a. Wetting loss ΔP_w

Wetting losses are due to precipitation retaining or sticking to the sides of the gauge and cannot be poured out and measured in precipitation observation. Wetting losses are gauge specific and vary by precipitation type and the number of times the gauge is emptied. According to the methods of observation in China (Chinese Meteorological Administration 1979), the Chinese gauge is used with a funnel and a glass bottle (container) for rain measurements. For solid precipitation measurements the funnel and container are removed. Wetting loss experiments in China show that the average wetting loss of the CSPG per observation was 0.23 mm for rainfall measurement, and 0.30 and 0.29 mm for snow and mixed precipitation, respectively (Yang et al. 1991; Yang 1988). Precipitation is measured twice a day over China, at 0800 and 2000 LT (local, Beijing, time). To be conservative, we correct wetting losses once a day.

b. Trace precipitation ΔP_t

A precipitation event of less than 0.10 mm is beyond the resolution of the CSPG and is recorded as trace amount of precipitation. Trace events are counted as precipitation days, but quantitatively they are treated as zero amounts. Considering the wetting loss, the gauge cannot detect the precipitation events that are less than the wetting loss. Precipitation observations in China show that sometimes two trace precipitation events are reported in a single trace precipitation day; thus, it is not unreasonable to assume that a trace event could be a small precipitation amount of 0.05 to 0.15 mm. To be conservative, trace precipitation was corrected on a daily basis in this study; for example, for any given trace day, regardless of the number of the trace observations reported, a value of 0.10 mm was assigned and added to the monthly total.

c. Evaporation loss ΔP_e

Evaporation loss is the water lost by evaporation of the contents in the gauge before the observation is made. Comprehensive assessment of evaporation losses of many national gauges was conducted during the WMO intercomparison project, and the results indicated that average daily losses varied by gauge type and time of the year (Aaltonen et al. 1993; Goodison et al. 1998). Evaporation losses are strongly dependent on weather condition and methods of observations, such as the number of times of daily observations (Sevruk 1982). For the CSPG tested in western China, the annual evaporation loss was estimated to be 18.6 mm (4.4%) in high mountains (3640 m msl with about 300 mm of annual mean potential evaporation and 421 mm of annual precipitation) and 4.4 mm (1.6%) in low elevations (918 m msl with about 2000 mm of annual mean potential evaporation and 277 mm of annual precipitation) (Yang 1988). The funnel and container can greatly reduce evaporation losses. The evaporation losses usually are less than other losses such as wetting loss, trace precipitation, and wind-induced error even in the arid region with high potential evaporation when the funnel and container were used with the gauge in the raining season. The higher evaporation losses in high-mountain regions are due to the CSPG used without a funnel and a container in mountain regions for observations of wet snowfall in the summer season. In these cases, the evaporation loss may be greater than the trace loss.

Daily variation and seasonal change in evaporation losses are great and can be very site dependent. It is difficult to estimate with any confidence the daily evaporation losses at regional climate station networks by using the average evaporation amount obtained from other experimental sites. Therefore evaporation loss in China was not corrected in this study. The annual evaporation loss in China is expected to be small due to the use of a funnel and a container in the high-evaporation period. Neglecting this error in this study will therefore not significantly affect the results of the bias corrections.

d. Wind-induced gauge undercatch

The wind-induced errors are caused by the wind field deformation over the gauge orifice (Sevruk and Hamon 1984). The catch ratio of snow is less than that of rain because the snowfall can be more easily affected by wind than rainfall (Larson and Peck 1974; Goodison et al. 1998; Yang et al. 1995, 1998b). To correct for wind-induced gauge undercatch, wind speed is required. Wind measurements were made at the standard height of 10 m at the observational network in China, and these data were available for this study. It has been suggested that gauge exposure should be considered when reducing wind from the standard height to the gauge level (Sevruk 1982). Gauge exposure depends on the average vertical angle of obstacles around the gauge; it can be directly measured or estimated by a classification system based on metadata archives (Sevruk 1982). The station metadata are not available for this work or most global-scale bias corrections. Site exposure was not accounted for in wind speed estimates at the gauge height. This may introduce some uncertainty in the estimation of gauge catch efficiency.

