• Aaltonen, A., E. Elomaa, A. Tuominen, and P. Valkovuori, 1993: Measurement of precipitation. Symposium on Precipitation and Evaporation, B. Sevruk and M. Lapin, Eds., Slovak Hydrometeorological Institute, 4246.

    • Search Google Scholar
    • Export Citation
  • Ding, Y., D. Yang, B. Ye, and N. Wang, 2007: Effects of bias correction on precipitation trend over China. J. Geophys. Res., 112, D13116, https://doi.org/10.1029/2006jd007938.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ding, Y. H., S. W. Wang, J. Y. Zheng, H. J. Wang, and X. Q. Yang, 2013: Climate of China (in Chinese). Science Press, 557 pp.

  • Duchon, C. E., and C. J. Biddle, 2010: Undercatch of tipping-bucket gauges in high rain rate events. Adv. Geosci., 25, 1115, https://doi.org/10.5194/adgeo-25-11-2010.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Folland, C. K., 1988: Numerical models of the raingauge exposure problem, field experiments and an improved collector design. Quart. J. Roy. Meteor. Soc., 114, 14851516, https://doi.org/10.1002/qj.49711448407.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Goodison, B. E., P. Y. T. Louie, and D. Yang, 1998: WMO solid precipitation measurement intercomparison: Final report. WMO Tech. Doc. 872, 212 pp.

    • Search Google Scholar
    • Export Citation
  • Jiang, Y., Y. Luo, and Z. C. Zhao, 2009: Evaluation of wind speeds in China as simulated by global climate models (in Chinese with English abstract). Acta Meteor. Sin., 67, 923934, https://doi.org/10.11676/qxxb2009.090.

    • Search Google Scholar
    • Export Citation
  • Jiang, Y., Y. Luo, and Z. C. Zhao, 2013: Maximum wind speed changes over China. Acta Meteor. Sin., 27, 6374, https://doi.org/10.1007/s13351-013-0107-x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kochendorfer, J., and Coauthors, 2017: Errors and adjustments for single-Alter shielded and unshielded weighing gauge precipitation measurements from WMO-SPICE. Hydrol. Earth Syst. Sci., 21, 35253542, https://doi.org/10.5194/hess-21-3525-2017.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kurtyka, J. C., 1953: Precipitation measurements study. Department of Registration and Education and States Water Survey Division, Investigation 20, 163 pp.

    • Search Google Scholar
    • Export Citation
  • Larson, L. W., 1971: Precipitation and its measurement: A state of the art. University of Wyoming Water Resources Research Institute, Water Resource Series 24, 74 pp., http://library.wrds.uwyo.edu/wrs/wrs-24/wrs-24-co.html.

  • Larson, L. W., and E. L. Peck, 1974: Accuracy of precipitation measurements for hydrologic modeling. Water Resour. Res., 10, 857863, https://doi.org/10.1029/WR010i004p00857.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, N., Y. Li, and N. Yao, 2018: Bias correction of the observed daily precipitation and redivision of climatic zones in China. Int. J. Climatol., 38, 33693387, https://doi.org/10.1002/joc.5506.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, Y. L., G. Y. Ren, and H. M. Yu, 2012: Climatology of snow in China (in Chinese with English abstract). Sci. Geogr. Sin., 32, 11761185.

    • Search Google Scholar
    • Export Citation
  • Nitu, R., and Coauthors, 2018: WMO Solid Precipitation Intercomparison Experiment (SPICE) (2012–2015). Instruments and Observing Methods Rep. 131, 452 pp., https://library.wmo.int/doc_num.php?explnum_id=5686.

  • Pan, X. C., and Coauthors, 2016: Bias corrections of precipitation measurements across experimental sites in different ecoclimatic regions of western Canada. Cryosphere, 10, 2347, https://doi.org/10.5194/tc-2016-122.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rasmussen, R., and Coauthors, 2012: How well are we measuring snow: The NOAA/FAA/NCAR Winter Precipitation Test Bed. Bull. Amer. Meteor. Soc., 93, 811829, https://doi.org/10.1175/BAMS-D-11-00052.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ren, G. Y., J. Guo, M. Z. Xu, Z. Y. Chu, L. Zhang, X. K. Zou, Q. Q. Li, and X. N. Liu, 2005: Climate changes of China’s mainland over the past half century (in Chinese with English abstract). Acta Meteor. Sin., 63, 942956.

    • Search Google Scholar
    • Export Citation
  • Ren, G. Y., Y. J. Zhan, Y. Y. Ren, Y. Chen, T. Wang, Y. J. Liu, and X. B. Sun, 2015a: Spatial and temporal patterns of precipitation variability over mainland China: I: Climatology (in Chinese with English abstract). Adv. Water Sci., 26, 299310.

    • Search Google Scholar
    • Export Citation
  • Ren, G. Y., Y. Y. Ren, Y. J. Zhan, X. B. Sun, Y. J. Liu, Y. Chen, and T. Wang, 2015b: Spatial and temporal patterns of precipitation variability over mainland China: II: Recent trends (in Chinese with English abstract). Adv. Water Sci., 26, 451156.

    • Search Google Scholar
    • Export Citation
  • Ren, G. Y., and Coauthors, 2016: Spatial and temporal patterns of precipitation variability over mainland China: III: Causes for recent trends (in Chinese with English abstract). Adv. Water Sci., 27, 327348.

    • Search Google Scholar
    • Export Citation
  • Ren, Z. H., and M. Q. Li, 2007: Errors and correction of precipitation measurements in China. Adv. Atmos. Sci., 24, 449458, https://doi.org/10.1007/s00376-007-0449-3.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ren, Z. H., G. L. Wang, F. L. Zou, and H. Z. Zhang, 2003: The research of precipitation measurement errors in China (in Chinese with English abstract). Acta Meteor. Sin., 61, 621627.

    • Search Google Scholar
    • Export Citation
  • Scaff, L., D. Q. Yang, Y. P. Li, and E. Mekis, 2015: Inconsistency in precipitation measurements across the Alaska–Yukon border. Cryosphere, 9, 24172428, https://doi.org/10.5194/tc-9-2417-2015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sevruk, B., 1982: Method of correction for systematic error in point precipitation measurement for operational use. WMO Tech. Doc. 589, 91 pp.

  • Sevruk, B., and W. R. Hamon, 1984: International comparison of national precipitation gauges with a reference pit gauge. WMO Instrument and Observing Methods Rep. 17, 111 pp.

    • Search Google Scholar
    • Export Citation
  • Sevruk, B., and L. Zahlavova, 1994: Classification system of precipitation gauge site exposure: Evaluation and application. Int. J. Climatol., 14, 681689, https://doi.org/10.1002/joc.3370140607.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sun, X. B., G. Y. Ren, Z. H. Ren, and Z. C. Shen, 2013: Effects of wind-induced errors on winter snowfall and its trends (in Chinese with English abstract). Climatic Environ. Res., 18, 178186.

    • Search Google Scholar
    • Export Citation
  • Westra, S., and Coauthors, 2014: Future changes to the intensity and frequency of short-duration extreme rainfall. Rev. Geophys., 52, 522555, https://doi.org/10.1002/2014RG000464.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yang, D., Y. F. Shi, E. S. Kang, Y. S. Zhang, and X. Y. Yang, 1991: Results of solid precipitation measurement intercomparison in the Alpine area of Urumqi River Basin. Chin. Sci. Bull., 36, 11051109.

