean surface air temperature [TMP; °C (10 yr)−1] from 1979 to 2018 derived from different datasets for each in-situ station.

    Fig. 6.  Anomalies in the (a) GSL, (b) GSS, and (c) GSE over the Tibetan Plateau from 1979 to 2018.

    Fig. 7.  As in Fig. 5, but for the GSL [day (10 yr)−1].

    Table 1.  Location and information about the national reference climate stations in China

    Station
    No.
    Station nameStation IDLongitude
    (°E)
    Latitude
    (°N)
    Altitude
    (m)
    1Shiquanhe5522880324278
    2Gaize5524884324414
    3Pulan5543781303900
    4Nielamu5565585283810
    5Anduo5529491324800
    6Naqu5529992314507
    7Suoxian5610693314022
    8Shenzha5547288304672
    9Rikaze5557888293836
    10Langkazi5568190284431
    11Cuona5569091274280
    12Linzhi5631294282991
    13Chayu5643497282366
    14Changdu5613797313315
    15Lenghu5260293382770
    16Dachaidan5271395373173
    17Gangcha52754100 373301
    18Geermu5281894362807
    19Nuomuhong5282596362790
    20Zaduo5601895324066
    21Maduo5603398344272
    22Nangqian5612596323643
    Download: Download as CSV
  • [1]

    Berrisford, P., D. Dee, D. P. Poli, et al., 2011a: The ERA-Interim Archive Version 2.0. ERA Report Series. 1. ECMWF, Reading, 23 pp. Available online at https://www.ecmwf.int/node/8174. Accessed on 7 July 2021.
    [2]

    Berrisford, P., P. Kållberg, S. Kobayashi, et al., 2011b: Atmospheric conservation properties in ERA-Interim. Quart. J. Roy. Meteor. Soc., 137, 1381–1399. doi: 10.1002/qj.864.
    [3]

    Burrows, M. T., D. S. Schoeman, L. B. Buckley, et al., 2011: The pace of shifting climate in marine and terrestrial ecosystems. Science, 334, 652–655. doi: 10.1126/science.1210288.
    [4]

    Chen, Y. Y., K. Yang, J. He, et al., 2011: Improving land surface temperature modeling for dry land of China. J. Geophys. Res. Atmos., 116, D20104. doi: 10.1029/2011JD015921.
    [5]

    Dee, D. P., S. M. Uppala, A. J. Simmons, et al., 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597. doi: 10.1002/qj.828.
    [6]

    Dee, D. P., M. Balmaseda, G. Balsamo, et al., 2014: Toward a consistent reanalysis of the climate system. Bull. Amer. Meteor. Soc., 95, 1235–1248. doi: 10.1175/BAMS-D-13-00043.1.
    [7]

    Deng, X. H., P. M. Zhai, and C. H. Yuan, 2010: Comparative analysis of NCEP/NCAR, ECMWF and JMA reanalysis. Meteor. Sci. Technol., 38, 1–8. doi: 10.3969/j.issn.1671-6345.2010.01.001. (in Chinese)
    [8]

    Ding, Y. H., and L. Zhang, 2008: Intercomparison of the time for climate abrupt change between the Tibetan Plateau and other regions in China. Chinese J. Atmos. Sci., 32, 794–805. doi: 10.3878/j.issn.1006-9895.2008.04.08. (in Chinese)
    [9]

    He, J., K. Yang, W. J. Tang, et al., 2020: The first high-resolution meteorological forcing dataset for land process studies over China. Sci. Data, 7, 25. doi: 10.1038/s41597-020-0369-y.
    [10]

    Jiang, H., M. C. Tang, and X. Q. Gao, 2003: Contribution of multi-hotspring regions over Qinghai−Xizang Plateau to air temperature field. Plateau Meteor., 22, 640–642. doi: 10.3321/j.issn:1000-0534.2003.06.018. (in Chinese)
    [11]

    Kang, S. C., Y. W. Xu, Q. L. You, et al., 2010: Review of climate and cryospheric change in the Tibetan Plateau. Environ. Res. Lett., 5, 015101. doi: 10.1088/1748-9326/5/1/015101.
    [12]

    Kato, T., 2016: Chapter 4-Prediction of photovoltaic power generation output and network operation. Integration of Distributed Energy Resources in Power Systems, T. Funabashi, Ed., Academic Press, Pittsburgh, 77–108, doi: 10.1016/B978-0-12-803212-1.00004-0.
    [13]

