3.1. Temporal Variability in Rainfall Erosivity
Sen’s slope estimator analysis indicated that rainfall erosivity in the Tibetan Plateau exhibited an increasing trend from 1980 to 2015, but the trend was not significant. Sen’s slope of annual rainfall erosivity was 3.05 (
p = 0.32). However, variability at each station was high, and 13 stations exhibited a decreasing trend (34% of the stations), and 25 stations an increasing trend (66% of the stations) (
Figure 2a,
Table 1). The climate tendency rate had a significantly increasing trend at the Gerze station in the northern part of the Tibetan Plateau, Shenza station in the middle, and Markam station in the eastern part of the Minjiang River basin (at the 0.05 significance level), and the rate increased by 20 MJ·mm·ha
−1·h
−1·10a
−1, 32 MJ·mm·ha
−1·h
−1·10a
−1, and 50 MJ·mm·ha
−1·h
−1·10a
−1, respectively, at the three stations. Increasing rainfall erosivity was mainly found in the southern Tibetan valley, Yarlung Zangbo River basin, and Hengduan Mountains in the Tibetan Plateau, especially in the Lancang River Basin. An annual rainfall erosivity anomaly showed that it fluctuated. The rainfall erosivity was small before 1987, and the value increased after 2010. In the study period, two-thirds of the annual rainfall erosivity was higher than the average; the minimum appeared in 1992, and the maximum appeared in 1998 (
Figure 2b). The results of the M–K test showed a non-significant increasing trend in rainfall erosivity in the Tibetan Plateau since 1984 (
Figure 2c).
The distribution of rainfall erosivity over the 24 and a half months is an input factor for many soil erosion model calculations, including the Chinese Soil Loss Equation (CSLE) [
70] and for calculating vegetation cover and biological measure factors. Data from 1981 to 2010 indicated that rainfall erosivity in the Tibetan Plateau was mainly concentrated in June–September, accounting for 81% of the year (
Figure 3). This was closely related to the monsoon climate of the Tibetan Plateau, as monsoons account for 58.5% of the annual precipitation [
52]. The proportion of rainfall in May–October accounted for 90% of the rain in the eastern part of Tibet [
49]. Therefore, the rainy season was also a frequent period of soil water erosion.
Rainfall erosivity at most meteorological stations in spring showed an increasing trend; only 5 of the 38 meteorological stations had a decreasing trend (
Figure 4a,
Table 2). The climate tendency rate indicated a significant increasing trend (at the 0.05 significance level) at Cona and Nyingchi stations in the southern part of the Tibetan Plateau, and a highly significant increase (at 0.01 significance level) at the Sog station in the upper reaches of the Nu River Basin; the climate tendency rate increased by 10, 20, and 11 MJ·mm·ha
−1·h
−1·10a
−1, respectively, for the three stations. Sen’s slope analysis indicated that rainfall erosivity in the Tibetan Plateau increased during the study period. Sen’s slope value for spring rainfall erosivity was 1.85 (
p = 0.005 < 0.01). Rainfall erosivity with a decreasing trend was found in the western part of the Tibetan Plateau where the Zuogong station had an insignificant decreasing trend; its erosivity decreased by 40 MJ·mm·ha
−1·h
−1·10a
−1 (
Table 2). A spring rainfall erosivity anomaly showed that rainfall erosivity was smaller before 2000, and larger after 2000. During the study period, spring rainfall erosivity was below average 67% of the time; the minimum appeared in 1987 and the maximum in 2010 (
Figure 4b). The M–K test results showed an increasing trend in spring erosivity in the Tibetan Plateau since 1987, a significant increasing trend after 2010, and the abrupt year occurred in 2005 (
Figure 4c).
Increasing rainfall erosivity in summer was observed at more stations than did decreasing, with 21 of 38 meteorological stations (55%) showing increasing trends in rainfall erosivity (
Figure 5a,
Table 3). Based on the climate tendency rate, Shenza station in the middle of the Tibetan Plateau, and the Qamdo station of the Lancang River Basin in the eastern Hengduan Mountains showed a significant trend (at 0.05 significance level) with an increase of 31 and 41 MJ·mm·ha
−1·h
−1·10a
−1, respectively; the trend at Gerze station in the west and Markam station in the Lancang River Basin in the eastern Hengduan Mountains was highly significant (at the 0.01 significance level) at 24 and 69 MJ·mm·ha
−1·h
−1·10a
−1, respectively. The results of Sen’s slope estimator indicated an increasing but not significant trend during the study period. The Sen’s slope value of summer rainfall erosivity was 1.41 (
p = 0.41). Rainfall erosivity showing a decreasing trend was observed mainly in the western part of the Tibetan Plateau, the southern part, and the western part of Hengduan Mountains, with the largest decrease at Gyangze and Lhorong stations where the trend coefficient decreased by 34 and 24 MJ·mm·ha
−1·h
−1·10a
−1, respectively (
Table 3). A summer rainfall erosivity anomaly showed that summer rainfall erosivity was above average in most years (57%), with the minimum in 1983 and the maximum in 1998 (
Figure 5b). The M–K test results indicated an insignificant increasing trend of summer rainfall erosivity in the Tibetan Plateau since 1983 (
Figure 5c).
