**3. Results**

### *3.1. Spatio-Temporal Variation of Snow Disasters*

### 3.1.1. Temporal Variation Characteristics

According to the above division criteria for the grade of snow disaster, the frequency and areal proportion of snow disasters on the QTP from 1980 to 2019 were summarized (Figure 1). Over the past 40 years, 36,330 snow disasters happened at 724 grid points on the QTP, that is, 908 snow disasters every year on average. The annual average frequency was the lowest, at only 687, in the 2010s (2010–2019), while that in the 1990s (1990–1999) was relatively high at about 1034. Although the climate tendency rate shows that the overall frequency of snow disasters declined significantly (−94.9 times/decade, *p* < 0.01), the variations showed interdecadal characteristics. The frequency of snow disasters increased with volatility (74.2 times/decade, *p* < 0.05) in the 1980s and 1990s, while, after entering the 21st century, it exhibited a monotonic decreasing trend (−285.5 times/decade, *p* < 0.01). Variations of the frequency of different grades of snow disaster also showed remarkable differences: 354 slight snow disasters happened annually on average, with the lowest number in 2018 (168 times), while the largest number occurred in 1999 (622). The frequency of slight snow disasters showed a decreasing trend in the long run (−16.8 times/decade) (not passing the significance test). However, it also demonstrated a tendency to increase first, then decrease during the aforementioned interdecadal variation; the frequency increased significantly in the 1980s and 1990s by 186 times/decade (*p* < 0.01), while it significantly declined since the beginning of the 21st century by −168.4 times/decade (*p* < 0.01). Variations in the frequency of moderate snow disasters also showed similar characteristics: the frequency declined significantly (−19.4 times/decade, *p* < 0.05); it significantly increased before the beginning of the 21st century (39.5 times/decade, *p* < 0.05) and decreased significantly (−67.5 times/decade, *p* < 0.01) thereafter. In terms of the long-term variation trend, the frequencies of heavy and extremely heavy snow disasters were, sepa-

rately, −28.7 times/decade (*p* < 0.01) and −29.1 times/decade (*p* < 0.01) without significant interdecadal variation.

**Figure 1.** Variation of the frequency (**a**) and areal proportion of snow disasters (**b**) on the QTP over past 40 years.

Regarding the areal proportion of snow disasters, an average of 34.3% of the area of the QTP suffered from snow disasters every year, while the areal proportion of snow disasters shrank at −1.9%/decade (*p* < 0.05). This decline mainly occurred since the 2010s. Although the areal proportions of different grades of snow disaster changed with certain differences over the last 40 years, they all showed a decreasing trend. The areal proportions of slight and moderate snow disasters decreased slightly and did not pass the significance test, while those of heavy and extremely heavy snow disasters declined significantly by −2.1%/decade and −1.8%/decade, separately, passing the significance tests at the 0.01 and 0.05 levels.

As to the long-term variations in the annual frequencies of different grades of snow disasters at various grid points (Figure 2), the long-term variation trends of different grades of snow disasters and that of all snow disasters showed similar spatial distribution over the last 40 years. That is, the frequency of snow disasters showed a decreasing trend for the majority of the QTP in the context of global warming, with areas with the most significant decrease distributed mainly in the center and south of the Tibetan Autonomous Region and the north of Qinghai Province. The areas with an increasing frequency of snow disasters were dispersed and relatively concentrated in the north-western and south-eastern parts of the QTP with a slightly increasing trend.

The duration of snow disasters and snow depth are also important indices used for measuring snow disasters. Figure 3 illustrates the regional average variations of the duration of snow disasters and snow depth on the QTP over the past 40 years and the corresponding spatial distribution of long-term variation trends. The figure shows that the average duration of snow disasters in the past 40 years was 116 d, with the longest being 151 d (1982) and the shortest being 69 d (2006). The average snow depth was 73 mm, with the deepest being 89 mm (1992) and the shallowest being 60 mm (2006). Compared with variations of the snow depth (Figure 3c), the duration of snow disasters showed more obvious interdecadal variation characteristics. It can be seen from Figure 3a that the duration of snow disasters showed a significant decreasing trend before the beginning of the 21st century (−22.7 d/decade, *p* < 0.01); the snow disasters lasted for the shortest time in 2006, while their duration rapidly lengthened in the several years following until 2010 when the duration showed stable volatility. Correspondingly, the snow depth did not fluctuate greatly over the past 40 years and did not have significant interdecadal variation; however, the long-term variation trends of the duration of snow disasters and snow depth at different grid points (Figure 3b,d) exhibited significant spatial differences, and the two showed similar spatial distributions. The majority of the area of the QTP was found to have decreased duration of snow disasters and snow depth. The areas with the most