    Liang, X., L. P. Jiang, Y. Pan, et al., 2020: A 10-yr global land surface reanalysis interim dataset (CRA-Interim/Land): Implementation and preliminary evaluation. J. Meteor. Res., 34, 101–116. doi: 10.1007/s13351-020-9083-0.
    [14]

    Liu, X. D., and B. D. Chen, 2000: Climatic warming in the Tibetan Plateau during recent decades. Int. J. Climatol., 20, 1729–1742. doi: 10.1002/1097-0088(20001130)20:14<1729::AID-JOC556>3.0.CO;2-Y.
    [15]

    Liu, X. D., Z. G. Cheng, L. B. Yan, et al., 2009: Elevation dependency of recent and future minimum surface air temperature trends in the Tibetan Plateau and its surroundings. Global Planet. Change, 68, 164–174. doi: 10.1016/j.gloplacha.2009.03.017.
    [16]

    Liu, Z. Q., C. X. Shi, Z. J. Zhou, et al., 2017: CMA global reanalysis (CRA-40): Status and plans. Proc. 5th International Conference on Reanalysis, 13–17 November, Natl. Meteor. Int. Center, Rome, Italy, 1–16.
    [17]

    Ming, S. H., Z. K. Qin, and Y. Huang, 2019: Climate trend of upper troposphere temperature revealed by satellite data over the Qinghai−Tibetan Plateau. Plateau Meteor., 38, 264–277. (in Chinese)
    [18]

    Pal, R., 2017: Validation methodologies. Predictive Modeling of Drug Sensitivity, R. Pal, Ed., Academic Press, Pittsburgh, 83–107, doi: 10.1016/B978-0-12-805274-7.00004-X.
    [19]

    Pang, H. X., S. G. Hou, W. B. Zhang, et al., 2020: Temperature trends in the northwestern Tibetan Plateau constrained by ice core water isotopes over the past 7000 years. J. Geophys. Res. Atmos., 125, e2020JD032560. doi: 10.1029/2020JD032560.
    [20]

    Qin, J., K. Yang, S. L. Liang, et al., 2009: The altitudinal dependence of recent rapid warming over the Tibetan Plateau. Climatic Change, 97, 321. doi: 10.1007/s10584-009-9733-9.
    [21]

    Smith, R. B., 1979: The influence of mountains on the atmosphere. Adv. Geophys., 21, 87–230. doi: 10.1016/S0065-2687(08)60262-9.
    [22]

    Song, Y. L., C. Y. Wang, H. W. Linderholm, et al., 2019: Agricultural adaptation to global warming in the Tibetan Plateau. Int. J. Environ. Res. Public Health, 16, 3686. doi: 10.3390/ijerph16193686.
    [23]

    Sun, Y. T., Q. J. Gao, and J. Z. Min, 2013: Comparison of reanalysis data and observation about summer/winter surface air temperature in Tibet. Plateau Meteor., 32, 909–920. (in Chinese)
    [24]

    Tang, M. C., H. L. Zhong, and D. L. Li, 2003: The standard of marking off the four seasons along Qinghai−Xizang railway and its temperature variation analysis. Plateau Meteor., 22, 440–444. doi: 10.3321/j.issn:1000-0534.2003.05.002. (in Chinese)
    [25]

    Taylor, K. E., 2001: Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res. Atmos., 106, 7183–7192. doi: 10.1029/2000JD900719.
    [26]

    Wang, C. Z., 2016: A remote sensing perspective of alpine grasslands on the Tibetan Plateau: Better or worse under “Tibet Warming”? Remote Sens. Appl. Soc. Environ., 3, 36–44. doi: 10.1016/j.rsase.2015.12.002.
    [27]

    Wilks, D. S., 2006: Statistical Methods in the Atmospheric Science. Academic Press, Pittsburgh, 627 pp. Available online at https://www.sciencedirect.com/book/9780128158234/statistical-methods-in-the-atmospheric-sciences. Accessed on 7 July 2021.
    [28]

    Yang, K., and J. He, 2016: China meteorological forcing dataset (1979–2015). A Big Earth Data Platform for Three Poles, doi: 10.3972/westdc.002.2014.db.
    [29]