It has been well documented that for the same wind speed, gauge undercatch of snow is much higher than for rain (Larson and Peck 1974; Goodison et al. 1998; Yang et al. 1995, 1998b). Classifying the type of precipitation is therefore necessary in order to apply the best wind-loss correction. In this work, information of precipitation type was available in the metadata before 1980. After 1980, precipitation type is classified by daily air temperature. The temperature criteria were set at −2° and +2°C, that is, snow and rain for daily temperature below −2° and above +2°C, respectively, and mixed precipitation when daily temperature was in between −2° and +2°C (Yang et al. 2001).

In the WMO intercomparison, catch ratio was defined as the ratio of the amount of precipitation caught by a gauge (including the recorded amount and wetting loss) to the true precipitation (Goodison et al. 1998). Based on the experimental observations in the Urumqi River basin, relation of catch ratio as a function of wind speed has been developed for the CSPG (Yang et al. 1991). It was found in the WMO experiment that wind speed was the most important factor determining gauge catch when precipitation was classified into snow, mixed precipitation, and rain. The results of daily gauge catch ratio (CR; percent) versus daily mean wind speed (Ws; meters per second) at 10-m height are presented below for snow and rain precipitation:

View Expanded

For mixed precipitation, the CR is estimated by a liner relationship of snow and rain percentages determined by daily temperature (T_d):

View Expanded

Once daily wind speed at the 10-m heights was determined, the daily CR for the CSPG was calculated using the regression equations (2)–(4) for snow, rain, or mixed precipitation. The daily catch ratio can be determined according to the precipitation type determined by daily temperature or direct observations. The wind-loss correction coefficient (K) was calculated as K = 1/CR. The daily corrected precipitation amount can be obtained from the daily precipitation data and daily wind-loss coefficient:

View Expanded

a. Monthly bias correction

The monthly bias correction is the total amount of the daily correction for each month. They vary greatly over China because of different climate characteristics, particularly the amount and type of precipitation and wind conditions. To demonstrate the seasonal cycle of the bias corrections and the associated changes in precipitation amount, monthly trace amounts, wetting loss, and wind-induced errors are presented here for January, April, July, and October.

The monthly trace amounts are directly calculated from data of the monthly trace precipitation days. Monthly mean trace amounts are small, usually less than 1 mm over China (Fig. 2). Seasonal and spatial distributions of trace amounts and the correction factors (CFs; ratio of corrected amount to measured amount of precipitation) are different because of the precipitation variation. Generally, the CFs decline from northwest China to southeast China in all months, and the mean CFs are higher in winter months and lower in the summer season. For instance, the CF in January is usually less than 40% in north and west China and as high as 200% in some areas over southwest China because of variability of precipitation. On the other hand, the CF in July is usually less than 4%, with the maximum of 32% in northwest China. The CF of high precipitation in southeast China is less than 1% in every month. The areas with CF less than 1% are very large in July and small in January, indicating that trace corrections are more important in the winter season.

The monthly wetting loss is the sum of daily wetting loss determined by the daily and precipitation type. The monthly mean wetting loss amounts are usually less than 8 mm over China (Fig. 3). In northwest China, the wetting loss corrections are small in all months, that is, less than 2 mm in January, April, and October, and less than 3 mm in July. High wetting losses (over 4 mm) are found in southeast China in January and April and in middle regions of south China in July and October.

