3.2. Trends variation of presummer precipitation in UAs
In order to further investigate the effect and contribution of climate change and urbanization on the presummer precipitation in different periods, we divide the five UAs into urban and rural areas separately according to the map of artificial impervious area cover. The UA area is covered by the artificial impermeable area, and the rural areas are mapped by using a buffer method. A buffer zone of 20 kilometers around the urban area is defined as the rural buffers (Fig. 1). The precipitation differences between the urban and the rural areas are calculated and compared in this section. The 30-m resolution land cover map is regridded into a 0.1° GAIA map according to the spatial resolution of the CMFD dataset. For each grid of the new GAIA map, if the area of the original urban grids accounts for over 50% of its total area, it is defined as an urban grid; otherwise, it is defined as a rural grid. The precipitation indices are then spatially averaged over the urban and the rural grids separately, and the trend of each index is calculated based on the spatial mean during 1979–2018.
The analysis of section 3.1 reveals that the urbanization plays a role in amplifying the local climate change. Although the climate change is the main contributor to the urban precipitation trend, the effect of urbanization has been gradually identified in recent years. We apply the Student’s t-test to detect the change point of the precipitation over the LP and divide the study periods into two stages. Most of the extreme precipitation indices in the five UAs demonstrate a sudden change point in the middle or late 1990s (Table 1). This is in accord with the previous studies indicating an obvious inter-decadal variation of the precipitation over the LP, with the significant turning point in the middle and late 1990s. During this period, the UAs had been growing more rapidly. Therefore, it is inferred that the climate change plays a primary role in the first stage of the precipitation series, and the urbanization effect gradually emerges in the second stage.
Figures 4 to 7 and Table 1 display the evolution of Rmax, R10mm, R95p and the light rain days in the five UAs and their rural areas. In the first stage, the extreme precipitation exhibits consistent trends in the urban and the rural areas of the five UAs (Fig. 4 and Table 1). However, the discrepancies between the urban and the rural widen in the second stage. The Rmax, R10mm and R95p keep growing slowly in both the urban and the rural areas of Xi’an before 2001, however, the trends of extreme precipitation become discrepant between the urban and the rural areas after 2001, with increase in the urban areas and decrease in the rural areas. The Rmax, R10mm and R95p in the urban and the rural areas of Taiyuan maintain a decreasing trend, which becomes steeper in the urban areas than in the rural areas in the second stage. For example, the trends of R95p in the urban and the rural areas of Taiyuan after 2001 are −1.09 mm·year −1and −0.06 mm·year −1, respectively (Table 2). The Rmax, R10mm and R95p in Xining, Hohhot and Luoyang show a consistent increase trend in both the urban and the rural areas. In the second stage, the increase rates in the urban areas get greater than that in the rural areas, all of which are significant at a 0.10 significance level (Table 2). The linear trends of the number of light rain days in the five UAs show similar characteristics in the two stages, but the variation is significant. In the second stage, the number of light rain days in the rural areas of Hohhot, Taiyuan and Xining increases more significantly than that in the urban areas, on the contrary, that in the urban areas of Xi’an decreases much more than that in the rural areas. As a result, the number of light rain days increases less in the urban areas than in the rural areas. The above analyses illustrate that the climate change is the dominant factor for the urban precipitation in the first stage, but the urbanization effect is further revealed in the second stage.