    Yang, M. X., F. E. Nelson, N. I. Shiklomanov, et al., 2010: Permafrost degradation and its environmental effects on the Tibetan Plateau: A review of recent research. Earth-Sci. Rev., 103, 31–44. doi: 10.1016/j.earscirev.2010.07.002.
    [30]

    You, Q. L., D. L. Chen, F. Y. Wu, et al., 2020: Elevation dependent warming over the Tibetan Plateau: Patterns, mechanisms and perspectives. Earth-Sci. Rev., 210, 103349. doi: 10.1016/j.earscirev.2020.103349.
    [31]

    Zhao, T. B., and C. B. Fu, 2006: Preliminary comparison and analysis between ERA-40, NCEP-2 reanalysis and observations over China. Climatic Environ. Res., 11, 14–32. doi: 10.3878/j.issn.1006-9585.2006.01.02. (in Chinese)
    [32]

    Zhao, T. B., and C. B. Fu, 2009: Applicability evaluation of surface air temperature from several reanalysis datasets in China. Plateau Meteor., 28, 594–606. doi: 10.3969/j.issn.1001-7313.2009.01.009. (in Chinese)
    [33]

    Zhao, T. B., W. D. Guo, and C. B. Fu, 2008: Calibrating and evaluating reanalysis surface temperature error by topographic correction. J. Climate, 21, 1440–1446. doi: 10.1175/2007JCLI1463.1.
    [34]

    Zhao, T. B., J. H. Wang, and A. G. Dai, 2015: Evaluation of atmospheric precipitable water from reanalysis products using homogenized radiosonde observations over China. J. Geophys. Res. Atmos., 120, 10703–10727. doi: 10.1002/2015JD023906.
  • Panmao ZHAI and Jiaxi YANG.pdf

  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Performance of the CRA-40/Land, CMFD, and ERA-Interim Datasets in Reflecting Changes in Surface Air Temperature over the Tibetan Plateau

    Corresponding author: Panmao ZHAI, pmzhai@cma.gov.cn
  • 1. Institute of Urban Meteorology, China Meteorological Administration, Beijing 100089
  • 2. State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, China Meteorological Administration, Beijing 100081
Funds: Supported by the Second Tibetan Plateau Scientific Expedition and Research Program (2019QZKK1001) and Science Funds from Beijing Meteorological Service (BMBKJ202003008)

Abstract: We analyzed the spatiotemporal variations in surface air temperature and key climate change indicators over the Tibetan Plateau during a common valid period from 1979 to 2018 to evaluate the performance of different datasets on various timescales. We used observations from 22 in-situ observation sites, the CRA-40/Land (CRA) reanalysis dataset, the China Meteorological Forcing Dataset (CMFD), and the ERA-Interim (ERA) reanalysis dataset. The three datasets are spatially consistent with the in-situ observations, but slightly underestimate the annual mean surface air temperature. The daily mean surface air temperature estimated by the CRA, CMFD, and ERA datasets is closer to the in-situ observations after correction for elevation. The CMFD shows the best performance in simulating the annual mean surface air temperature over the Tibetan Plateau, followed by the CRA and ERA datasets with comparable performances. The CMFD is relatively accurate in simulating the daily mean surface air temperature over the Tibetan Plateau on an annual scale, whereas both the CRA and ERA datasets perform better in summer than in winter. The increasing trends in the annual mean surface air temperature over the Tibetan Plateau from 1979 to 2018 reflected by the CRA dataset and the CMFD are 0.5°C (10 yr)−1, similar to the in-situ observations, whereas the warming rate in the ERA dataset is only 0.3°C (10 yr)−1. The trends in the length of the growing season derived from the in-situ observations, the CRA, CMFD, and ERA datasets are 5.3, 4.8, 6.1, and 3.2 day (10 yr)−1, respectively. Our analyses suggest that both the CRA dataset and the CMFD perform better than the ERA dataset in modeling the changes in surface air temperature over the Tibetan Plateau.