significant decrease in the duration were mainly distributed in the center and south of the Tibetan Autonomous Region. The duration of snow disasters in these high-value centers was found to have a decreasing trend in the range of −40 to −67 d/decade. In addition, a sub-high-value center was found in the north of Qinghai Province, where the duration of snow disasters decreased in a trend from −20 to −40 d/decade. The areas with the most significant decreasing trend ( −1 to −2 cm/decade) of snow depth were mainly concentrated in the south of the Tibetan Autonomous Region. Areas where the duration of snow disasters showed an increasing trend were dispersed, with relatively concentrated areas in the north-west and south-east of the QTP, varying within 20 d/decade. Areas with increasing snow depth where the snow depth increase did not exceed 10 mm/decade, were more dispersed.

The annual grade of snow disaster is the highest grade of snow disaster in a year. In this way, the variations in regional average grades of snow disaster and the spatial distribution of long-term variations over the past 40 years were calculated (Figure 4); the multi-year variations in the grade of snow disaster on the QTP showed significant interannual volatilities and a slight decreasing trend over the past 40 years (not passing the significance test). However, the grade tended to decrease then increase in different sections, somewhat akin to the variations in the duration of snow disasters. That is, the grade of snow disasters exhibited a significant decreasing trend before the early 21st century, growing significantly for several years thereafter, and showing slight volatility in the 2010s. The long-term variations at different grid points were found to have an uneven spatial distribution; the grade of snow disaster slightly rose in most areas of the QTP, with the most significant increase at the south-eastern margin of the plateau. Areas where the grade of snow disaster declined were mainly distributed at the southern margin of the plateau and in the region of the Qaidam Basin, particularly the former, where the grade declined most significantly.

### 3.1.2. Spatial Distribution

From the spatial distribution of the overall frequency of snow disasters on the QTP over the past 40 years (Figure 5a), it can be seen that snow disasters were very unevenly spread across the plateau. In the west of the QTP, there is an obvious, low-value center from Ngari Prefecture in the Tibetan Autonomous Region (hereinafter shorted to Ngari) to Hotan Prefecture in the Xinjiang Uygur Autonomous Region (Hotan) to northern Nagqu County in the Tibetan Autonomous Region (Nagqu) to northern Yushu Tibetan Autonomous Prefecture in Qinghai Province (Yushu). Fewer than 10 snow disasters happened in most areas along the zone in the past 40 years, and most areas did not suffer from any snow disasters. In addition, there is also a low-value center with very few snow disasters in the north-east of the QTP from Xining City in Qinghai Province (Xining) to Haidong Prefecture in Qinghai Province (Haidong) to the Tibetan Autonomous Prefecture of Huangnan in Qinghai Province (Huangnan) to Gannan Tibetan Autonomous Prefecture in Gansu Province (Gannan) to northern Aba Tibetan and Qiang Autonomous Prefecture in Sichuan Province (Aba). There are three relatively concentrated high-value centers and one subhigh-value center with relatively high frequencies of snow disasters. The areas with the highest frequency of snow disasters are mainly found in the west of Shigatse City in the Tibetan Autonomous Region (Shigatse) and south-eastern Ngari in the south of the QTP, extending north-eastward to the center of Nagqu. This zone is where snow disasters happened most frequently on the QTP, with as many as 150 to 252 snow disasters over the past 40 years. Eastern Yushu in the middle of the QTP is also a high-value center and had around 150 to 200 snow disasters. Moreover, there is also a high-value center that had 150 to 200 snow disasters in the north-western end of the QTP from Kashgar Prefecture to Kizilsu Kirgiz Autonomous Prefecture (both in the Xinjiang Uygur Autonomous Region, Kashgar and Kizilsu Kirgiz). Snow disasters also happened frequently in the south-east of the QTP, which is a sub-high-value zone, with cumulative snow disasters amounting to 50 to 150 therein.

**Figure 2.** Spatial distribution of long-term variation trends for annual frequencies of different grades of snow disasters ((**<sup>a</sup>**–**<sup>e</sup>**) represent total snow disasters, slight, moderate, heavy, and extremely heavy snow disasters, respectively).