    • Search Google Scholar
    • Export Citation
  • Yang, D., and Coauthors, 1995: Accuracy of Tretyakov precipitation gauge: Result of WMO intercomparison. Hydrol. Processes, 9, 877895, https://doi.org/10.1002/hyp.3360090805.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yang, D., B. E. Goodison, J. R. Metcalfe, V. S. Golubev, R. Bates, T. Pangburn, and C. L. Hanson, 1998: Accuracy of NWS 8 standard nonrecording precipitation gauge: Results and application of WMO intercomparison. J. Atmos. Oceanic Technol., 30, 5768, https://doi.org/10.1175/1520-0426(1998)015<0054:AONSNP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Yang, D., and Coauthors, 2001: Compatibility evaluation of national precipitation gauge measurements. J. Geophys. Res., 106, 14811492, https://doi.org/10.1029/2000JD900612.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yang, D., D. Kane, Z. Zhang, D. Legates, and B. Goodison, 2005: Bias-corrections of long-term (1973–2004) daily precipitation data over the northern regions. Geophys. Res. Lett., 32, L19501, https://doi.org/10.1029/2005GL024057.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yao, N., Y. Li, N. Li, D. Yang, and O. Ayantobo, 2018: Bias correction of precipitation data and its effects on aridity and drought assessment in China over 1961–2015. Sci. Total Environ., 639, 10151027, https://doi.org/10.1016/j.scitotenv.2018.05.243.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ye, B., D. Yang, Y. Ding, T. Han, and T. Koike, 2004: A bias corrected precipitation climatology for China. J. Hydrometeor., 5, 11471160, https://doi.org/10.1175/JHM-366.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ye, B., D. Yang, and L. Ma, 2012: Effect of precipitation bias-correction on water budget calculation in Upper Yellow River, China. Environ. Res. Lett., 7, 025201, https://doi.org/10.1088/1748-9326/7/2/025201.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, Z. F., S. Xi, N. Liu, Y. Yu, S. H. Fan, and M. N. Feng, 2015: Snowfall change characteristics in China from 1961 to 2012 (in Chinese with English abstract). Ziyuan Kexue, 37, 17651773, http://www.resci.cn/CN/Y2015/V37/I9/1765.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    (a) Distribution of the 2379 CMA national stations (red dots) over mainland China and seven stations with no relocations (blue pluses) chosen to assess the bias correction results, and (b) temporal evolution of the number of available stations with at least 75%, 90%, and 99% valid daily records within each particular year.

  • View in gallery

    Frequency distribution of (a) the maximum and (b) mean bias correction for the total correction of daily precipitation in all stations.

  • View in gallery

    Measured/corrected climatology and correction percentages for the climatology of (a) annual precipitation (mm), (b) daily precipitation intensity (mm day−1); (c),(d) light rain amount (mm) and intensity (mm day−1), respectively; and (e),(f) rainstorm amount (mm) and intensity (mm day−1), respectively, during the period of 1961–2013 for 7 representative gauges with no relocations.

  • View in gallery

    Frequency of occurrence distribution of bias corrected (blue lines), measured (red lines) daily precipitation, and the frequency gap (black lines) between them for 7 representative gauges with no relocations.

  • View in gallery

    Mean of seasonal (left) correction amount (mm) and (right) percentage (%) for trace precipitation for 1961–2013 over mainland China.

  • View in gallery

    As in Fig. 5, but for wetting losses.

  • View in gallery

    As in Fig. 5, but for wind-induced undercatch.

  • View in gallery

    The per 10 a trend of annual mean wind speed for precipitation days [m s−1 (10 yr)−1] for 1961–2013 over mainland China, and the regions having significant changes at 0.05 confidence level are stippled.

  • View in gallery

    (a) Climatology of bias corrected annual precipitation (mm), (b) specific contours of annual precipitation (mm), (c) climatology of precipitation intensity (mm day−1), (d) specific contours of precipitation intensity (mm day−1), and (e) correction percentage of precipitation intensity (%) for 1961–2013 over mainland China. Blue (red) lines in (b),(d) represent climatologies with (without) bias correction.

  • View in gallery

    Climatology of (left) seasonal [(a)–(c) for DJF; (d)–(f) for MAM; (g)–(i) for JJA; (j)–(l) for SON] precipitation after bias correction (mm), (middle) specific contours of seasonal precipitation (mm) with (blue lines) and without (red lines) bias correction, and (right) the corresponding correction percentage (%) for 1961–2013 over mainland China.

  • View in gallery

    As in Fig. 9, but for annual light rain.

  • View in gallery

    As in Fig. 9, but for annual rainstorm.

  • View in gallery

    The total number of increased rainstorm events (days) due to bias correction for 1961–2013 over mainland China.

  • View in gallery

    As in Fig. 9, but for annual snowfall.

  • View in gallery

    The corrected snowfall (mm, red lines) by using method I and the absolute difference between two correction methods (mm, blue lines) for each snow event at station (a) 50564, (b) 51358, (c) 53594, (d) 55248, and (e) 57279.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 421 351 25
PDF Downloads 371 307 23

Bias Correction of Gauge Data and its Effect on Precipitation Climatology over Mainland China

View More View Less
  • 1 Laboratory for Climate Studies, National Climate Center, China Meteorological Administration, Beijing, China
  • | 2 Department of Atmospheric Science, School of Environmental Studies, China University of Geosciences, Wuhan, and Laboratory for Climate Studies, National Climate Center, China Meteorological Administration, Beijing, China
  • | 3 Laboratory for Climate Studies, National Climate Center, China Meteorological Administration, Beijing, China
© Get Permissions
Full access

Abstract

Typical rain gauge measurements have long been recognized to underestimate actual precipitation. Long-term daily precipitation records during 1961–2013 from a dense national network of 2379 gauges were corrected to remove systematic errors caused by trace precipitation, wetting losses, and wind-induced undercatch. The corrected percentage was higher in cold seasons and lower in warm seasons. Both trace precipitation and wetting loss corrections were more important in arid regions than in wet regions. A greater correction percentage for wind-induced error could be found in cold and arid regions, as well as high wind speed areas. Generally, the annual precipitation amounts as well as the annual precipitation intensity increased to varying degrees after bias correction with the maximum percentage being about 35%. More importantly, the bias-corrected snowfall amount as well as the rainstorm amount increased remarkably by percentages of more than 50% and 18%, respectively. Remarkably, the total number of actual rainstorm events during the past 53 years could be 90 days more than the observed rainstorm events in some coastal areas of China. Therefore, the actual amounts of precipitation, snowfall, and intense rainfall were much higher than previously measured over China. Bias correction is thus needed to obtain accurate estimates of precipitation amounts and precipitation intensity.

© 2019 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: G. Ren, guoyoo@cma.gov.cn

Abstract

Typical rain gauge measurements have long been recognized to underestimate actual precipitation. Long-term daily precipitation records during 1961–2013 from a dense national network of 2379 gauges were corrected to remove systematic errors caused by trace precipitation, wetting losses, and wind-induced undercatch. The corrected percentage was higher in cold seasons and lower in warm seasons. Both trace precipitation and wetting loss corrections were more important in arid regions than in wet regions. A greater correction percentage for wind-induced error could be found in cold and arid regions, as well as high wind speed areas. Generally, the annual precipitation amounts as well as the annual precipitation intensity increased to varying degrees after bias correction with the maximum percentage being about 35%. More importantly, the bias-corrected snowfall amount as well as the rainstorm amount increased remarkably by percentages of more than 50% and 18%, respectively. Remarkably, the total number of actual rainstorm events during the past 53 years could be 90 days more than the observed rainstorm events in some coastal areas of China. Therefore, the actual amounts of precipitation, snowfall, and intense rainfall were much higher than previously measured over China. Bias correction is thus needed to obtain accurate estimates of precipitation amounts and precipitation intensity.

