An evaluation of vegetation loss due to the super typhoon Sarika in Hainan Island of China

With the continued strengthening of global climate change, various extreme climate events have become increasingly prominent. Typhoons are the most disastrous weather event that cause severe losses to the economy, agriculture, transportation, communication, and ecosystem in coastal regions. The super typhoon Sarika that hit Hainan Island on October 18, 2016, lasted for 15 h, and it was the most severe typhoon that hit this island in October since 1970. The coastal land of Hainan Island experienced gales with an average speed of 35 m/s. In this study, we evaluated the impact of the super typhoon Sarika on vegetation by performing normalized difference vegetative index (NDVI) difference analysis using MODIS multi-temporal images acquired before and after the typhoon. The assessment of typhoons depends on the land-use types and landscape topography of slope, aspect, and altitude. The results indicated that the super typhoon Sarika seriously hit forestry, agriculture, shrubs, plantations, and orchards on Hainan Island. Overall, 79% of vegetation exhibited a negative change, whereas only 21% of vegetation exhibited a positive change in NDVI after the super typhoon Sarika. Agriculture was most severely impacted by the typhoon, where more than 81% of areas exhibited a decrease in NDVI, followed by plantations and orchards, where 77% of areas exhibited a decrease in NDVI. Additionally, the impact of the typhoon on vegetation was affected by the degree of NDVI decrease with the altitude, slope, and aspect. In conclusion, vegetation damage is associated with land cover types, altitude, aspect, and slope. NDVI decreased more in low-altitude plain and coastal areas than in higher altitude montane forest areas.


Introduction
Typhoons are the greatest threat to life and have caused enormous economic losses in China. They can also cause vast damages to forest vegetation. Under global warming, their increasing frequencies and intensities pose challenges for effective ecological management and protection against natural hazards. Typhoons are characterized by heavy rainfall and strong wind, and typhoon winds can greatly influence forest structure and dynamics. The effect of tropical cyclones on individual trees ranges from defoliation and branch and canopy damage to bole snapping and uprooting. The level of damage depends on the intensity of the cyclone and the resistance properties of trees (Niklas 2000;Butler et al. 2012;Uriarte et al. 2019;Peereman et al. 2020). It can cause severe damage to trees having a large diameter, resulting in bark tearing, drying, branch breaking, falling of several leaves and uprooting (Yin et al. 2013;Liu et al. 2012;Dunham et al. 2000;Gardiner and Quine 2000;Xu et al. 2008), and early seed falling. It can even cause diseases and pests invasion (Cai et al. 2008). Falling of trees, broken branches, and fallen leaves overwhelm other plants, producing several forest windows (Zhao et al. 2004). This results in reduced canopy density, enhanced light transmittance, and habitat heterogeneity, eventually affecting the primary productivity, hydrological characteristics, nutrient cycle, vegetation renewal, animal community succession, species composition and structure, and ecosystem stability in the community Wang et al. 2007). Many researchers have studied the impact of typhoons on the forest ecosystem, focusing on vegetation dynamics, forest canopy and density, and mechanical damage to the forest community (Zimmerman et al. 1994;Everham 1996;Li et al. 1998;Sánchez and Islebe 1999).
Vegetation indices derived from satellite images have been used for monitoring vegetation changes, which have become the primary sources of information for monitoring vegetation conditions and mapping changes in land cover. Vegetation indices are widely used to determine vegetation damage and its severity. The leading vegetation indices include the normalized difference vegetation index (NDVI) (Teillet et al. 1997;Oindo and Skidmore 2002;Wang et al. 2010;Chu et al. 2014;Hu and Smith 2018;Pettorelli et al. 2005), enhanced vegetation index (EVI) (Brun and Barron 2013;Mandal and Hosaka 2020), the normalized difference infrared index (NDII) (Hardisky et al. 1983;Anderson et al. 2004;Cheng et al. 2006), and soil-adjusted vegetation index (SAVI) (Lawrence 1998). Of these vegetation indices, the NDVI is most commonly used, and it represents a function of red and near-infrared spectral bands, reflecting the vegetation changes post damage.
Hainan Island, referred to as the China's typhoon corridor, is deeply affected by typhoons, and a typhoon serves as an essential natural interference factor for this island (Wang et al. 2007). Changes in the frequency and intensity of cyclones can have significant implications for ecosystems (Zhang et al. 2013;Chiwen et al. 2018). Therefore, it is important to understand how tropical cyclones affect ecosystem dynamics. However, few attempts have been made to assess the damages caused by typhoons to forests and vegetation, and studies have focused mainly on the effect of typhoons on fatalities, major economic losses, and disaster risks in Hainan Island (Huang et al. 2019;Liu and Xu 2020). Lin et al. reported that from 1989 to 2018, 32% of tropical cyclones occurred in Northwestern Pacific, but only 18% of studies were conducted in this region and only two studies determined the impact of cyclone on forests in Hainan Island (Lin et al. 2020). Thus, the extent and pattern of vegetation damage caused by a typhoon on Hainan Island remain poorly understood. The increased frequency and severity of typhoon damages in recent years require rapid and accurate regional vegetation damage assessment to support post-typhoon recovery and reconstruction. Therefore, assessing the impact of typhoon is essential to support effective post-hurricane management approaches. Investigating the extent and magnitude of landscape and property damage by using traditional field survey approaches is time-consuming and expensive. Accessing the mountain region of Hainan Island is difficult, and satellite imagery offers an optimum means of studying large-scale forest damages. Remote sensing is a cost-effective tool for assessing vegetation damage caused by typhoon events at various scales Gouvenain and Silander 2003; Wang and Xu 1 3 2018), which can provide useful information to facilitate the relief process and manage infrastructure, communities, and natural resources in disaster management.
This study investigates the impact of strong natural damages of a super typhoon on vegetation in a tropical island with a monsoon climate. Taking the super typhoon Sarika as an example, we used multi-temporal remote sensing images to evaluate the impact of typhoon Sarika on the vegetation losses and gains in Hainan Island. The objectives of this study were as follows: (1) Analyze short-term impacts on vegetation by using vegetation indices derived from MODIS; (2) Evaluate the damage extents on different land covers and the severity of damage; and (3) Evaluate the spatial distributions of vegetation damage in relation to altitude, slope, aspect, and wind speed of the typhoon Sarika.