CRA-40/Land、CMFD和ERA-Interim数据集在反映青藏高原气温变化中的表现

本文基于22个站点的实测数据资料,对比了CRA-40/Land(CRA)再分析数据集、中国气象强迫数据集(CMFD)和ERA-Interim再分析数据集,分析了1979–2018年青藏高原气温和关键气候变化指标的时空变化特征,并评价了不同数据集在反映不同时间尺度气温变化上的表现。3个数据集在空间上与实测资料基本一致,但对年平均气温有轻微低估。经海拔高度插值后,CRA、CMFD和ERA数据集估算的日平均气温与实测数据更接近。CMFD对青藏高原年平均气温的模拟效果最好,CRA和ERA-Interim数据次之。CMFD在年尺度上对青藏高原日平均气温的模拟较为准确,而CRA和ERA-Interim数据在夏季的模拟效果优于冬季。1979–2018年,CRA数据集和CMFD数据集反映的青藏高原年平均气温上升趋势为0.5℃/10a,与实测数据一致,而ERA-Interim数据集的增温速率仅为0.3℃/10a。另外,实测数据、CRA、CMFD和ERA数据的生长季长度变化趋势分别为5.3、4.8、6.1和3.2 days/10a。总之,CRA数据集和CMFD数据集对青藏高原气温变化的表现效果均优于ERA数据集。
    • The Tibetan Plateau is extremely sensitive to global climate change, and as a result of its unique geographical location and environmental characteristics, its climatic factors are more representative of climate change than those of low-altitude areas at the same latitude (Liu and Chen, 2000). Climate change over the Tibetan Plateau had affected the water resources and threatened the livelihoods of people living on the plateau and in the surrounding regions. The potential impacts of climate change on both fragile natural ecosystems and socioeconomic development are becoming increasingly apparent (Ding and Zhang, 2008; Kang et al., 2010). The variation in the surface air temperature over the Tibetan Plateau also affects climate change in East Asia (Smith, 1979; Burrows et al., 2011) and may provide important insights into the Holocene temperature conundrum resulting from the enhanced sensitivity of high-altitude regions to global climate change (Pang et al., 2020).

      Changes in surface air temperature and their influence on the Tibetan Plateau have long been of concern. This region has experienced the highest increase in annual air temperature in China in recent years (Song et al., 2019). Warming is greater at high elevations over the Tibetan Plateau than at lower elevations, especially during winter (Liu et al., 2009; Yang et al., 2010). However, existing estimates of the impact of climate change are highly uncertain due to the lack of observational sites in the harsh terrain, especially in the western and central regions of the plateau, and the short time period over which climate data have been collected. Gridded reanalysis datasets are urgently needed to give a more comprehensive understanding of the changes in the surface air temperature over the Tibetan Plateau.

      Reanalysis datasets, multi-source merged datasets, and satellite remote sensing data are widely used in climate change research, but the quality of different datasets varies greatly. Large-coverage satellite imagery is becoming the predominant source of data used to monitor the temperature trend in this unique environment as a result of limited physical access to the Tibetan Plateau (Wang, 2016). However, most of the reanalysis products show a relatively limited performance over the Tibetan Plateau (Zhao et al., 2015).

      Public datasets from the ECMWF are generally recognized as excellent sources of reanalysis data. Zhao et al. (2009) found that the ERA-40 dataset was significantly superior to the NCEP/NCAR reanalysis dataset in long-term studies of climate change in China. Studies have also shown that the ERA-40 and Japanese Re-Analysis 25-yr (JRA-25) datasets are closer to the actual observed values in climate studies (Zhao et al., 2006; Deng et al., 2010). The ERA-Interim (ERA) dataset is a third-generation reanalysis product using a much-improved atmospheric model and assimilation system from those in the ERA-40 dataset (Dee et al., 2011, 2014). The ERA dataset is generally more applicable to the study of temperature change in Tibet than the NCEP/NCAR reanalysis dataset (Sun et al., 2013). At the same altitude, the ERA-40 reanalysis dataset is more applicable and referential than other datasets, but from the vertical perspective, the NCEP/NCAR reanalysis dataset is more universal and accurate (Ming et al., 2019).

      China has recently produced CRA-40/Land (CRA), a land surface reanalysis dataset providing high-quality land element information (Liang et al., 2020). The CRA dataset is an important land surface component of China’s first generation 40-yr global atmospheric reanalysis product (CRA-40; Liu et al., 2017; Liang et al., 2020). The China Meteorological Forcing Dataset (CMFD; Chen et al., 2011; Yang and He, 2016) is the first high-resolution meteorological forcing merged dataset for land processes in China (He et al., 2020). The quality and applicability of these two new datasets over the Tibetan Plateau are worth exploring.

      We carried out comparative analyses of the CRA, CMFD, and ERA datasets and compared their performance with in-situ observational data. We then evaluated the applicability of the three datasets to simulations of the spatiotemporal variations in some key climate change indicators of the surface air temperature over the Tibetan Plateau.