© 2019 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: G. Ren, guoyoo@cma.gov.cn

1. Introduction

Precipitation is the most widely recorded hydroclimatic phenomenon and an essential component of the hydrological cycle. Precipitation observations are obtained from radars, satellites, and rain gauges, with rain gauges having been widely and continuously utilized for centuries around the world. Deficiencies in existing gauge measurements, however, have long been recognized because they usually underestimate the actual precipitation (Kurtyka 1953; Larson 1971; Rasmussen et al. 2012; Nitu et al. 2018). Rain gauge measurements of precipitation suffer from a range of mostly negative biases, especially trace precipitation, wetting losses, evaporation losses, and wind undercatch. Bias correction for precipitation could affect the precipitation climatology and trend, as well as water budget (Ye et al. 2012) and drought assessment (Yao et al. 2018).

Negative biases arise because of the inability to measure the very small amounts of precipitation that are reported as “trace” and ignored when calculating monthly or longer-term totals. Wetting loss occurs when a gauge is emptied into a measuring device to obtain the precipitation total. The small amount left in the gauge (sticking to its sides) is the wetting loss. Overall wetting losses therefore depend on how often the gauge is emptied as well as the type of gauge and the type of precipitation. For example, according to the measurements by Aaltonen et al. (1993), the average wetting losses for the Hellmann gauge is 0.14 and 0.1 mm per measurement of rain and snow when the gauge is emptied once, respectively. This is the minimum correction since generally more than one observation was made in a precipitation day. The biases due to trace precipitation are greater in percentage terms at sites where precipitation is predominantly light and the total precipitation is small, for example in polar and cold-desert climates. Like the “trace”-related bias, wetting losses are greater in percentage terms where precipitation is predominantly light.

Gauges record too little precipitation, especially snow. Surrounding winds are strong because of the deformation of the wind field by the gauge (Folland 1988). The severity of the undercatch depends on the shape of the orifice (Folland 1988) and the exposure of the site, which affects the wind strength. Sevruk and Zahlavova (1994) reported that the average undercatch at a shielded site of Swiss 600 m above sea level was 2% in summer and 7% in winter; at an exposed site, the corresponding losses were 5% and 18%. At high and exposed sites (2000 m above sea level) where winds are strong and snow is more likely to occur, the undercatch reached 25% in summer and even 60% in winter. Besides, 80%–120% correction factors were found in winter over the region north of 40°N (Yang et al. 2005). The errors in the unadjusted shielded measurements were also generally smaller than the unadjusted unshielded measurements, because the shields are designed to reduce the horizontal wind impacting a gauge inside the shield, and thereby reduce the effects of the gauge on the flow around it (Kochendorfer et al. 2017).

However, our understanding of the factors causing undercatch in the shields was not sufficiently well advanced to allow for the optimal shield design and only allowed the development of empirical correction factors (Rasmussen et al. 2012). The Solid Precipitation Intercomparison Experiment (SPICE) was conducted as an internationally coordinated project during 2013–15. SPICE focused on recommendations for adjustments that account for the undercatch of solid precipitation due to gauge exposure, and presented a function of data available at operational sites (Nitu et al. 2018). The winter snowfall in northeast China was generally undervalued due to wind-induced undercatch, and the average of annual wind-induced error was 34.1% during the period of 1960–2009 (Sun et al. 2013). Besides, the wind-induced error obviously resulted in a general overestimation of long-term winter snowfall trend in northeast China, and this is due to the weakening of measured near-surface wind speed (Sun et al. 2013). Scaff et al. (2015) discovered significant inconsistency in the precipitation measurements across the United States and Canada border because of the different instruments and observation methods used. This discontinuity was greater for snowfall than for rainfall, as gauge snowfall observations had large errors in the windy and cold conditions. In addition, Pan et al. (2016) found the bias corrections varied greatly in different eco-climatic regions of western Canada.

Evaporation from the gauge leads to a further negative bias, which is highly dependent on the weather conditions and the site environment. In particular, the surface of the orifice may receive some moisture, which later evaporates without entering the gauge. Water may splash out of the orifice leading to further undercatch, or into it from the ground leading to potential overcatch (Folland 1988). Likewise, snow may blow out of or into the orifice. Tipping-bucket gauges usually underestimate very heavy rain because some precipitation is missed during the bucket-tipping process (Duchon and Biddle 2010). The loss depends on precipitation intensity, and according to Duchon and Biddle (2010), precipitation missed in this way is significant only at intensities exceeding 50 mm h−1 and is about 4% in a 75 mm h−1 event. In addition, mechanical errors may also affect tipping-bucket gauge measurements (Westra et al. 2014).

Compensation of rain gauge data for systematic biases ideally requires metadata on instrumentation, siting (which affects wind strength), and observing practices. Ren et al. (2003) and Ren and Li (2007) assessed precipitation measurement biases in China using a parallel observational dataset and pointed out great observational bias in the gauge records. Ye et al. (2004) adjusted daily Chinese rain gauge data from 710 stations during 1951–98, which was followed by analysis of the effects of bias correction on precipitation trend (Ding et al. 2007). Based on the adjusted data, new precipitation climatology was also generated. Li et al. (2018) have updated the former results of Ye et al. (2004) at 553 sites in China, and redivided the climate zones of China according to the bias-corrected data. However, precipitation bias correction from the dense national network of 2379 gauges has not yet been conducted so far.

Precipitation is one climatic element that is sensitive to the local environment and topography. A higher-density network can describe the spatial characteristics of precipitation in details. More importantly, the near-surface wind speeds over China have been found to be reduced in recent decades (Ren et al. 2005; Jiang et al. 2009), and this may further complicate the understanding of precipitation bias in this country. Therefore, assessment and adjustment of the precipitation bias for data from the denser network of 2379 stations is urgently needed. Because the bias-correction method for precipitation could be gauge specific, of course, the uncertainty in corrections is up in the air. In addition, the impact of bias correction on extreme precipitation climatology is still unclear.

The objective of this study is to assess and adjust the precipitation data bias for the 2379 national stations over China. The structure of this study is as follows. In section 2, the dataset and bias-correction methods are introduced. Section 3 presents the results of bias correction. In section 4, we focus on the impact of bias correction on the annual precipitation amount and intensity, and on extreme precipitation climatology. Section 5 discusses the potential uncertainty in snowfall corrections by employing two existing methods for a subset of the observation sites. Conclusions are given in section 6.

2. Data and methods

a. Data

We used measured daily cumulative precipitation amounts, precipitation type (rainfall/snowfall/mixed), daily mean temperature, and daily mean wind speed at 10-m of a dense national network of 2379 stations from 1961 to 2013 provided by the National Meteorological Information Centre (NMIC) of the China Meteorological Administration (CMA). The data have been used in various studies on climate change over China (Ren et al. 2005; Ding et al. 2013; Jiang et al. 2013; Ren et al. 2015a,b; Zhang et al. 2015; Ren et al. 2016). The daily precipitation data were obtained operatively by applying the Chinese standard precipitation gauge (CSPG) for manual observation during 1961–2003, which can measure all types of precipitation. After 2003 tipping-bucket sensors were used for automatic rainfall observation. The most commonly used was the double bucket tipping rainfall sensor. While in winter, the snowfall was measured by weighing precipitation sensor, which can obtain the precipitation amount by weighing without heating the instrument. The information of precipitation type was available in the metadata. The CSPG is a cylinder of galvanized iron with a diameter of 20 cm, and the automatic observation instrument is a tipping-bucket precipitation sensor installed with a water holding device with a diameter of 20 cm. All the gauges were placed 0.7 m above the ground but without wind shelters after 1960. This means wind-induced undercatch was substantial. According to the criterion of surface meteorological observation in China, precipitation was recorded twice a day at 0800 and 2000 [local time (LT) in Beijing], and daily precipitation released by NMIC is the accumulated precipitation amount during one day. Mean wind speed was recorded at 0200, 0800, 1400, and 2000 LT, and is defined as the mean value during the 2-min period before the hour. The available daily mean wind speed released by NMIC is the average of the mean wind speed at the four individual measurement times. The daily mean temperature released by NMIC is the average value of the temperatures measured at the four time points, the same as the daily mean wind speed.

