-
To explore the contribution of EP to total precipitation, we analyze the interannual variability in EP and total precipitation averaged for the chosen stations in NWC. Figure 2a shows the annual total and extreme precipitation amounts (EPA) in NWC. In addition to interannual fluctuations, the EPA show a significant increasing trend of 3.6 mm (10 yr)−1, while the total precipitation increases at a rate of 3.2 mm (10 yr)−1, which is relatively slower. In addition to EPA, the extreme precipitation days (EPD) also exhibit a significant increasing trend over nearly 56 years, especially against the background of decreasing total precipitation days (Fig. 2b). This result means that almost all of the increase in total precipitation occurs due to EP, and the precipitation in NWC is more likely to appear in the form of EP in recent years. This result is confirmed by the interannual variability in the contribution rate of EP to total precipitation (Figs. 2c, 2d), with an increase rate of 0.7% (10 yr)−1 for EPA and a rate of 0.4% (10 yr)−1 for EPD, with both values passing the significance test.
Figure 2. Time series of annual precipitation amounts (a, units: mm) and days (b, units: d) in NWC, the contribution rate (units: %) of EPA to total precipitation amounts (c), and the contribution rate of EPD to total precipitation days (d). The blue, red, and black dotted lines represent the linear trends in the total precipitation, EP and contribution rate of EP, respectively. The asterisk indicates that the trend is significant at the 0.05 level. The blue and red shading denotes the 95% confidence intervals for total precipitation and EP, respectively.
Furthermore, the variability in EP among the four seasons demonstrates upward trends in both EPA and EPD, passing the significance test in summer and winter (Table 1). The EP in summer grows fastest at rates of 2.64 mm (10yr)−1 in terms of amount and 0.13 d (10 yr)−1 in terms of days. Although the average EPA and EPD in winter are only 1.4 mm and 0.2 d, their increasing trend magnitudes reach 0.38 mm (10 yr)−1 and 0.04 d (10 yr)−1, respectively, which indicates that winter has the most obvious upward trend, passing the significance test at the 0.01 level.
season EPA EPD Average (mm) Trend [mm (10 yr−1)] Average (d) Trend [d (10 yr−1)] Spring 23.1 0.61 1.5 0.05 Summer 90.5 2.64* 4.9 0.13* Autumn 31.5 0.01 1.8 0.03 Winter 1.4 0.38** 0.2 0.04** Annual 146.5 3.64* 8.4 0.26** Table 1. Average EPA and EPD and their trends in each season in NWC during 1961–2016. Double asterisk and single asterisk represents significance at the 0.01 level and 0.05 levels, respectively.
To analyze the details of the distribution of precipitation changes in NWC, the variation trends in annual and seasonal precipitation are tested with the nonparametric Mann-Kendall method at each site. As presented in Figs. 3a and 3b, the total precipitation amounts and days all show similar distribution characteristics in NWC, rising in the central and western regions and falling in the eastern region. Notably, most of the decline in total precipitation days occurs in the eastern region. The EPA show a rising trend at 80% of the stations (Fig. 3c), concentrated in Xinjiang, Qinghai, western Gansu, and northern and southern Shaanxi Province or autonomous regions, with the largest trend of 26.5 mm (10 yr)−1 in Urumqi. Conversely, the decreasing trends generally range from
$ - $ 10.0 mm (10 yr)−1 to$ - $ 1.0 mm (10 yr)−1, with the largest trend occurring in Lintao [$ - $ 18.5 mm (10yr)−1]. The trends in EPD occur in areas that are similar to the spatial distribution of EPA (Fig. 3d). A slight difference is that there are more stations with a downward trend for EPD in the eastern region. Additionally, less than one-third of the site trends are significant, with greatest EPD clustering in Xinjiang and northern Qinghai. The trends in EP in Gansu, Ningxia, and Shaanxi largely fail to pass the significance test.Figure 3. Spatial distribution of the trends for total precipitation (a, b) and EP (c, d) in NWC. Dots, triangles, and squares represent the stations in the WZ, PZ, and MZ, respectively. A larger colored symbol represents significance at the 0.05 level. The black plus symbol indicates no trend.