Description of the study area
Hainan Island (18° 10ʹ-20° 10ʹ N and 108° 37ʹ-111° 03ʹ E) is located at the southernmost tip of China, surrounded by the South China Sea. It looks like an ellipse and is the secondlargest island in China, with an area of approximately 34,500 km 2 and a coastline area of more than 1500 km. It is a provincial administrative region having 18 counties with a resident population of 9.45 million. A marine tropical monsoon climate with high temperatures is prevalent in this island. The average annual temperature is 24.8 °C. The seasonal temperature changes are not significant. The average annual precipitation is 1808 mm, and the seasonal and regional distributions are uneven. The terrain is high in the middle and low in the surrounding, with Wuzhi and Yingge Mountains at the center, with an altitude of 1867 m; the altitude gradually decreases outwards (Fig. 1). The southern island is close to the South China Sea and the Pacific Ocean to the East.
Typhoons generated in the South China Sea and the Pacific Ocean may threaten Hainan Province, with those developed in the ocean having the most vigorous intensity. On average, 7.2 typhoons affect Hainan Island every year. Among the typhoons that seriously affected Hainan Island, 86% were generated in the Pacific Ocean, and 93% were developed in the Pacific Ocean (Wan et al. 2016;Chen, 2000).

Meteorological data for super typhoon Sarika
The super typhoon Sarika originated over the western North Pacific Ocean, made landfall on Luzon (Philippines), and continued across the South China Sea toward Hainan Island. The typhoon became severe in the afternoon of October 15, 2016 and transformed into a super typhoon at midnight, making landfall at Wanning, Hainan Island, at 9:00 am on October 18, 2016. Further, it passed over Qiongzhou, Danzhou, and other cities and then moved from Danzhou to the northern sea of Beibu Gulf. The typhoon's track, direction, and wind data were obtained from the Chinese typhoon center (tcdata.typhoon.org.cn). Overall, 435 weather stations monitored the wind and precipitation. To study the evolution of the super typhoon Sarika and the intensity and distribution of gale, we obtained hourly gale data from the Chinese meteorological information center from October 15, 2016 to 1 3

Remote sensing data
To evaluate the severity and spatial patterns of super typhoon Sarika's impacts on vegetation in the study area, vegetation changes before and after the super typhoon were analyzed and compared. MODIS images with a 250-m spatial resolution remote sensing data were acquired for the comparison. These were on-demand terrain-corrected surface reflectance images from the Earth Explorer and USGS, application-ready products to support monitoring and assessing vegetation changes. The post-typhoon images were acquired on October 31, 2016 and November 16, 2016, and the control images were acquired on September 13, 2016 and September 29, 2016.