All the climate data used in this work were subject to quality control by the NMIC including extreme-value and internal consistency checks. This lead to the selection of 2379 CMA national stations having valid records for at least 30 years (Fig. 1a). Figure 1a indicates that more stations were in the southern and southeastern regions of China, while in the northern and northwestern parts the network was sparse. Figure 1b shows the number of available stations year by year with different thresholds for at least 75%, 90%, and 99% valid daily records within each year. The number of stations grew until 1961, when it was maintained at more than 2000. This is one of the reasons why we chose 1961–2013 as the study period.

Fig. 1.
Fig. 1.

(a) Distribution of the 2379 CMA national stations (red dots) over mainland China and seven stations with no relocations (blue pluses) chosen to assess the bias correction results, and (b) temporal evolution of the number of available stations with at least 75%, 90%, and 99% valid daily records within each particular year.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

b. Bias-correction methods

According to Sevruk and Hamon (1984), bias correction for gauge precipitation mainly includes trace precipitation, wetting losses, evaporation losses, and wind undercatch errors. The general formula for precipitation bias correction was modified as follows (Sevruk and Hamon 1984; Yang et al. 2001; Ye et al. 2004):
Pc=K(Pg+ΔPw+ΔPe+ΔPt),
where Pc is the corrected precipitation; Pg is measured gauge precipitation; ΔPw, ΔPe, and ΔPt are wetting losses, evaporation losses, and trace precipitation, respectively; and K is the correction coefficient for wind-induced errors, which is defined as 1/CR, and CR is the catch ratios due to wind-induced undercatch.
To determine the systematic biases in gauge measurements for precipitation, a gauge intercomparison study between CSPG and reference gauges was carried out in Urumqi River basin, which is located in northwest China (Yang et al. 1991). According to the intercomparison study, catch ratios due to wind-induced undercatch have been developed for different precipitation types (i.e., snow, rain, and mixed) (Yang et al. 1991; Ye et al. 2004):
CRs(snow)=exp(0.056Ws)×100,
CRr(rain)=exp(0.041Ws)×100,and
CRm(mixed)=CRs(CRsCRr)×(Td+2)/4,
where CRs, CRr, and CRm are catch ratios for snow, rainfall, and mixed precipitation, respectively; Ws is the daily 10-m mean wind speed; and Td is the daily mean temperature. It is noteworthy that the catch ratio for mixed precipitation has a linear relationship with catch ratios for snow and rainfall and daily mean temperature (Td). The correction factor of wind-induced errors for snowfall in Yang et al. (1991) is more conservative than that in Sun et al. (2013). Therefore, formulas (2)(4) developed by Yang et al. (1991) were adopted, and the wind-loss correction is defined as 1/CR. Finally, the corrected daily precipitation can be obtained through the following formula (Ye et al. 2004):
Pc={(Pg+ΔPw)/CRΔPt,
where Pc is the corrected precipitation; Pg is measured gauge precipitation; ΔPw is the wetting loss correction; ΔPt is trace precipitation correction; and CR is the catch ratio of wind-induced undercatch obtained from Eqs. (2)(4). The upper section in Eq. (5) is for measurable precipitation events, and the one below is for trace precipitation events.

According to the experiments by Yang et al. (1991), the average wetting loss of per observation was 0.23 mm for rainfall measurement, 0.30 for snowfall, and 0.29 mm for mixed precipitation. Therefore, we corrected the wetting losses once for each precipitation day according to the above criteria. For gauge observations in China, a trace event is recorded when precipitation is less than 0.1 mm, which is usually below the measured resolution. Trace precipitations are counted as precipitation days, but treated as zero quantitatively. Two trace precipitation events are sometimes reported in a single trace precipitation day in China. Following Yang et al. (1991) and Ye et al. (2004), to be conservative, we assigned a value of 0.10 mm to a given trace day regardless of the number of the trace events reported in a day.

Wind speed is the most important factor for gauge precipitation catch. To correct wind-induced undercatch, wind speed is required. The data of wind measurements at a standard height of 10 m were available for correction. It has been suggested that gauge exposure should be considered when reducing wind from the standard height to the gauge height (Sevruk 1982). Gauge exposure depends on the average vertical angle of obstacles around the gauge. The relationship between the gauge height and 10 m height wind varies with site exposure, which can be directly measured or estimated by a classification system based on metadata archives (Sevruk 1982). The station metadata are not available for precipitation bias corrections, so site exposure was not accounted for in wind speed estimates at the gauge height. This may introduce some uncertainties in the estimation of gauge catch efficiency. Depending on the different precipitation type (rainfall, snow, and mixed precipitation), the wind-induced undercatch could be different in quantity. It has been documented that for the same wind speed, gauge undercatch of snow is much higher than rain (Larson and Peck 1974; Goodison et al. 1998; Yang et al. 1995, 1998; Ye et al. 2004; Rasmussen et al. 2012; Sun et al. 2013; Nitu et al. 2018).

Though the correction formulas we adopted were determined by an intercomparison experiment between CSPG and reference gauges, uniform bias correction was applied both for CSPG and automatic gauges. Uniform bias correction of wind-induced undercatch was applied both for the manual and the automatic observation instruments, because both gauges had the same diameter and were installed at the same height of 0.7 m without wind shelters. However, the biases are mostly gauge specific, and applying uniform bias correction may lead to some potential uncertainties. The same correction for trace precipitation was used both for manual and automatic gauges, because a precipitation event of less than 0.10 mm is both below the resolution of the CSPG and tipping-bucket rainfall sensor. The biases of wetting losses of automatic gauges and CSPG may be slightly different theoretically. But the corresponding intercomparison experiment has not yet been carried out. This is a source of uncertainty for wetting loss correction. We hope to improve this in the future work. Tipping buckets have the unfortunate property of underestimating rainfall in high rain rates, because the collected rainfall would loss during tipping when rain is not being measured during the finite time required for bucket to tip from one side to the other. It is hard to evaluate the undercatch of tipping buckets, so this kind of undercatch was ignored after the beginning of 2003. Evaporation loss is time varying and site dependent, and it is unreasonable to estimate daily evaporation losses at large regional observation networks by using experimental results obtained from a few gauge sites (Ye et al. 2004). Therefore, only trace precipitation, wetting losses, and wind undercatch errors for measured gauge precipitation were corrected in our study.

c. Statistical analysis methods

In all following sections, the seasonal or annual bias correction is the total amount of daily correction for each season [December–February (DJF), March–May (MAM), June–August (JJA), September–November (SON)] or a full year, and the corresponding climatology is the average from 1961 to 2013. Annual/seasonal precipitation is the cumulative amount during the corresponding period. The precipitation amount divided by the number of precipitation days is defined as precipitation intensity. The correction percentage is the ratio of the corrected amount to the total measured precipitation. These definitions of precipitation were also applied to snowfall, the units of which refer to liquid equivalents.