    2.   Data and methods
    • The in-situ observed data are real-time observational data from 22 national reference climate stations on the Tibetan Plateau: 14 stations are located in the Tibetan Autonomous Region and 8 stations in Qinghai Province. These data are archived by the National Meteorological Information Center of the China Meteorological Administration after quality control. Figure 1 shows the location of the reference stations and Table 1 provides data about their locations.

      Figure 1.  Location of the national reference climate stations in China.

      Station
      No.
      Station nameStation IDLongitude
      (°E)
      Latitude
      (°N)
      Altitude
      (m)
      1Shiquanhe5522880324278
      2Gaize5524884324414
      3Pulan5543781303900
      4Nielamu5565585283810
      5Anduo5529491324800
      6Naqu5529992314507
      7Suoxian5610693314022
      8Shenzha5547288304672
      9Rikaze5557888293836
      10Langkazi5568190284431
      11Cuona5569091274280
      12Linzhi5631294282991
      13Chayu5643497282366
      14Changdu5613797313315
      15Lenghu5260293382770
      16Dachaidan5271395373173
      17Gangcha52754100 373301
      18Geermu5281894362807
      19Nuomuhong5282596362790
      20Zaduo5601895324066
      21Maduo5603398344272
      22Nangqian5612596323643

      Table 1.  Location and information about the national reference climate stations in China

    • The CRA dataset covers 40 yr from 1979 to 2018. It is based on an assimilation algorithm, a multi-source fusion method, the Noah3.3 land surface model, and the establishment of core technologies, such as surface parameter optimization by the National Meteorological Information Center. It is China’s first generation global land surface reanalysis dataset. The spatial resolution of the CRA-40 dataset is about 34 km (1152 × 576, Gaussian grid) and the temporal resolution is 3 h. The CRA dataset includes two types of data: atmosphere-driven fusion products and land surface products. The surface air temperature product is developed by ensemble assimilation. The data sources of the surface air temperature product are the global ground observational dataset [China basic data, Climate Forecast System Reanalysis (CFSR), Integrated Surface Database (ISD), and the CRA dataset (http://data.cma.cn/analysis/cra40)]. The CRA dataset is a useful supplement to the land surface elements in the CRA dataset as it uses assimilation and fusion algorithms to improve the quality of near-surface atmosphere-driven data and to optimize surface vegetation/soil parameters (Liang et al., 2020).

    • The CMFD is a reanalysis dataset of near-surface meteorological and environmental elements developed by the Institute of Qinghai–Tibet Plateau, Chinese Academy of Sciences (https://data.tpdc.ac.cn/zh-hans/data/8028b944-daaa-4511-8769-965612652c49/). The dataset is based on the existing international Princeton reanalysis dataset, the Global Land Data Assimilation System (GLDAS) dataset, the World Climate Research Programme Global Energy and Water Exchanges Surface Radiation Budget (GEWEX-SRB) radiation dataset, and the Tropical Rainfall Monitoring Mission (TRMM) precipitation dataset as the background field. The CMFD is constructed by integrating conventional meteorological observational data from the China Meteorological Administration. The temporal resolution is 3 h and the horizontal spatial resolution is 0.1°. The accuracy of this dataset is between that of meteorological observational data and satellite remote sensing data. The CMFD contains seven elements (variables): the 2-m air temperature, surface pressure, specific humidity, 10-m wind speed, downward shortwave radiation, downward longwave radiation, and precipitation rate. It can provide data to drive land surface process simulations for China (Chen et al., 2011; Yang and He, 2016).

    • The ERA dataset provided by ECMWF (https://apps.ecmwf.int/datasets/data/interim-full-daily/levtype=sfc/) is a reanalysis dataset for the global atmosphere covering the data-rich period since 1979. The ERA dataset originally ran from 1989 to 31 August 2019, but a 10-yr extension (1979–1988) was produced in 2011 (Berrisford et al., 2011a). The ERA dataset uses a fixed version of a numerical weather prediction system [the Integrated Forecasting System (IFS)-CY31r2] to produce the reanalysis data. The fixed version ensures that no spurious trend is caused by an evolving numerical weather prediction system, although the changing observing system can create such trends (Berrisford et al., 2011b). The numerical weather prediction system blends or assimilates observations with a previous forecast to obtain the best fit to both. The ERA reanalysis dataset has a resolution of 0.75° × 0.75°; the vertical direction is divided into 60 levels and uses a T55 grid. It provides connectivity between the ERA-40 dataset and the next generation of products.