According to the spatial distribution of frequencies of slight, moderate, and heavy snow disasters (Figure 5b–d), despite different frequencies of different grades of snow disaster, the spatial distribution of the frequencies of these snow disasters was similar to that of the overall frequency of all snow disasters. That is, there are three relatively concentrated high-value centers, one sub-high-value center, and two low-value zones. In comparison, the frequency of extremely heavy snow disasters showed a very different spatial distribution (Figure 5e), mainly occurring in relatively decentralized high-value centers, and the relatively concentrated high-value centers are mainly located in two regions: the south-eastern end of the QTP and the north-western end of the plateau along Kashgar

to Kizilsu Kirgiz. The cumulative numbers of extremely heavy snow disasters in the two high-value centers were mainly between 50 and 95.

**Figure 3.** The regional average variation sequences and the spatial distribution of long-term variations of the duration of snow disasters and snow depth ((**<sup>a</sup>**,**b**) represent the duration of snow disasters; (**<sup>c</sup>**,**d**) represent the snow depth, respectively).

**Figure 5.** Spatial distribution of frequencies of different grades of snow disasters on the QTP over the past 40 years ((**<sup>a</sup>**–**<sup>e</sup>**) represent total snow disasters, slight, moderate, heavy, and extremely heavy snow disasters, respectively).

Figure 6 shows the spatial distribution of the average duration of snow disasters and the corresponding average snow depth on the QTP over the past 40 years. The low-value zones are distributed in areas consistent with the frequency of snow disasters, while the high-value zones are distributed in different areas. The area with the longest average duration of snow disaster was in the north-western end of the QTP from Kashgar to Kizilsu Kirgiz, and the average duration of snow disaster in the high-value center was between 150 and 229 d. High-value centers with a long duration of snow disaster are also present in the south-east and the middle (eastern Yushu) of the QTP. Although snow disasters occur frequently in the south of the QTP, from the western Shigatse–south-eastern Ngari line, the duration does not tend to be any longer than in the aforementioned areas, so it is a

sub-high-value zone. The spatial distribution of the average snow depth was similar to that of the duration of snow disasters; the high-value center is located in the north-western end of the QTP along the Kashgar–Kizilsu Kirgiz line, followed by the south-east of the plateau.

**Figure 6.** Spatial distribution of the average duration of snow disasters (**a**) and the average snow depth (**b**) on the QTP over the past 40 years.

The spatial distribution of multi-year average grades of snow disaster on the QTP over the past 40 years (Figure 7) indicates that the average grades of snow disaster were distributed with multiple high-value centers. The highest-value center is in the northwestern end of the QTP from Kashgar to Kizilsu Kirgiz, where the multi-year average grade of snow disasters always reached a level concomitant with extremely heavy snow disasters every year. There is also an area with a relatively high average grade of snow disaster in the south-east of the QTP, with the core area being from Lhoka City–Nyingchi City in the Tibetan Autonomous Region (Nyingchi) to Diqing Tibetan Autonomous Prefecture in Yunnan Province (Diqing). In that area, heavy snow disasters may take place. The zone

from Yushu to the Tibetan Autonomous Prefecture of Golog in Qinghai Province (Golog) to Aba in the center of the QTP is also a continuous area with a high average grade of snow disaster, with snow disasters in the central zone reaching a heavy grade. The zone with a high average grade of snow disasters and covering a large area is found in the south of the QTP along the western Shigatse–south-eastern Ngari line, with a moderate grade of snow disaster on average.

**Figure 7.** Spatial distribution of the average grade of snow disaster on the QTP.

### *3.2. Risk Assessment of Snow Disasters for Husbandry*

The risk of snow disasters for animal husbandry on the QTP was assessed from four perspectives: hazard factors, hazard-inducing environments, hazard-affected bodies, and disaster prevention and mitigation capacity. Four hazard factors were selected in the research, including the duration, snow depth, frequency, and grade of snow disaster (spatio-temporal variation characteristics of each factor are provided above). By using the weighted comprehensive evaluation method, the risk of hazard factors of snow disasters for animal husbandry on the QTP was zoned. This mainly reflects the intensity and probability of hazard factors that cause snow disasters to affect animal husbandry and is the leading precondition for snow disasters. The larger the risk of hazard factors, the greater the intensity thereof and the higher the probability of ensuing damage. Figure 8a shows the spatial distribution of the risk of hazard factors of snow disasters for animal husbandry on the QTP; the risk of hazard factors is low in the north while high in the south on the whole, having multiple high-value centers. The high-risk zone of snow disasters is mainly concentrated in Kashgar in the north-western end of the QTP, which is characterized by a long duration of snow, large snow depth, and high grade of snow disaster. The subhigh-value zone is mainly concentrated in the south-east of the QTP, including southern Lhoka and Nyingchi, Nujiang Lisu Autonomous Prefecture in Yunnan Province (Nujiang), Diqing, and southern Garzê Tibetan Autonomous Prefecture in Sichuan Province (Garzê). Snow disasters in the zone are mainly characterized by a high grade and a high frequency of extremely heavy snow disasters.