Daily measured precipitation between 0.1 and 10.0 mm is defined as light rain. Light rain amounts and intensity are the cumulative amount classified as light rain and the ratio of the total amount of light rain to the number of light rain days. The related definition of rainstorm is the same as that for light rain, but for daily measured precipitation more than 50.0 mm. We used the unadjusted precipitation classification for the adjusted precipitation—in other words, we kept the original classification intact, and evaluated the changes in precipitation based solely on their original classifications.

3. Results of bias correction

By applying the methods described above, we corrected the bias of daily precipitation for the 2379 stations for the period of 1961–2013. Figure 2 illustrates the frequency distribution of the maximum and mean bias correction for all gauges. We found that most stations had a daily precipitation correction of 20–40 mm with the largest correction exceeding 200 mm (Fig. 2a). The most frequent mean correction amount was between 0.6 and 0.8 mm, with the highest value at about 2.2 mm (Fig. 2b).

Fig. 2.
Fig. 2.

Frequency distribution of (a) the maximum and (b) mean bias correction for the total correction of daily precipitation in all stations.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

Seven stations with no relocations and one for each of the seven regions (i.e., northeast, northwest, north, central, southwest, and south China, and Tibet) were chosen to assess the results of bias correction in detail (Fig. 1a and Table 1). The seven stations are located in arid regions (annual precipitation is less than 200 mm), semiarid regions (annual precipitation between 200 and 400 mm), semihumid regions (annual precipitation between 400 and 800 mm), and humid regions (annual precipitation more than 800 mm), respectively (Fig. 3a). In general, the annual bias correction is small for all stations, while the correction percentage is large (between 5% and 25%), both for annual precipitation and daily precipitation intensity (Figs. 3a,b). The lower the annual precipitation/daily precipitation intensity is, the higher the correction proportion. Both Lingqiu and Sunwu stations are located in semihumid regions, but the correction percentage of annual precipitation/daily intensity at Lingqiu station is clearly lower than that at Sunwu station. This is due to the larger proportion of snowfall at Sunwu located in higher latitude than at Lingqiu. The correction percentage of annual precipitation/daily intensity at Lingqiu station is similar to that at Zaoyang station, while the difference of precipitation climatology between them is quite large. The reason is that the snowfall percentage at Lingqiu station (4.7%) is higher than that at Zaoyang station (1.9%).

Table 1.

Locations of 7 representative gauges with no relocations chosen to assess bias correction.

Table 1.
Fig. 3.
Fig. 3.

Measured/corrected climatology and correction percentages for the climatology of (a) annual precipitation (mm), (b) daily precipitation intensity (mm day−1); (c),(d) light rain amount (mm) and intensity (mm day−1), respectively; and (e),(f) rainstorm amount (mm) and intensity (mm day−1), respectively, during the period of 1961–2013 for 7 representative gauges with no relocations.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

The features of correction amount/percentage for light rain (Figs. 3c,d) are quite similar to that for unclassified precipitation (Figs. 3a,b). The only caveat is that the difference of amount/percentage among stations for light rain is smaller, compared to that for unclassified precipitation. Due to a lack of heavy precipitation for some of the arid and semiarid stations, the effects of the corrections on rainstorm amount and intensity could not be determined (Figs. 3e,f). It makes sense that rainstorm amount/intensity is much higher in humid-area stations (e.g., Jingdong, Zaoyang, and Guangning). However, the correction percentage of rainstorm intensity in some semihumid areas (e.g., Sunwu) is much higher than that in humid regions (e.g., Zaoyang).

Generally, the frequency of occurrence decreases with precipitation amount for all selected gauges (Fig. 4). After bias correction, the corrected frequency of occurrence increased for most precipitation grades and most stations. The difference of frequency between bias corrected and measured precipitation falls with the increase of precipitation grade. Meanwhile, the frequency gap is larger in arid (e.g., Gaize) and semiarid (e.g., Wulanwusu) regions than semihumid (e.g., Sunwu and Lingqiu) and humid (e.g., Jingdong, Zaoyang, and Guangning) regions. For the same climate type, this gap is larger in colder regions (e.g., Sunwu) than warmer regions (e.g., Lingqiu), due to more snowfall in the colder environment.

Fig. 4.
Fig. 4.

Frequency of occurrence distribution of bias corrected (blue lines), measured (red lines) daily precipitation, and the frequency gap (black lines) between them for 7 representative gauges with no relocations.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

To understand the seasonal differences of bias correction, seasonal correction amount and correction percentage for each correction component are also evaluated. Overall, the trace precipitation correction amount differs slightly in the four seasons (Fig. 5). In DJF, MAM, and SON, maximum trace precipitation correction appears in southwest China (including the eastern part of the Tibetan Plateau), the northern part of northwest China, and the western part of northeast China. In JJA, the corrected amount shows the largest values in northwest China. The correction percentage has a quite different spatial and temporal pattern from corrected amount due to different rainfall climatology (Fig. 5). Generally, the correction percentage is higher in DJF and lower in JJA, and the percentage value decreases from northwest to southeast China. Though the absolute magnitude of the trace precipitation correction is not quite large (less than 3.0 mm) across China, the correction percentage is dramatic, especially in the northwestern region of China.

Fig. 5.
Fig. 5.

Mean of seasonal (left) correction amount (mm) and (right) percentage (%) for trace precipitation for 1961–2013 over mainland China.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

Figure 6 shows that the wetting loss amount in JJA is more than that in DJF. This is due to the number of precipitation days in summer being much more than those in winter over most regions of China. The correction amount exhibits a remarkable seasonal and spatial variation: the maximum value in winter appears in southeast China; in spring another maximum value center arises in southwest China, which becomes obviously prominent in summer; while in autumn, the maximum center begins to shift back to southeast China, which is finally formed in winter. The spatial pattern of correction percentage for wetting loss is similar to that for trace precipitation: the correction percentage is higher in winter, lower in summer, and the percentage value declines from northwest to southeast China. Additionally, except for extremely arid areas in northwest China, the correction percentage for wetting loss is generally larger than for trace precipitation, indicating that wetting loss correction is more important than trace precipitation correction in most humid regions of China. In extremely arid areas, trace precipitation correction is as important as wetting loss correction.

Fig. 6.
Fig. 6.

As in Fig. 5, but for wetting losses.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

The correction amount for wind-induced loss is large in wet regions (southeast China) and small in arid regions (northwest China) (Fig. 7). This spatial distribution pattern is quite unambiguous in the dry season (DJF and MAM), which precisely reflects the monsoon rainfall characteristics. In the wet season (JJA and SON), extremely high correction for wind-induced loss is found in coastal regions where typhoons often occur. It is found that large correction percentage for wind-induced undercatch occurs in the Tibetan Plateau, north China, and northeast China, where winter climate is cold and snowy or dry, and the wind speed is large (Ding et al. 2013). At the same time, the mean correction amount for wind-induced error in period of 1961–2013 was not as large as that during 1951–98 compared to the results of Ye et al. (2004). This is probably due to the gradual reduction of annual mean wind speed on precipitation days over most regions of China (Fig. 8). The significant decline in near-surface wind speed in recent decades has been reported in previous studies (e.g., Ren et al. 2005; Jiang et al. 2009, 2013).

Fig. 7.
Fig. 7.

As in Fig. 5, but for wind-induced undercatch.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

Fig. 8.
Fig. 8.