    • The physical quantity selected for this study here is the daily mean surface air temperature at a 2-m height. A bilinear interpolation method is used to interpolate the reanalysis data at different spatial resolutions to the corresponding national reference climate station. The values of the four grid points around the site are weighted and linear interpolation is used to obtain the value for the site.

      As a result of the drastic change in altitude over the plateau, the difference between the altitude of the reanalysis data and the observational site may cause an abnormal deviation in the air temperature. The surface air temperature of the three datasets therefore needs to be corrected to take into account the decrease in temperature with altitude (Zhao et al., 2008):

      $$\hspace{-68pt} T=t+\gamma \Delta z{,} $$ (1)

      where $ T $ is the surface air temperature after correction for elevation, $ t $ is the surface air temperature after bilinear interpolation, $ \gamma $ is the rate of decrease in temperature with altitude, and $ \Delta z $ is the difference between the altitude of the interpolated model terrain and the actual altitude of the site.

      Common metrics used to evaluate the accuracy of forecasts include the mean bias error (MBE or bias), mean absolute error, and root-mean-square error (RMSE; Kato, 2016). The RMSE is the square root of the MSE. The MBE is primarily used to estimate the average bias in the model and to decide whether steps need to be taken to correct the model bias (Pal, 2017). The RMSE is used to describe the error between two values and mainly reflects the degree of dispersion of the data. We used the MBE, RMSE, and other calculation methods to evaluate the daily mean surface air temperature of different datasets.

      The equation for the MBE is:

      $$\hspace{-22pt} {\rm{MBE}}=\frac{1}{N}{\sum _{i=1}^{N}}\left(M_i-O_i \right){.} $$ (2)

      The equation for the RMSE is:

      $$ {\rm{RMSE}}=\sqrt{\frac{1}{N}\sum _{i=1}^{N}{(M_i-O_i)}^{2}}{,} $$ (3)

      where $ M $ is the value of the reanalysis and merged data, $ O $ is the value of the in-situ observed data, and $ N $ is the number of samples in the analysis period (Wilks, 2006).

      Taylor diagrams provide a visual framework for comparing model results with a reference model or more commonly, to observations (Taylor, 2001). Because different variables may have widely varying numerical values, the model results are normalized by the reference variables. The ratio of the normalized variances indicates the relative amplitude of the model and the observed variations. The pertinent statistics are the weighted pattern correlation and the normalized RMSE differences.

      Based on the thermal growing season length (GSL) defined by the Expert Team on Climate Change Detection and Indices, the start of the thermal growing season (GSS) is defined as the first occurrence in a year of at least six days with a daily mean air temperature > 5°C. Similarly, the first occurrence of at least six days with a daily mean air temperature < 5°C after 1 July indicates the growing season end (GSE). The GSL is measured by the length of the interval between the GSS and the GSE. If the GSS is not found, GSS/GSE will be considered as undefined and the GSL will be set to 0. If no GSE is found, the GSL is counted until the end of the year.

    3.   Results
    • Figure 2 shows that the three datasets are generally consistent in describing the spatial pattern of the daily mean surface air temperature over the Tibetan Plateau. The three gridded datasets (CRA, CMFD, and ERA) show that the overall temperature of the Tibetan Plateau is significantly lower than that of the surrounding areas. There are two clear centers of low temperature over the plateau (the Kunlun and Qilian mountain regions) and the centers of high temperature are located in the Qaidam Basin. All the data show that the daily mean surface air temperature over the Tibetan Plateau is generally less than 0°C. The daily mean surface air temperature is generally less than −5°C over the northwest and southwest of the Tibet Autonomous Region in the Kunlun mountains and the Himalaya. The daily mean surface air temperature in northwest Qinghai Province in the Qaidam Basin is significantly higher than that in other areas, with values more than 5°C in the hinterland of the basin.

      Figure 2.  Spatial patterns of the annual mean surface air temperature over the Tibetan Plateau averaged from 1979 to 2018 for (a) the in-situ observation sites, (b) CRA dataset, (c) CMFD dataset, and (d) ERA dataset.