As suggested above, total precipitation and EP have similar distributions for the most part, with increases in the central and western regions and decreases in the eastern region. The variability in EP plays a vital role in changes in total precipitation, and its effect reflects obvious regional differences. Undoubtedly, Xinjiang is the region with the most obvious increase in rates of change regardless of total precipitation or EP, whereas the variability in EP essentially determines the variation characteristics of total precipitation.
-
The results presented in the previous section show an increase of EP in most of NWC, but there are still great differences in variability among the sites. To understand the specific differences between the climate zones, we analyze the average EPA during the first 20 years and the last 20 years of the study period over different regions. For comparative convenience, the same EP threshold is used in comparing EP statistics for different 20-year periods. As shown in Fig. 4, comparing average EPA during1961–80 and during 1997–2016, EP significantly increases in NWC, the WZ, and the PZ, but there is a nonsignificant decrease in the MZ, which also confirms the previous results. The largest increase in EP occurs in the WZ, with the increases in EPA and EPD both exceeding 30%.
Figure 4. EPA (a, units: mm) and EPD (b, units: d) averaged for NWC and different climate zones during different periods. The asterisk indicates significance at the 0.05 level.
The variation trends in EP over the WZ, PZ, and MZ are discussed in this section. As shown in Figs. 3c and 3d, the overwhelming majority of stations have increasing trends in the WZ, with 49.3% and 53.6% of the stations showing statistical significance in EPA and EPD, respectively (Table omitted). Thus, the regional average trends in EPA and EPD in the WZ demonstrate significant positive trends with slopes of 6.2 mm (10 yr)−1 and 0.5 d (10 yr)−1, respectively (Figs. 5a, 5b). For the PZ, approximately 85% of the stations have positive trends. The regional positive trend in the PZ is slightly weaker than that in the WZ, but it also passes the significance test. For the MZ, a nonsignificant downward trend occurs for a large percentage of the stations, with 48.9% and 60.0% of stations showing nonsignificant trends in EPA and EPD, respectively. Regionally, the trends in EPA and EPD in the MZ are decreasing and fail the significance test (Figs. 5a, 5b).
Figure 5. Time series of EPA (a, units: mm) and EPD (b, units: d) in different climate zones. The red, yellow, and blue dotted lines represent the extreme precipitation linear trends in the WZ, PZ, and MZ, respectively. The red, yellow, and blue shading denotes the 95% confidence intervals for different climate zones. The asterisk indicates that the trend is significant at the 0.05 level.
The trends in different zones vary throughout the seasons, underlining the need to examine each season and region. Increasing trends are observed in all seasons in the WZ. Over 20% of the stations with significant positive trends in extreme summer and autumn precipitation are in the WZ (Table 2). The largest increase in EP occurs in summer, with rates reaching 3.1 mm (10 yr)−1 and 0.2 d (10 yr)−1 for EPA and EPD, respectively (Table omitted). For the PZ, the seasonal trend in summer is also larger than that in other seasons, with more than 20% of the stations exhibiting obvious increasing tendencies. On the other hand, the seasonal trends in EP in the MZ are very weak, with few sites showing significant increasing trends.
Index Season NWC WZ PZ MZ EPA Spring 9.1 13.0 13.8 0.0 Summer 21.0 29.0 27.6 4.4 Autumn 11.2 20.3 6.9 0.0 Winter 8.4 17.4 0.0 0.0 Annual 28.7 49.3 24.1 0.0 EPD Spring 7.0 8.7 13.8 0.0 Summer 18.2 27.5 20.7 2.2 Autumn 11.2 20.3 6.9 0.0 Winter 9.1 18.8 0.0 0.0 Annual 32.2 53.6 31.0 0.0 Table 2. Percentages of stations with significant increasing seasonal trends in EP in different climate zones.