The per 10 a trend of annual mean wind speed for precipitation days [m s−1 (10 yr)−1] for 1961–2013 over mainland China, and the regions having significant changes at 0.05 confidence level are stippled.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

4. Effect of bias correction on precipitation climatology

As shown in Fig. 9, the distribution pattern of bias-corrected annual precipitation is a gradual reduction from southeast to northwest China, which has not changed much compared to that of unadjusted precipitation (figure not shown here). However, annual precipitation increased over China after bias correction to varying degrees. The area with precipitation less than 50 mm is reduced, and the 400, 800, and 1600 mm contours are extended northwestward (Fig. 9b). The spatial pattern of bias-corrected precipitation intensity (Fig. 9c) is quite similar to that of the annual precipitation amount, featured by the heavy precipitation often occurring in southeast China and weak precipitation in arid regions. After bias correction, precipitation intensity increased to varying degrees (Fig. 9d). Figure 9e indicates that the impact of bias correction on precipitation intensity is greatest in dry regions where the corrected precipitation intensity has increased by 35%.

Fig. 9.
Fig. 9.

(a) Climatology of bias corrected annual precipitation (mm), (b) specific contours of annual precipitation (mm), (c) climatology of precipitation intensity (mm day−1), (d) specific contours of precipitation intensity (mm day−1), and (e) correction percentage of precipitation intensity (%) for 1961–2013 over mainland China. Blue (red) lines in (b),(d) represent climatologies with (without) bias correction.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

The value of the DJF correction percentage is clearly higher than those of the other three seasons, and the JJA value is the lowest (Fig. 10). This means that bias correction poses a greater influence in the cold and dry season over China. Because the catch ratio of snowfall is less than that of rainfall, the bias correction of precipitation has the greatest impact over northwest and northeast China in winter, with the highest DJF correction percentage reaching more than 80%. The climatology of MAM precipitation shows the highest correction percentage (more than 60%) in the Tibetan region. In summer and autumn, bias correction has a greater influence in arid regions where the highest value is more than 40%.

Fig. 10.
Fig. 10.

Climatology of (left) seasonal [(a)–(c) for DJF; (d)–(f) for MAM; (g)–(i) for JJA; (j)–(l) for SON] precipitation after bias correction (mm), (middle) specific contours of seasonal precipitation (mm) with (blue lines) and without (red lines) bias correction, and (right) the corresponding correction percentage (%) for 1961–2013 over mainland China.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

From Fig. 11, we find that light rain is most likely to occur in southwest China where the annual light rain amount is more than 400 mm. In addition, light rain intensity shows a higher value (more than 2.1 mm day−1) in southeast China and some regions of southwest China. After bias correction, the light rain amount and intensity clearly increased over the whole of China. Figure 11e indicates that the bias correction has the greatest influence on arid regions, and the light rain intensity after bias correction could increase by 40% in some extremely arid regions.

Fig. 11.
Fig. 11.

As in Fig. 9, but for annual light rain.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

The bias-corrected rainstorm amount and intensity gradually decrease from southeast to northwest China (Fig. 12). In general, the correction percentage increases gradually from southeast to northwest China, and the correction percentage has larger differences in arid regions than wet regions. The maximum correction percentage could reach as high as 18% in some regions of northwest and northeast China. Though rainstorms are rare in arid regions, the actual amount and intensity of rainstorm events is really higher than the observed value, especially because rainstorms are often accompanied by strong winds. Please note that we used the unadjusted precipitation classification for the adjusted precipitation, so the precipitation events after correction that reached the light rain or rainstorm level were not included in Figs. 11 and 12. Therefore, besides the stronger rainstorm described in Fig. 12e, we should also pay attention to those new emerging rainstorm events due to the correction (Fig. 13). During the period of 1961–2013, the total number of actual rainstorm events could be 90 days more than that of observed rainstorm events (about 500 days) in some coastal areas of China.

Fig. 12.
Fig. 12.

As in Fig. 9, but for annual rainstorm.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

Fig. 13.
Fig. 13.

The total number of increased rainstorm events (days) due to bias correction for 1961–2013 over mainland China.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

The bias correction has a larger influence on snowfall (Fig. 14) than on unclassified precipitation (Fig. 9) and rainfall. This is due to the gauge undercatch of snowfall, which is much higher than rainfall undercatch for the same wind speed. After bias correction, the annual snowfall amount is larger in wet and cold regions (northern part of northwest and northeast China) as well as high-altitude regions (Tibetan Plateau) (Fig. 14a), and the area with more than 10 and 80 mm of annual snowfall is much larger than that derived from measured snowfall data (Fig. 14b). The spatial pattern of bias-corrected snowfall intensity is quite different from that of the snowfall amount (Figs. 14a,c). Stronger snowfall is found in the Yangtze and Huaihe River basins (25°–35°N, 110°–120°E). This is due to the abundant moisture content there, though the annual total snowfall is less than that in the far north (Liu et al. 2012; Zhang et al. 2015). In addition, the corresponding influence of bias correction is much greater in dry and cold regions than that in wet and warm regions (Fig. 14e). The maximum correction percentage of snowfall is up to 50% or more, which is much higher than that of precipitation, including solid state and rainfall.

Fig. 14.
Fig. 14.

As in Fig. 9, but for annual snowfall.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

5. Discussion

Because of the diversity of observational environments and reference gauge types, the bias-correction methods could be different. For example, Sun et al. (2013) developed the correction factors of wind-induced errors for snowfall by comparing parallel observation data from three reference gauges in northeast China during 1992–98, which is presented below:
CRs(snow)=exp(0.12Ws)×100,
where CRs is catch ratios for snow, and Ws is the daily 10-m mean wind speed. Comparing formula (6) and (2), the functions of wind speed and catch ratio for snow developed by Yang et al. (1991) and Sun et al. (2013) are similar but with different factors (i.e., −0.056 vs −0.12). This indicates the robust relationship between the wind speed and the catch ratios for snow characterized by different coefficients derived from different experiments. Different correction factors may lead to some uncertainties, however. Hence, in this section, we evaluated the uncertainty in different correction methods by applying the two existing correction functions to observed snow events at five stations (ID 50564, 51358, 53594, 55248, and 57279). Two of the seven representative stations (ID 56856 and 59271) were excluded because of no snowfall detected. In the followings, the correction method used in this study is called method I, and the correction method developed by Sun et al. (2013) is called method II.

During the past 53 years, a total of 3614, 2693, 1148, 1664, and 446 snow events were detected at station 50564, 51358, 53594, 55248, and 57279, respectively. For station 50564, the average of corrected daily snowfall is 1.64 and 1.92 mm by employing methods I and II, respectively. The average of corrected snowfall for other 4 stations can also be seen in Table 2. The corrected snowfall by using method II is more than that by method I, which is clear at all selected stations (Table 2). The absolute difference at station 57279 is the largest (i.e., 0.46 mm), while stations 51358 and 53594 register the smallest difference (i.e., 0.13 mm). Figure 15 presents the corrected snowfall by using method I, as well as the absolute difference between method II and I for each snow event. The absolute difference is quite small in most snow events; however, it reaches 10 mm or more for a few of events at station 50564. Besides, the absolute difference of corrected daily snowfall between method II and I divided by corrected snowfall by applying method I is termed as the relative difference. On average, the smallest relative difference appears at station 51358 (i.e., 5.5%), while station 57279 sees the largest relative difference (i.e., 13.9%). This indicates that the corrected snowfall by using method II is more than that by method I, and the excess is less than 10% in general. Therefore, using different methods for snowfall correction could produce some uncertainty, but the difference of the results using the two methods is less than 10%. Because the data in reference gauge are unavailable for the time being, we are not able to develop new correction method for other precipitation types (i.e., rainfall and mixed precipitation) in this study. Compared with method II, method I developed by Yang et al. (1991) covers a variety of precipitation types (i.e., rainfall, snowfall and mixed). Besides, the correction of wind-induced errors for snowfall in method I is more conservative than that in method II. In view of the above reasons, we adopted the method I in this study.