      The simulation of the annual mean surface air temperature in the CRA dataset is generally more than 0°C, especially in central Tibet, which is significantly higher than that in the two other datasets. However, the observational data in Fig. 2a show that the daily mean surface air temperature in central Tibet from 1979 to 2018 is mostly more than 0°C. The annual mean surface air temperature in central Tibet in the CRA dataset is closer to the in-situ observed data than the two other datasets.

    • Considering the complex topography over the Tibetan Plateau, the terrain height of the reanalysis datasets may differ from the actual altitude, leading to a deviation in the surface air temperature between the reanalysis and merged datasets and the in-situ observed data. According to Tang et al. (2003) and Jiang et al. (2003), the rate of decrease of air temperature with altitude over the Tibetan Plateau is roughly 0.65°C (100 m)−1. We therefore performed altitude interpolation for the three reanalysis temperature datasets based on this rate of decrease.

      Figures 3a and 3b show the average deviation (indicated by the MBE) and RMSE, respectively, of the daily mean surface air temperature (after elevation correction) between the in-situ observations and the gridded data over the Tibetan Plateau after interpolation. Without correction for elevation, the CRA, CMFD, and ERA datasets generally underestimate the daily mean surface air temperature. The annual mean deviation of the datasets from the observational data is −3.0, −1.1, and −3.0°C, respectively, and the RMSE is 4.5, 1.8, and 4.0°C, respectively. After correction for elevation, the daily mean air temperature derived from the CRA, CMFD, and ERA datasets is more comparable, but on average, has a lower annual mean deviation (0.2, 0.09, and 0.2°C, respectively) and lower RMSE (2.7, 1.1, and 2.7°C, respectively). Therefore, after correction for elevation, the three datasets give a better description of the daily mean surface air temperature over the Tibetan Plateau.

      Figure 3.  (a) MBE and (b) RMSE of the daily mean surface air temperature between the in-situ observations and the three gridded datasets for each site. Dataset names without (with) brackets (0.65) indicate results before (after) correction for elevation.

    • After correction for elevation using a rate of decrease in temperature with altitude of 0.65°C (100 m)−1, the three datasets give a better description of the daily mean surface air temperature over the Tibetan Plateau. We therefore use these corrected data in the following evaluation.

    • The Taylor diagrams in Fig. 4a show that the three datasets can fit the daily mean surface air temperature over the Tibetan Plateau on the annual scale. The average correlation of the CRA, CMFD, and ERA datasets and the observational data on the daily scale is 0.96, 0.99, and 0.97, respectively. The mean ratio of the standard deviation for the three datasets is 1.07, 1.01, and 1.00, respectively. The correlation between the CRA, CMFD, and ERA datasets and the observational data for the daily mean air temperature in summer is 0.79, 0.93, and 0.80, respectively (Fig. 4b). The mean ratio of the standard deviation for the CRA, CMFD, and ERA datasets is 0.99, 1.05, and 0.95, respectively. The correlation between the CRA, CMFD, and ERA datasets and the observational data in winter is 0.74, 0.92, and 0.79, respectively. The mean ratio of the standard deviation is 1.00, 1.01, and 0.88, respectively (Fig. 4c).

      Figure 4.  Taylor diagrams comparing the CRA, CMFA and ERA datasets with the in-situ observations of the daily mean surface air temperature (a) on an annual scale, (b) during summer (June–August), and (c) during winter (December–the following February).

      Based on the mean correlation coefficient and the RMSE on the seasonal scale, the CMFD merged dataset is relatively accurate in simulating the daily mean surface air temperature over the Tibetan Plateau in both summer and winter and the performance of the other two datasets is slightly better in summer than in winter. In summer, the surface air temperatures derived from CRA dataset underestimate the actual temperature, whereas the surface air temperatures derived from the ERA dataset are evenly distributed around the observations. This indicates that the ERA dataset gives a more stable simulation of the surface air temperature in summer. In winter, however, the difference between CRA and ERA data is not significant. In summary, CMFD has the best performance in simulating the annual mean surface air temperature over the Tibetan Plateau, followed by the CRA and ERA datasets with comparable performances.