-
The SDEP, the EDEP, and their changes have a great impact on the region. Advancements or delays in the SDEP and the EDEP affect regional climates and human life both directly and indirectly. Hence, on the basis of calculating the annual SDEP and EDEP and durations of extreme precipitation (DEP) at every site, we analyze the spatial distribution characteristics of their trends in NWC (Fig. 6). Based on Fig. 6a, the stations with early trends in SDEP are mainly distributed in most parts of Xinjiang, Qinghai, Ningxia, the northwestern Hexi Corridor, and northern Shaanxi, and the stations with delayed trends are concentrated in southern Gansu and central Shaanxi. The stations in northern Xinjiang show the largest advance with more than 10 d (10 yr)−1. In addition, a delay in EDEP is found at most sites in NWC (Fig. 6b), which is essentially the opposite to the spatial distribution of SDEP, with the largest delay of more than 9.0 d (10 yr)−1 found in Xinjiang. Due to advances in the SDEP and delays in the EDEP at most of the stations, the DEP exhibits more obvious extended trends in NWC, especially in Xinjiang and southern Qinghai, where the trends exceed 10 d (10 yr)−1 by a great percentage (Fig. 6c). At the same time, the DEPs in southern Gansu and Central Shaanxi decrease at a rate of more than 5 d (10 yr)−1.
Figure 6. Spatial distribution of the trends [units: d (10 yr)−1] in the SDEP (a), EDEP (b), and DEP (c). Dots, triangles, and squares represent the stations in the WZ, PZ, and MZ, respectively. A larger colored symbol represents significance at the 0.05 level. The black plus symbol indicates no trend.
Based on a comparison of the multiyear average SDEP in the three climate zones, EP usually occurs first in the WZ on 30 April, followed by the MZ and PZ (Table 3). The average end dates of EP for the WZ and PZ occur at almost the same time in middle September, and the latest EDEP occurs in late September in the MZ. However, in recent years, the SDEP, EDEP, and DEP for each climate zone have changed and differ considerably from their average values. For the largest advance in the SDEP [−3.6 d (10 yr)−1] and the largest postponement of the EDEP [5.9 d (10 yr)−1], the DEP in the WZ has the fastest increasing trend of 9.5 d (10 yr)−1. As a result, the average DEP in the WZ extends from 114.6 d in the 1960s to 150.4 d in recent years (2011–16). The regional SDEP and EDEP in the PZ also exhibit advancing and delayed trends, respectively. Hence, the DEP in the PZ increases by approximately 16.5 d from 1961 to 2016. On the other hand, due to the mixed pattern and heterogeneity in spatial distributions of the SDEP and EDEP, the variation in the DEP is not obvious in the MZ. Recently, the DEP values in the WZ and MZ are almost equal.
Average Trend SDEP (date) EDEP (date) DEP (d) SDEP [d (10 yr−1)] EDEP [d (10 yr−1)] DEP [d (10 yr−1)] WZ 30 Apr 12 Sept 134.7 −3.6** 5.9** 9.5** PZ 16 May 13 Sept 119.6 −2.9** 0.9 3.9** MZ 10 May 27 Sept 140.5 −0.7 0.1 0.9 Table 3. Average values of the SDEP, EDEP, and DEP and their trends in different climate zones. Double asterisk represents significance at the 0.01 level.
season | EPA | EPD | |||
Average (mm) | Trend [mm (10 yr−1)] | Average (d) | Trend [d (10 yr−1)] | ||
Spring | 23.1 | 0.61 | 1.5 | 0.05 | |
Summer | 90.5 | 2.64* | 4.9 | 0.13* | |
Autumn | 31.5 | 0.01 | 1.8 | 0.03 | |
Winter | 1.4 | 0.38** | 0.2 | 0.04** | |
Annual | 146.5 | 3.64* | 8.4 | 0.26** |