The number of stations with increasing rainfall erosivity in autumn was slightly lower than that with a decreasing trend; 17 of the 38 meteorological stations (45%) showed an increasing trend for rainfall erosivity (
Figure 6a,
Table 4). Bases on the climate tendency rate, the Konggar station in the Yarlung Zangbo River valley in the southern part of the Tibetan Plateau (near Zetang station) showed a significant increasing trend (at the 0.05 significance level), and the Lhunze station showed a highly significant increasing trend (at the 0.01 significance level), with an increase of 16 and 13 MJ·mm·ha
−1·h
−1·10a
−1, respectively. The results of autumn rainfall erosivity indicated an insignificant decreasing trend during the study period. The Sen’s slope value of autumn rainfall erosivity was −0.02 (
p = 0.98). In addition to the Yarlung Zangbo River Basin, autumn rainfall erosivity in other areas was mainly decreasing (
Table 3). Autumn rainfall erosivity at the Bome and Zuogong stations in the Hengduan Mountains in the eastern part of the Tibetan Plateau decreased by 46 and 9 MJ·mm·ha
−1·h
−1·10a
−1, respectively. An autumn rainfall erosivity anomaly shows that autumn rainfall erosivity was above average in most years (50%), with the minimum in 1992 and the maximum in 1985 (
Figure 6b). The M–K test results indicated an insignificant increasing trend in the Tibetan Plateau from 1984 to 2003, and a slowly increasing trend after 2008 (
Figure 5c).
Winter rainfall erosivity in the Tibetan Plateau was very low, and about 66% of the stations had a value of 0 (
Figure 7a,
Table 5). The results indicated an insignificant decreasing trend in winter rainfall erosivity during the study period. Meanwhile, the Sen’s slope value of winter rainfall erosivity was −0.03 (
p = 0.89). Based on the climate tendency rate, the largest decreasing trend was observed at the Lhunze station and Qamdo in the Hengduan Mountains in the eastern part of the Tibetan Plateau, as well as at the Cona station in the southern region, where it decreased almost 100%. The anomaly showed that winter rainfall erosivity was below average in most years (66%), with the minimum in 1992 and the maximum in 1989 (
Figure 7b). The M–K test results demonstrated that the winter rainfall erosivity in the Tibetan Plateau had an insignificant decreasing trend since 1991 (
Figure 7c).
3.2. Spatial Distribution of Rainfall Erosivity in the Tibetan Plateau
Spatial distribution of rainfall erosivity in the Tibetan Plateau showed a decreasing trend from the southeast to the northwest (
Figure 8). The average annual rainfall erosivity at 38 meteorological stations varied from 61 to 1776 MJ·mm·ha
−1·h
−1 during the study period, with an average of 714 MJ·mm·ha
−1·h
−1. The largest annual rainfall erosivity was observed at the Bome station in the eastern part of the Tibetan Plateau, and the smallest at the Shiquanhe station in the western region (
Figure 8). Rainfall erosivity was <500 MJ·mm·ha
−1·h
−1 at 24% of the stations, 500–1000 MJ·mm·ha
−1·h
−1 at 55%, and >1000 MJ·mm·ha
−1·h
−1 at 21% of the stations. Relatively high rainfall erosivity was mainly distributed in the Hengduan Mountains in the eastern part of the Tibetan Plateau and in the low-elevation areas between the Yarlung Zangbo and Nu rivers.
The seasonal distribution of rainfall erosivity varied greatly across the Tibetan Plateau; the average summer rainfall erosivity was 482 MJ·mm·ha
−1·h
−1, accounting for a maximum of 67.5% of the annual rainfall erosivity, followed by autumn and spring, accounting for 18.5% and 11.5%, respectively. The proportion of winter rainfall erosivity was the smallest at only 2.5%. Seasonal variability in rainfall erosivity varied among meteorological stations, but generally followed rain distribution in summer > autumn > spring > winter (
Figure 9). The proportions of spring to annual total rainfall erosivity differed among meteorological stations. The Shiquanhe station had the smallest proportion at 0 and the Zayu station in the Hengduan Mountains had a proportion of 46.5%. Meanwhile, the proportion in the south was higher than in the north, and higher in the east than in the west, resulting in a slight decreasing trend from the southeast to the northwest (
Figure 9a). The proportion of summer rainfall erosivity varied from the smallest of 16.4% at Nyalam, to the largest of 94.7% at Shiquanhe station. In addition, 71% of the stations accounted for more than 70% of summer rainfall erosivity, and summer rainfall erosivity in most meteorological stations in the southern Tibet Valley contributed more than 80% to the total (
Figure 9b). The proportion of autumn rainfall erosivity at each meteorological station varied between 5.3% and 39.4% with the highest at Purang station, followed by the Nyalan and Cona stations; these stations were located on the southernmost edge of the Tibetan Plateau (
Figure 9c). The proportion of winter rainfall erosivity to the total was small, with less than 4% of the stations contributing >90% of the winter erosivity. In that, the Nyalam station contributed the most at 33.9%, followed by Purang, both located on the southwestern edge of the Tibetan Plateau (
Figure 9d).