    • Figure 5 shows that the change in the annual mean surface air temperature of most of the stations has a significantly positive trend, with the changes in surface air temperature all less than 1°C (10 yr)−1, consistent with the change in temperature in the three datasets. The linear trend of the annual mean surface air temperature from 1979 to 2018 estimated by the in-situ observations is 0.5°C (10 yr)−1. The increasing trend in the annual mean surface air temperatures derived from the CRA and CMFD datasets during 1979–2018 is 0.5°C (10 yr)−1, whereas the ERA dataset slightly underestimates the warming rate [0.3°C (10 yr)−1]. These results support previous studies showing elevation-dependent warming over the Tibetan Plateau (Qin et al., 2009; You et al., 2020). The exception was Langkazi station (Fig. 5j), where the warming rate was less than at some stations at lower elevations. This may be because the warming rate is not only related to elevation, but also to the topography and vegetation cover.

      Figure 5.  Changes in the daily mean surface air temperature [TMP; °C (10 yr)−1] from 1979 to 2018 derived from different datasets for each in-situ station.

    • The annual mean thermal GSL of the Tibetan Plateau estimated by the observational data extended by 5.3 day (10 yr)−1 from 1979 to 2018 (Fig. 6a). The CRA, CMFD, and ERA datasets consistently show a trend in the GSL of 4.8, 6.1, and 3.2 day (10 yr)−1, respectively. Both the observational and reanalysis datasets show an earlier start of the GSS over the Tibetan Plateau from 1979 to 2018 (Fig. 6b). On average, the advancing rate of the GSS estimated by the observational data and the CRA, CMFD, and ERA reanalysis datasets is 3.4, 2.5, 3.6, and 1.8 day (10 yr)−1, respectively. In terms of the end of the GSS, the in-situ observations show that the GSE advanced by an average of 1.9 day (10 yr)−1, whereas the CRA, CMFD, and ERA datasets show that the GSE is delayed on average by 2.3, 2.5, and 1.4 day (10 yr)−1, respectively (Fig. 6c).

      Figure 6.  Anomalies in the (a) GSL, (b) GSS, and (c) GSE over the Tibetan Plateau from 1979 to 2018.

      Spatially, the changes in the GSL from 1979 to 2018 from the ERA reanalysis dataset at different stations are clearly smaller than the observational data. The changes in the GSL of the CMFD at different stations are generally larger than the observational data. By comparison, the changes in the GSL estimated by the CRA dataset are closer to those in the in-situ observations (Fig. 7).

      Figure 7.  As in Fig. 5, but for the GSL [day (10 yr)−1].

    4.   Conclusions and discussion
    • We evaluated the daily mean surface air temperature derived from three reanalysis datasets (CRA, CMFD, and ERA) and merged data over the Tibetan Plateau from 1979 to 2018 against in-situ observations. Our main conclusions are as follows: (1) The spatial patterns of the annual mean surface air temperature in the CRA, CMFD, and ERA datasets are generally consistent with the in-situ observations over the Tibetan Plateau. Regionally, the annual mean surface air temperature in the CRA dataset is closer to the in-situ observations, especially in central Tibet. (2) All three gridded datasets slightly underestimate the annual mean surface air temperatures. However, after correction for elevation, these datasets are comparable with the station-based in- situ observations. The average differences between these three datasets (CRA, CMFD, and ERA) and the in-situ observations after correction for elevation are 0.2, 0.09, and 0.2°C with RMSEs of 2.7, 1.1, and 2.7°C, respectively. This suggests that correction for elevation can further improve the performance of the reanalysis datasets in reflecting the surface air temperatures over the complex topography of the Tibetan Plateau. The CRA dataset has a comparable performance to the ERA dataset for the daily surface air temperature. (3) The increasing trends in the annual mean surface air temperature derived from the CRA and CMFD datasets during 1979−2018 are both 0.5°C (10 yr)−1, the same as the in-situ observations, but the temperature trend for the ERA dataset is only 0.3°C (10 yr)−1. (4) The GSL trends reflected in the CRA, CMFD, and ERA datasets are 4.8, 6.1, and 3.2 day (10 yr)−1, respectively. Quantitively, the change in the GSL reflected in the CRA dataset is closer to the observed trend 5.3 day (10 yr)−1.

      The unique topography of the Tibetan Plateau and the sparse distribution of observational stations mean that there is a lack of observational data in this important region. We have shown here that reanalysis datasets and merged data for the Tibetan Plateau complement the sparse in-situ observations. The differences in surface air temperature among the three datasets are mainly due to the different reanalysis/merging methods, data sources, and spatial resolution. The performance of both the CRA and CMFD datasets in reflecting the changes in surface air temperature is better than that of the ERA dataset.

Reference (34)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return