The mean monthly CFs for wetting loss range from 0.8% to 120% over China (Fig. 3). The mean CF is large in winter months and small in the summer season. The CFs usually are less than 80% in north and west China. There are several high CF centers in west China that correspond with low precipitation and high trace precipitation corrections. Usually, the CFs are less than 20% in southeast China. In July, the CF is usually less than 15% in northwest China and less than 3% in southeast China of high-precipitation regions. The CFs in April and October are moderate, between those for January and July. Generally, the CFs for wetting losses are larger than those for trace correction, except for some extreme low-precipitation regions.

The monthly mean wind-induced errors vary from 0 to 120 mm over China during 1951–98. Similar to the monthly precipitation, the spatial distribution of monthly mean wind-induced errors are characterized by low corrections in northwest China and high corrections in southeast China (Fig. 4). For instance, in northwest China, the wind-induced errors are less than 1, 3, 5, and 2 mm in January, April, July, and October, respectively. However, over southeast China, the errors are greater, that is, 4, 10, 15, and 5 mm for January, April, July, and October, respectively.

The monthly mean CFs for the wind-induced errors show very similar spatial patterns between months due to similar patterns of monthly mean daily wind speed for precipitation days. They generally correspond to annual mean wind speed for precipitation days and are enhanced by snow percentage due to difference in gauge catch of snow and rain. There is a high CF region in west-central China, in Qinhai-Xizang Plateau. Other high CF areas are in north China where the climate is cold and dry. The monthly CF of wind-induced error is usually 1%–40%, 3%–39%, 2%–31%, and 1%–40% in January, April, July, and October, respectively. The difference is due to the variations in wind speeds and snow percentage.

We also calculated the monthly total correction and its CF for wetting loss, trace, and wind-induced error. Figure 5 presents the results for January, April, July, and October and shows regional patterns over China. The monthly total corrections decreased from northwest China to southeast China. In northwest China, the total corrections are less than 4, 5, 10, and 5 mm in January, April, July, and October, respectively. However, over southeast China, the errors are greater, that is, 4, 10, 15, and 10 mm for January, April, July, and October, respectively. The CF increases from south to north in east China and has high centers around midwest China because of dry and cold climate at the high elevations.

To clearly show the difference between months, several stations have been chosen to represent the regional characteristics. The stations include the northern region with high snow percentage (Fig. 6a), the south region without any snow (Fig. 6b), high elevation with wet snow sometimes in summer months (Fig. 6c), and the extreme dry region (Fig. 6d). Monthly snow, rain, and mixed precipitation, corrected amounts (Fig. 6, left), correction factors (Fig. 6, middle), and wind speed for precipitation days (Fig. 6, right) are shown for the stations. The important common features of the bias corrections can be summarized as follows: 1) In each month, the absolute monthly amounts of wetting losses and wind-induced errors are usually greater than trace precipitation adjustments except for the extreme dry conditions. The trace precipitation events are important in the months with precipitation less than 0.5 mm. 2) Wind-induced errors are generally greater than wetting loss for the months with more than 15–20 mm of precipitation. 3) A clear seasonal variation of the monthly CF: high CF values for the cold season and low CF values for rain in the warm season. The CF in winter in north and west China is higher than in southeast China because of the higher wind loss for snow than for rain and due to the smaller amount of absolute precipitation in the cold season in north and west China. The CF seasonal variation at the stations with high precipitation (>1000 mm yr⁻¹) is small (Fig. 6b).

b. Yearly bias correction

Based on the monthly results, the annual bias correction has been determined and summarized in Fig. 7. The averaged yearly correction for trace precipitation varied from 1.4 to 10 mm with an overall mean of 4.9 mm at the 710 stations, or about 0.1%–26% with an average of 1.4% of the gauge-measured annual precipitation over China. Correction for trace precipitation is thus important especially in northwest China because of low precipitation. The trace amount is several millimeters and the CFs are very small (less than 1%) in most of east China where high precipitation occurs. The yearly mean corrections for wetting loss ranged from 3.1 to 78 mm with an overall mean of 31 mm at the 710 stations, and the CFs are about 1.6%–23.7% with an average of 5.6% of the gauge-measured annual precipitation over China.