Table 2.

The average of the corrected snowfall (mm) by two correction methods, and the relative difference between two correction methods (%) for all corrected snow events at five stations.

Table 2.
Fig. 15.
Fig. 15.

The corrected snowfall (mm, red lines) by using method I and the absolute difference between two correction methods (mm, blue lines) for each snow event at station (a) 50564, (b) 51358, (c) 53594, (d) 55248, and (e) 57279.

Citation: Journal of Applied Meteorology and Climatology 58, 10; 10.1175/JAMC-D-19-0049.1

The corrections for wind-induced undercatch as reported in this study may have other uncertainties of the same magnitude as caused by the applied correction method, because wind speed can vary during 24-h period of a day, and the daily mean wind speed in some cases may not represent the simultaneous wind speed with the occurrence of precipitation. Hourly data are preferred, but high-quality hourly precipitation and wind speed data are unavailable at present. Besides, the current methods [e.g., methods developed by Yang et al. (1991) and Sun et al. (2013)] were developed by using daily precipitation data from reference gauges, and the data of hourly reference gauge measurements have not been applied yet for developing correction method. A verification of the effectiveness of the corrections using hourly data should be carried out on the premise of available hourly data and mature methods in the future. One important aim of SPICE is to improve the understanding and reliability of solid precipitation measurements using automatic gauges, and a series of methods for solid precipitation measurement and adjustment in China are being studied by Chinese scientists. In the future, precipitation bias correction in the country will be improved by applying the new methods.

6. Conclusions

Rain gauge estimates of precipitation suffer from a range of biases due to several factors. This results in an underestimate of the actual precipitation. Long-term daily data of precipitation, precipitation type, mean temperature and mean wind speed from 1961 to 2013 at 2379 national stations over China were used to conduct a bias correction for daily gauge records. Trace precipitation, wetting losses, and wind-induced undercatch were estimated for each precipitation day with different precipitation types, namely rainfall, snowfall, and mixed rainfall and snowfall. Based on this, a more reliable daily precipitation dataset was generated, and the influences of bias correction on the climatology of precipitation, snowfall, and extreme precipitation were also analyzed.

We found that the corrections of trace precipitation and wetting losses were more important in arid than wet regions. The correction percentage decreased from northwest (e.g., 35% for annual precipitation) to southeast (e.g., 10% for annual precipitation) China, and was higher in cold seasons and lower in warm seasons. Wetting loss correction was greater than trace precipitation in most wet regions, while trace precipitation correction had the equivalent or greater influence than wetting loss correction in extremely arid areas. It is worth noting that, unlike trace precipitation and wetting losses, larger correction percentages for wind-induced error were found in cold or arid regions (the Tibetan Plateau and north and northeast China), as well as in high wind speed areas. The mean correction amount for wind-induced error in period of 1961–2013 was not as large as that during 1951–98 probably due to the gradual reduction of near-surface wind speed on precipitation days over most regions of China.

Generally, the climatology of bias-corrected annual precipitation as well as precipitation intensity increased to varying degrees. The 400, 800, and 1600 mm annual precipitation contours extended northwestward. The correction percentage of both annual precipitation amount and intensity was smallest in south part of China with a minimum percentage value less than 10%, while it was the greatest in arid regions with a maximum percentage value up to 35% or more. In addition, the correction percentage in DJF is clearly higher than that in other seasons due to less precipitation and a higher proportion of snowfall. Gauge undercatch of snowfall is much higher than that of rainfall for the same wind speed, hence the correction percentage of snowfall is overall higher than precipitation (including solid and liquid) or rainfall (liquid) with a maximum percentage value up to 50% or more. Due to the bias correction, the correction percentage of light rain intensity increased by more than 40% in extremely arid regions. Though rainstorm events are rare in arid regions, the corrected amount and intensity of rainstorms is higher than the observed value, and the maximum correction percentage for rainstorm intensity could reach 18% or more. Remarkably, the total number of actual rainstorm events during the past 53 years could be 90 days more than the observed rainstorm events (about 500 days) in some coastal areas of China.

Moreover, bearing in mind the results derived from our analysis, it can be concluded that bias correction is truly needed to obtain accurate estimates of precipitation records. Bias-corrected precipitation, especially snowfall, as well as rainstorms increased to different extents. The relative adjustments were greater in cold seasons (10%–80% in DJF) than warm seasons (10%–40% in JJA) and in arid regions (e.g., more than 50% in DJF) than wet areas (e.g., less than 20% in DJF). Therefore, the corrected daily precipitation data of a high-density observational network will substantially improve the accuracy and reliability of large-scale precipitation analyses over China, including studies and assessments of precipitation climatology and climate variability.

Last but not least, some uncertainties in current study of bias-correction method for precipitation exist. The correction in this study may have been conservative due to the application of a relatively low correction factor and the daily mean near-surface wind speed data. The diversity of observational environments and reference gauge types may have been the other potential causes for the uncertainty of the correction. Further improvements will be achieved in the future when high-quality subdaily data are available and the relevant comparative experiments of precipitation measurement are conducted.

Acknowledgments

We are grateful to the two anonymous reviewers for their valuable comments and helpful advice. This work was sponsored by the National Key Research and Development Program of China (Grants 2016YFA0600301 and 2018YFA0605603), the China Meteorological Administration Special Public Welfare Research Fund (Grant GYHY201406017), and the China Meteorological Administration Special Foundation for Climate Change (Grants CCSF201803 and CCSF201917).

REFERENCES

  • Aaltonen, A., E. Elomaa, A. Tuominen, and P. Valkovuori, 1993: Measurement of precipitation. Symposium on Precipitation and Evaporation, B. Sevruk and M. Lapin, Eds., Slovak Hydrometeorological Institute, 4246.

    • Search Google Scholar
    • Export Citation
  • Ding, Y., D. Yang, B. Ye, and N. Wang, 2007: Effects of bias correction on precipitation trend over China. J. Geophys. Res., 112, D13116, https://doi.org/10.1029/2006jd007938.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ding, Y. H., S. W. Wang, J. Y. Zheng, H. J. Wang, and X. Q. Yang, 2013: Climate of China (in Chinese). Science Press, 557 pp.

  • Duchon, C. E., and C. J. Biddle, 2010: Undercatch of tipping-bucket gauges in high rain rate events. Adv. Geosci., 25, 1115, https://doi.org/10.5194/adgeo-25-11-2010.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Folland, C. K., 1988: Numerical models of the raingauge exposure problem, field experiments and an improved collector design. Quart. J. Roy. Meteor. Soc., 114, 14851516, https://doi.org/10.1002/qj.49711448407.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Goodison, B. E., P. Y. T. Louie, and D. Yang, 1998: WMO solid precipitation measurement intercomparison: Final report. WMO Tech. Doc. 872, 212 pp.

    • Search Google Scholar
    • Export Citation
  • Jiang, Y., Y. Luo, and Z. C. Zhao, 2009: Evaluation of wind speeds in China as simulated by global climate models (in Chinese with English abstract). Acta Meteor. Sin., 67, 923934, https://doi.org/10.11676/qxxb2009.090.

    • Search Google Scholar
    • Export Citation
  • Jiang, Y., Y. Luo, and Z. C. Zhao, 2013: Maximum wind speed changes over China. Acta Meteor. Sin., 27, 6374, https://doi.org/10.1007/s13351-013-0107-x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kochendorfer, J., and Coauthors, 2017: Errors and adjustments for single-Alter shielded and unshielded weighing gauge precipitation measurements from WMO-SPICE. Hydrol. Earth Syst. Sci., 21, 35253542, https://doi.org/10.5194/hess-21-3525-2017.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kurtyka, J. C., 1953: Precipitation measurements study. Department of Registration and Education and States Water Survey Division, Investigation 20, 163 pp.