**Figure 8.** Zoning of the risk of hazard factors (**a**), sensitivity of hazard-inducing environments (**b**), vulnerability of hazard-affected bodies (**c**), and disaster prevention and mitigation capacity (**d**) of snow disasters for husbandry on the QTP.

In terms of the hazard-inducing environment, three factors, including the altitude, slope, and slope aspect, were mainly considered. Research showed that, under same or similar conditions, the combination of snow and landform may further aggravate the influences of snow disasters and cause certain secondary disasters. Therein, small topographic factors, such as the altitude, slope, and slope aspect, exert more significant influences. Therefore, the three factors, i.e., altitude, slope, and slope aspect, were selected to analyze the sensitivity of the hazard-inducing environment of snow disasters for animal husbandry on the QTP (Figure 8b). The QTP lies at a high altitude, on the whole, and features steep mountains, so the hazard-inducing environment of snow disasters for animal husbandry is of high sensitivity, and areas of low sensitivity are mainly located in the north-east and the south-eastern margin.

The degree of damage caused by snow disasters is, in fact, closely related to the body affected by snow disasters. The loss caused by snow disasters is not only dependent on the intensity, duration, and frequency of the disasters, but also is greatly affected by the hazardaffected bodies. Generally, the higher the vulnerability of hazard-affected bodies, the more easily these bodies are affected and the greater the loss (and vice versa). The present research mainly focused on snow disasters and their effects on animal husbandry, so cropsown area and livestock inventories were selected as indices representing the vulnerability of hazard-affected bodies. Zones with a large crop-sown area are mainly located in the northern and eastern QTP, in which Kashgar has the largest crop-sown area, followed by the Bayingolin Mongol Autonomous Prefecture in the Xinjiang Uygur Autonomous Region (Bayingolin). Two zones have large livestock inventories: one is Kashgar–Hotan– Bayingolin–Nagqu–Shigatse in the west of the QTP, and the other is Garzê–Aba in the

east of the plateau. By combining these two indices, the zoning of the vulnerability of hazard-affected bodies in snow disasters for the husbandry on the QTP can be obtained (Figure 8c). Areas of high vulnerability of hazard-affected bodies are mainly distributed in Kashgar–Hotan–Bayingolin–Nagqu–Shigatse in the west of the plateau and Garzê, Hainan Tibetan Autonomous Prefecture in Qinghai Province (Hainan), and Gannan in the east.

The disaster prevention and mitigation capacities refer to both disaster resistance and post-disaster resilience, which are mainly represented by the local level of economic development and the economic capability of farmers and pastoralists. In this research, the (municipal/prefecture) GDP and the per capita net income of farmers and pastoralists were selected as indices to reflect the local disaster prevention and mitigation capacity, thus, finally obtaining a zoning map of the capacity (Figure 8d). As shown in the figure, areas with high disaster prevention and mitigation capacity include Xining, Bayingolin, Liangshan Yi Autonomous Prefecture in Sichuan Province (Liangshan), Aba, Kashgar, and Haixi.

The comprehensive risk index of snow disasters for husbandry on the QTP was calculated using Equation (2) for comprehensive risk assessment according to the weights of the four factors (the hazard factors, hazard-inducing environments, hazard-affected bodies, and disaster prevention and mitigation capacity) (Table 3). The comprehensive risk was graded using the natural breaks method, finally attaining the zoning map for the comprehensive risk of snow disasters for animal husbandry on the QTP (Figure 9). The risk of snow disasters can be divided into five grades: high-risk, sub-high-risk, medium-risk, low-risk, and sub-low-risk zones. The high-risk zone accounts for about 1.54% of the total area of the QTP, mainly located in Kashgar at the north-western end of the plateau; the sub-high-risk and medium-risk zones are mainly found in the south of the plateau and are distributed in a tripole pattern, separately accounting for 15.96% and 16.32% of the plateau. The three "poles" are located along the Lhoka–Nyingchi–Nujiang–Diqing–southern Garzê line in the south-east, western Shigatse in the south, and Yushu in the hinterland of the QTP. The northern QTP is mainly dominated by low-risk and sub-low-risk zones, which separately cover 43.06% and 23.12% of the plateau.