The patterns of the trace and wetting losses reflected the patterns of trace and measured precipitation days, respectively. The CF patterns are opposite to the patterns of the annual precipitation over China (Fig. 8). The CFs for both the trace and wetting losses decrease from northwest to southeast. It is important to note the annual corrected amounts of the wetting losses are generally larger than trace corrections except for some extremely dry climate regions in northwest China.

The annual average corrections for wind-induced errors have a much larger variation from 1.4 to 690 mm with an overall mean of 93 mm at the 710 stations over China. The amounts increased from less than 25 mm in northwest China to over 100 mm in southeast China. This difference from the annual precipitation patterns is due to variations in wind speed and the difference in the snow percentage. The annual CFs of the wind-induced errors are about 1.7%–32.1% with an average being 12.0% of the gauge-measured annual precipitation at the 710 stations over China. The mean annual total corrections ranged from 8 to 740 mm with an average of 130 mm at the 710 stations over China (Fig. 7). The amount varies from 8–50 mm in northwest China to 100–740 mm in southeast China. The amount is less then 100 mm in majority of China with annual precipitation less than 500–1000 mm. The mean annual CFs for total correction vary from 30%–62% in northwest China to 5%–15% in southeast China mainly due to the lower temperature, lower precipitation, higher snowfall proportion, and higher winds on precipitation days in the northern regions (Fig. 8). The overall mean annual CF at the 710 stations is 19%, which is higher than the global mean CF of 11.7% (Adam and Letternmaier 2003) and 11% (Legates and Willmott 1990).

Generally, the CF pattern for total correction is very similar to the CF pattern for the wind-induced error, although it has been enhanced by wetting loss and trace precipitation in west China, especially in the high CF regions in west-central China. The results show that the wind-induced errors are the largest bias, and trace and wetting losses are important particularly in the dry region of west China, that is, with yearly precipitation less than 250 mm. It is important to note that because of the effect of high elevation such as the Tianshan Mountains and Qinhai-Xizang Plateau in west China, the CF distribution over China shows not only poleward increase, which was found for the continental United States, Greenland, and northern Canada (Legates and DeLiberty 1993; Yang et al. 1998a, 1999b; Metcalfe et al. 1994), but also a westward increase due to the high-altitude region in west China. Similar results were found in the global monthly precipitation bias corrections (Adam and Lettenmaier 2003).

Maximum and minimum CF values are also provided in Fig. 9 to illustrate the interannual and interstation variation. The CF variation (5%–30% in east China and 15%–250% over the arid region in northwest China) is induced by year-to-year fluctuation of wind speed, frequency, and amounts of precipitation. This is particularly accentuated in west regions of China, where yearly precipitation is low (the annual total from 14 to 300 mm) and the trace and wetting loss are important components of the corrections. The annual CFs there can reach up to 60% or higher because of the low precipitation. The trace precipitation days are greater than the measurable precipitation days and mean daily precipitation is less than 3 mm. Similar results have been found in precipitation bias corrections for arctic regions such as northern Canada (Metcalfe and Goodison 1993) and Alaska (Yang et al. 1998a, 2001).

Observation data show that annual precipitation varies from 14 to 2800 mm over China (Fig. 8a). The long-term annual mean precipitation ranges from less 200 mm over northwest China to more than 1500 mm in the southern coastal regions, with mean snow percentage between 0% in the southeast regions to 46% in north and west region of China (Fig. 8c). Based on the bias-corrected results, yearly corrected precipitation maps were developed in this study (Fig. 10). Relative to the yearly precipitation map derived from the observed data (Fig. 8a), the bias-corrected yearly precipitation map (Fig. 10) shows much higher values over China. The area with less than 50 mm of precipitation is smaller, and the 500-m contour in the corrected precipitation map is similar to the 400-m contour in the gauge-measured precipitation map, although the precipitation distribution pattern has not changed much due to the bias corrections.