FIG. 3. Spatial distributions of the linear trends of (a) Rmax, (b) R>10mm, (c) R95p and (d) R<10mm over the Loess Plateau during the presummer season from 1979 to 2018.
The FDR-adjusted Mann-Kendall test is applied to detect the significance of the trend of the four presummer precipitation indices during 1979–2018 at each grid point over the LP. The black dots indicate the grid points with the trend passing the significance test at 90% confidence level. In general, the extreme precipitation increases in the northwest and decreases in the southeast of the LP (Fig. 3), which is basically consistent with the precvious studies (Zhao, Zhai et al. 2017, Zhao, Li et al. 2017). However, significant differences are found among the six UAs. Xi’an, Taiyuan and Luoyang are located in the Fenhe and Weihe River Basin in the southeastern LP, with the fastest urbanization development over the LP (Fig. 2). Under the climate background with decreasing precipitation, the extreme precipitation decreases more severely in Taiyuan UA and the rural areas. The rates of Rmax, R>10mm and R95p in Taiyuan UA are 0.2 mm·year −1, 0.4−0.6 mm·year−1 and 0.4−0.5 mm·year −1, respectively, which are lower than those in the rural areas. The most significant effect of urbanization is found in Taiyuan UA (Fig. 3c), where the decrease rates of R95p in the urban and the rural areas presents clearly different distributions. In obvious contrast with Taiyuan, Luoyang is under a climate background with increasing precipitation, and an obvious center with increasing extreme precipitation is found in Luoyang UA. The increase rates of Rmax, R>10mm and R95p in Luoyang UA are 0.3 mm·year −1, 0.7 mm·year −1 and 0.9 mm·year −1, respectively, which are higher than those in its rural region. Xi’an is different from the two above UAs. Xi’an UA is in a wet environment, but a dry core is formed in the UA center due to the urbanization effect, which is obviously reflected in the distribution of Rmax and R95p (Figs. 3a and 3c). The extreme precipitation rate decreases by 0.4−0.6 mm·year−1 in the center of Xi’an UA compared with that in the rural areas. Hohhot and Xining are two UAs in the northern and western parts of the LP, respectively, where the precipitation has increased significantly in recent years. In the past two decades, the urbanization of Xining has been relatively slow, but it is particularly sensitive to the urbanization as a plateau city. The extreme precipitation in Xining UA is more concentrated in the center, especially for Rmax and R>10mm. Hohhot has witnessed rapid urbanization in recent years, and its extreme precipitation characteristics are similar to those of Xining. The trend distribution of the number of light rain days is one of the most sensitive indicators for the urbanization effect. Different from the trend of extreme precipitation, the number of light rain days in the five UAs consistently decreases except for Luoyang, which corresponds to the change in the artificial impervious area (Figs. 2a and 4d).
The trends of extreme precipitation in Taiyuan, Luoyang, Hohhot and Xining UAs almost pass the significance test at 90% confidence level, while that in Xi’an does not pass the significance test, with opposite trend in the urban and the rural areas. The above analysis shows that the urbanization plays an accelerator role in the urban climate evolution, and the urban precipitation has become more intense and sensitive than that in the rural areas.