**3. Results**

### *3.1. Climatology of Precipitation*

Figure 2 shows the climatology of annual and seasonal precipitation on the SRTR. The annual precipitation ranges from 500 to 1000 mm/a, presenting a pattern of gradual decrease from southeast to northwest. Among the different seasons, summer (June-JulyAugust, JJA) precipitation dominates the pattern of annual precipitation, accounting for about 61.3% of the annual precipitation (Figure 3). In spring (March-April-May, MAM), summer and autumn (September-October-November, SON), the precipitation distribution presents a pattern of decrease from southeast to northwest, while in winter (December-January-February, DJF), precipitation has no significant spatial distribution characteristics. In addition, the summer precipitation is mainly in July and August, suggesting influences of water vapor transport by the summer monsoon.

**Figure 2.** Precipitation (mm/month) in (**a**) annual average, (**b**) June-July-August (JJA) average, (**c**) March-April-May (MAM) average, (**d**) June average, (**e**) September-October-November (SON) average, (**f**) July average, (**g**) December-January-Februrary (DJF) average, and (**h**) August average from 2001 to 2019.

In terms of precipitation occurrence ratio (i.e., proportion of precipitation days in the total days, Figure 4), the distribution characteristics for multi-year climatology, MAM, JJA and SON consistently decline from southeast to northwest, the same as precipitation amount shown in Figure 2. There is not much distinction between MAM and SON. However, for JJA, precipitation days occupy more than 60%, particularly in June, almost 80% of the days have precipitation events, followed by July and August. In the west of the SRTR, precipitation is suppressed most of the time. However, for the east part of the source region of the Yellow River, it always presents a relatively higher frequency of precipitation in all seasons, which has also become an important water supply area for the Yellow River. In DJF, precipitation only occurs in this region, suggesting a very dry condition in other regions of the SRTR.

**Figure 3.** Monthly accumulated precipitation on the SRTR (averaged from 2001 to 2019).

**Figure 4.** Precipitation occurrence ratio in (**a**) annual average, (**b**) June-July-August (JJA) average, (**c**) March-April-May (MAM) average, (**d**) June average, (**e**) September-October-November (SON) average, (**f**) July average, (**g**) December-January-Februrary (DJF) average, and (**h**) August average from 2001 to 2019.

### *3.2. Changes in Precipitation*

### 3.2.1. Precipitation Amount

Figure 5 shows differences between the climatology of precipitation in the two decades, i.e., 2010 to 2019 and 2001 to 2010. For the annual average precipitation, it shows a slight decrease in the north and west of the SRTR and a slight increase in the eastern and southern parts. In a large area of central SRTR, precipitation shows tiny variation. The most significant changes happened in the southeast of the source region of the Yellow River and the Lantsang River. Considering the contribution of different seasons, MAM, SON and DJF dominate the decreasing of precipitation in most area of the SRTR, while JJA contributes the most increases. Precipitation in spring presents a similar pattern, with the annual average, while in SON, precipitation in most of the area shows a drying trend

except the southeast of the source region of the Yellow and Lanstang Rivers. The winter presents a total drying pattern in contrast with the total wetting variation in most areas in summer. For different months in summer, precipitation shows a significant increase in June, especially in the center to the south, while it has an increase in the north in August. In July, it shows basically a drying trend in the northwest of the SRTR and a wetting trend in the south.

**Figure 5.** Difference between precipitation (**a**) annual average, (**b**) June-July-August (JJA) average, (**c**) March-April-May (MAM) average, (**d**) June average, (**e**) September-October-November (SON) average, (**f**) July average, (**g**) December-January-Februrary (DJF) average, and (**h**) August average from 2010 to 2019 and from 2001 to 2010 (mm/month).