    • Search Google Scholar
    • Export Citation
  • Larson, L. W., 1971: Precipitation and its measurement: A state of the art. University of Wyoming Water Resources Research Institute, Water Resource Series 24, 74 pp., http://library.wrds.uwyo.edu/wrs/wrs-24/wrs-24-co.html.

  • Larson, L. W., and E. L. Peck, 1974: Accuracy of precipitation measurements for hydrologic modeling. Water Resour. Res., 10, 857863, https://doi.org/10.1029/WR010i004p00857.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, N., Y. Li, and N. Yao, 2018: Bias correction of the observed daily precipitation and redivision of climatic zones in China. Int. J. Climatol., 38, 33693387, https://doi.org/10.1002/joc.5506.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liu, Y. L., G. Y. Ren, and H. M. Yu, 2012: Climatology of snow in China (in Chinese with English abstract). Sci. Geogr. Sin., 32, 11761185.

    • Search Google Scholar
    • Export Citation
  • Nitu, R., and Coauthors, 2018: WMO Solid Precipitation Intercomparison Experiment (SPICE) (2012–2015). Instruments and Observing Methods Rep. 131, 452 pp., https://library.wmo.int/doc_num.php?explnum_id=5686.

  • Pan, X. C., and Coauthors, 2016: Bias corrections of precipitation measurements across experimental sites in different ecoclimatic regions of western Canada. Cryosphere, 10, 2347, https://doi.org/10.5194/tc-2016-122.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rasmussen, R., and Coauthors, 2012: How well are we measuring snow: The NOAA/FAA/NCAR Winter Precipitation Test Bed. Bull. Amer. Meteor. Soc., 93, 811829, https://doi.org/10.1175/BAMS-D-11-00052.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ren, G. Y., J. Guo, M. Z. Xu, Z. Y. Chu, L. Zhang, X. K. Zou, Q. Q. Li, and X. N. Liu, 2005: Climate changes of China’s mainland over the past half century (in Chinese with English abstract). Acta Meteor. Sin., 63, 942956.

    • Search Google Scholar
    • Export Citation
  • Ren, G. Y., Y. J. Zhan, Y. Y. Ren, Y. Chen, T. Wang, Y. J. Liu, and X. B. Sun, 2015a: Spatial and temporal patterns of precipitation variability over mainland China: I: Climatology (in Chinese with English abstract). Adv. Water Sci., 26, 299310.

    • Search Google Scholar
    • Export Citation
  • Ren, G. Y., Y. Y. Ren, Y. J. Zhan, X. B. Sun, Y. J. Liu, Y. Chen, and T. Wang, 2015b: Spatial and temporal patterns of precipitation variability over mainland China: II: Recent trends (in Chinese with English abstract). Adv. Water Sci., 26, 451156.

    • Search Google Scholar
    • Export Citation
  • Ren, G. Y., and Coauthors, 2016: Spatial and temporal patterns of precipitation variability over mainland China: III: Causes for recent trends (in Chinese with English abstract). Adv. Water Sci., 27, 327348.

    • Search Google Scholar
    • Export Citation
  • Ren, Z. H., and M. Q. Li, 2007: Errors and correction of precipitation measurements in China. Adv. Atmos. Sci., 24, 449458, https://doi.org/10.1007/s00376-007-0449-3.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ren, Z. H., G. L. Wang, F. L. Zou, and H. Z. Zhang, 2003: The research of precipitation measurement errors in China (in Chinese with English abstract). Acta Meteor. Sin., 61, 621627.

    • Search Google Scholar
    • Export Citation
  • Scaff, L., D. Q. Yang, Y. P. Li, and E. Mekis, 2015: Inconsistency in precipitation measurements across the Alaska–Yukon border. Cryosphere, 9, 24172428, https://doi.org/10.5194/tc-9-2417-2015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sevruk, B., 1982: Method of correction for systematic error in point precipitation measurement for operational use. WMO Tech. Doc. 589, 91 pp.

  • Sevruk, B., and W. R. Hamon, 1984: International comparison of national precipitation gauges with a reference pit gauge. WMO Instrument and Observing Methods Rep. 17, 111 pp.

    • Search Google Scholar
    • Export Citation
  • Sevruk, B., and L. Zahlavova, 1994: Classification system of precipitation gauge site exposure: Evaluation and application. Int. J. Climatol., 14, 681689, https://doi.org/10.1002/joc.3370140607.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sun, X. B., G. Y. Ren, Z. H. Ren, and Z. C. Shen, 2013: Effects of wind-induced errors on winter snowfall and its trends (in Chinese with English abstract). Climatic Environ. Res., 18, 178186.

    • Search Google Scholar
    • Export Citation
  • Westra, S., and Coauthors, 2014: Future changes to the intensity and frequency of short-duration extreme rainfall. Rev. Geophys., 52, 522555, https://doi.org/10.1002/2014RG000464.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yang, D., Y. F. Shi, E. S. Kang, Y. S. Zhang, and X. Y. Yang, 1991: Results of solid precipitation measurement intercomparison in the Alpine area of Urumqi River Basin. Chin. Sci. Bull., 36, 11051109.

    • Search Google Scholar
    • Export Citation
  • Yang, D., and Coauthors, 1995: Accuracy of Tretyakov precipitation gauge: Result of WMO intercomparison. Hydrol. Processes, 9, 877895, https://doi.org/10.1002/hyp.3360090805.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yang, D., B. E. Goodison, J. R. Metcalfe, V. S. Golubev, R. Bates, T. Pangburn, and C. L. Hanson, 1998: Accuracy of NWS 8 standard nonrecording precipitation gauge: Results and application of WMO intercomparison. J. Atmos. Oceanic Technol., 30, 5768, https://doi.org/10.1175/1520-0426(1998)015<0054:AONSNP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Yang, D., and Coauthors, 2001: Compatibility evaluation of national precipitation gauge measurements. J. Geophys. Res., 106, 14811492, https://doi.org/10.1029/2000JD900612.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yang, D., D. Kane, Z. Zhang, D. Legates, and B. Goodison, 2005: Bias-corrections of long-term (1973–2004) daily precipitation data over the northern regions. Geophys. Res. Lett., 32, L19501, https://doi.org/10.1029/2005GL024057.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yao, N., Y. Li, N. Li, D. Yang, and O. Ayantobo, 2018: Bias correction of precipitation data and its effects on aridity and drought assessment in China over 1961–2015. Sci. Total Environ., 639, 10151027, https://doi.org/10.1016/j.scitotenv.2018.05.243.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ye, B., D. Yang, Y. Ding, T. Han, and T. Koike, 2004: A bias corrected precipitation climatology for China. J. Hydrometeor., 5, 11471160, https://doi.org/10.1175/JHM-366.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ye, B., D. Yang, and L. Ma, 2012: Effect of precipitation bias-correction on water budget calculation in Upper Yellow River, China. Environ. Res. Lett., 7, 025201, https://doi.org/10.1088/1748-9326/7/2/025201.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, Z. F., S. Xi, N. Liu, Y. Yu, S. H. Fan, and M. N. Feng, 2015: Snowfall change characteristics in China from 1961 to 2012 (in Chinese with English abstract). Ziyuan Kexue, 37, 17651773, http://www.resci.cn/CN/Y2015/V37/I9/1765.

    • Search Google Scholar
    • Export Citation
Save