**Figure 9.** Zoning of the comprehensive risk of snow disasters for husbandry on the QTP.

### *3.3. Possible Influences of Snow Disasters on FPWB*

To further explore the possible influences of major natural disasters, represented by snow disasters on FPWB, several typical regions (Figure 10) were selected for discussion based on the aforementioned analysis of the spatial-temporal variations of snow disasters in the QTP and zoning of the comprehensive risk of snow disasters for animal husbandry. These regions were Kashgar (I), Shigatse (II), Nagqu (III), Qamdo (IV), and Yushu (V). Among these regions, Kashgar represents the high-risk zone and is characterized by the high risk of hazard factors and high vulnerability of hazard-affected bodies while also having favorable disaster prevention and mitigation capacity. Shigatse mainly contains sub-high-risk and medium-risk zones, characterized by the moderate risk of hazard factors and moderate disaster prevention and mitigation capacity, while showing high sensitivity in its hazard-inducing environments and high vulnerability of hazard-affected bodies. Nagqu and Qamdo, sharing the similar comprehensive risk of snow disasters for animal husbandry and the risk of hazard factors, both belong to the medium-risk zone with above moderate vulnerability of hazard-affected bodies and poor disaster prevention and mitigation capacity. Western Yushu is a vast, depopulated zone, while the more heavily populated eastern Yushu region shows a sub-high comprehensive risk of snow disasters and a sub-high risk of hazard factors, moderate vulnerability of hazard-affected bodies, and poor disaster prevention and mitigation capacity.

**Figure 10.** Typical regions (**a**) and variations of the FPWB index and the risk of snow disasters for every typical region ((**b**–**f**) represent I, II, III, IV, and V, respectively).

According to the method of calculation of the FPWB index, variations in the FPWB index in various typical regions from 2000 to 2019 were calculated (the period from 2003 to 2017 was used for Yushu due to problems encountered in data acquisition). Meanwhile, the risk of hazard factors was used as the comprehensive assessment model for the severity of snow disasters. Variations in the risk of snow disasters of these typical regions in the same period were calculated, and the FPWB indices and variations of the risk of snow disasters are illustrated in Figure 10. From the long-term variation, the FPWB in these typical regions showed a significant ascending trend, which passed the significance test at the 0.01 level. The result indicates that the FPWB on the QTP has improved significantly since the 21st century, and the average growth rate of FPWB in the five typical regions is 3.6%, 1.9%, 2.8%, 1.3%, and 2.3%, respectively. The risk of snow disasters in regions I, II, and V exhibited a significant decreasing trend, while that in regions III and IV did not show significant variation. By using least-squares regression, the contribution of variations in the risk of snow disasters to the FPWB from 2000 to 2019 was quantitatively estimated (Table 5). The results sugges<sup>t</sup> that the risk of snow disasters had an adverse effect on variations of the FPWB. Every 1% increase in the risk of snow disasters in several typical regions corresponded to 0.186%, 0.768%, 0.378%, 0.109%, and 0.03% decreases in the FPWB index. Snow disasters affect the FPWB mainly by directly or indirectly impairing material resources (livestock inventories and meat production) and social and financial resources. Similar results were found in the research by Qiu et al. [22]. Because animal husbandry on the QTP is relatively unsophisticated and mainly depends on individual managemen<sup>t</sup> of farmers and pastoralists, the area is far from realizing mechanization, intensification, and modernization of its agricultural practices; it is heavily dependent on prevailing meteorological conditions. Natural disasters, represented by snow disasters, directly affect the livelihood of local farmers and pastoralists and exert adverse impacts on the FPWB. Once a snow disaster occurs, it is generally accompanied by low-temperature weather, and the snow cannot be removed for a long time. This, on the one hand, directly affects livestock and poultry (often killing and injuring many animals and birds); on the other hand, a snow disaster also causes insufficient supply of forage grass, which affects later feeding and managemen<sup>t</sup> and even causes death of livestock and poultry. At the same time, winter and spring are seasons with a high incidence of animal disease epidemics. Once a snow disaster occurs, the insufficient supply of forage grass may also lead to undernutrition and decreased immunity of livestock and poultry, rendering them more susceptible to infection, thus, influencing livestock inventories and meat production. In addition, in the process of snow disasters, the huge and rapid snowfall frequently causes collapse of livestock housing and breeding sheds, inducing deaths and injuries of livestock, bringing an economic loss to farmers and pastoralists and even causing causalities among farmers. Apart from the direct influences affecting development of animal husbandry, snow disasters also directly affect the life and production activities of residents, even causing major economic losses. For example, snow and ice heavily damage electric power facilities and transportation, hindering daily transportation of animal husbandry products, affecting the whole supply chain.

**Table 5.** Contribution of variations of the risk of snow disasters to the FPWB.