### 3.2.2. Precipitation Frequency

As to the changes in precipitation frequency, for the annual average, MAM, SON and DJF, most areas show a reduction in the precipitation occurrence ratio (Figure 6). Only in June and August do the precipitation occurrence ratios increase in most areas. In the west of the Yangtze River headwater region, a significant rise in the precipitation occurrence ratio is presented, although the precipitation amount does not increase correspondingly in this region. The spatial correlation coefficients between the variations of precipitation amount and precipitation occurrences ratio over the two decades are 0.689, 0.752, 0.48, 0.697 and 0.437 for the annual average, MAM, JJA, SON and DJF, respectively. The higher spatial correlation in MAM and SON suggests the possibility of precipitation reduction caused by

the decreasing precipitation frequency in these two seasons, while this is not the same in JJA and DJF. In June, July and August, the spatial correlation coefficients are 0.195, 0.604 and 0.611, indicating the inconsistency in changes in precipitation amount and frequency. '

**Figure 6.** Difference in precipitation occurrence ratio between 2010 and 2019 and from 2001 to 2010 in (**a**) annual average, (**b**) June-July-August (JJA) average, (**c**) March-April-May (MAM) average, (**d**) June average, (**e**) September-October-November (SON) average, (**f**) July average, (**g**) December-January-Februrary (DJF) average, and (**h**) August average.

### 3.2.3. Afternoon and Nighttime Precipitation

Figure 7 shows the changes in afternoon precipitation in the same period as Figure 5. To address the contribution of afternoon precipitation to precipitation changes in the two decades, we calculated spatial correlation coefficients between the afternoon precipitation differences and total precipitation differences between the two decades (Table 1). Comparing Figure 5 with Figure 7, there was a similar pattern in Year, MAM, SON and JJA, but the results were quite different in the three months in summer. In terms of the spatial correlation coefficients, the maximum is from the total precipitation changes between the two decades, and the correlation coefficients are 0.552, 0.438, 0.518 and 0.805 for MAM, JJA, SON, and DJF, respectively, which are all larger than that in each month in summer.

**Figure 7.** Difference between afternoon precipitation in (**a**) annual average, (**b**) June-July-August (JJA) average, (**c**) March-April-May (MAM) average, (**d**) June average, (**e**) September-October-November (SON) average, (**f**) July average, (**g**) December-January-Februrary (DJF) average, and (**h**) August average from 2010 to 2019 and from 2001 to 2010 (mm/month).

**Table 1.** Spatial correlation coefficients (Cor) between afternoon precipitation differences and total precipitation differences among the two decades in different periods.


As shown in Figure 8, the nighttime precipitation changes show similar patterns as in Figure 5, except the magnitude in the night is different. The spatial correlation coefficients are around 0.9 for almost all time periods (Table 2), suggesting a dominant contribution of nighttime precipitation to the total precipitation changes. Figure 9 shows diurnal changes in precipitation rates between the two decades. In the morning in the local time (i.e., 00:00 to 04:00 UTC), there are no significant changes between the two decades. For afternoon precipitation (04:00 UTC to 12:00 UTC), there is a slight reduction in the 2010s in contrast

with the 2000s, whilst there is a strong increase shown for the nighttime precipitation, emphasizing the contribution of nighttime precipitation variation to the total precipitation.

**Figure 8.** Difference between nighttime precipitation in (**a**) annual average, (**b**) June-July-August (JJA) average, (**c**) March-April-May (MAM) average, (**d**) June average, (**e**) September-October-November (SON) average, (**f**) July average, (**g**) December-January-Februrary (DJF) average, and (**h**) August average from 2010 to 2019 and from 2001 to 2010 (mm/month).

**Table 2.** Spatial correlation coefficients between nighttime precipitation differences and precipitation differences among the two decades in different time periods.


In order to quantify the causes of precipitation changes, we calculated the correlation coefficients of precipitation difference with precipitation probability, afternoon precipitation difference, nighttime precipitation difference, afternoon precipitation frequency difference and nighttime precipitation frequency difference (Table 3). It is very clear that the maximums of the correlation coefficients occur between the precipitation difference and the nighttime precipitation difference, suggesting the domination of nighttime precipitation changes in the total variations. When comparing the precipitation intensity and

frequency in the nighttime (Table 4), the nighttime precipitation intensity difference shows a higher correlation with the night precipitation changes, implying the domination of nighttime precipitation.

**Figure 9.** Diurnal changes in precipitation rate (mm/h) between the two decades.

**Table 3.** Correlation coefficients of precipitation difference with precipitation probability (Pp), afternoon precipitation difference (Pa), nighttime precipitation difference (Pe), afternoon precipitation frequency difference (Ppa) and nighttime precipitation frequency difference (Ppe).


**Table 4.** Correlation coefficients of nighttime precipitation difference with nighttime precipitation intensity difference (Se) and nighttime precipitation frequency difference (Pa).