Land-use and digital elevation model data
The land-use class data were obtained from the Chinese national earth system science data center, with a spatial resolution of 30 m. A three-level classification system was adopted. According to land resources and their use attributes, the first level was divided into six categories: cultivated land, forest land, grassland, water area, construction land, and unused land (Fig. 2). Field investigation and high-resolution remote sensing image sampling indicated that the remote sensing monitoring accuracy of the land-use class was higher than 90%. The spatial distribution data of altitude (DEM) were obtained from the shuttle radar topographic mission (SRTM) data of the space shuttle Endeavour. This dataset was based on the latest SRTM v. 4 1. The 30-m provincial data were generated through data splicing and sorting. SRTM was a mission jointly completed with NASA. It was completed by the SRTM system carried on the space shuttle Endeavour launched by the United States.

NDVI analysis
In this study, we analyzed vegetation changes by using remote sensing images. NDVI analysis was used to monitor the vegetation coverage. It can be used to accurately calculate the absorption of the red band and reflection of the near-infrared band by green plants (Lyon et al. 1998). The NDVI can be calculated using the following formula: NIR and Red represent the reflectance in NIR and Red wavebands, respectively. NDVI values range from − 1 to + 1, with values greater than 0 generally representing vegetation and those less than 0 indicating non-vegetated areas or stressed vegetation. Positive values of the NDVI area are highly correlated with green biomass and vegetation productivity (Deering 1978;Pettorelli et al. 2005).
We compared the NDVI values before and after the typhoon event; the difference is represented as ΔNDVI. The ΔNDVI can be used to describe the change in the distribution of vegetation coverage, which can be calculated using the following formula:

Fig. 2 The land cover classes on Hainan Island
where NDVI p and NDVI b represent after-and before-typhoon NDVI images, respectively. The loss in vegetation or stress in vegetation was calculated by dividing the NDVI difference between the control and post-typhoon image with the NDVI of the control image. Positive values indicate normal conditions, whereas negative values indicate the stress or reduction in vegetation. A positive ΔNDVI indicates that the vegetation coverage has increased, and vice-a-versa. (Volcani et al. 2005;Sawaid et al. 2020).

Analysis of the magnitude of typhoon influence
The resulting image loss or gain values of NDVI were obtained in the range from − 1 to + 1 and were divided into six classes, namely, extremely stressed [− 1, − 0.5), severely stressed [− 0.5, − 0.1), slightly stressed [− 0.1, 0), slightly healthy [0, 0.1), moderately healthy [0.1, 0.5), and healthy [0.5, 1). The impact categories and degrees are presented in Table 1. All datasets were resampled to 250 m resolution for subsequent analyses. The loss or gain values of NDVI were superimposed over the layers of land cover classes, elevation, aspect, and slope topographic position indices, and the area statistics of impact severity levels were calculated using area tabulation in ArcGIS 10.1.

Wind speed in the super typhoon Sarika
Sarika was generated at 8:00 pm on October 13, 2016; with rapidly increased strength and the intensity level, Sarika upgraded to the levels of a strong tropical storm at 1:00 am on October 14, a typhoon at 05:00 am on October 15, a strong typhoon at 5:00 pm on October 15, and a super typhoon at 11:00 pm on October 15. The minimum central pressure was 935 hPa. The severe tropical storm was directly strengthened into the super typhoon level within a day on October 15, and the rapid intensity enhancement was rare. After landing in the Philippines, Sarika was slightly weakened; however, it maintained the typhoon level after entering the South China Sea and strengthened into a strong typhoon. At 09:00 am on October 18, Sarika landed on the coast of Hele Town, Wanning City, Hainan Island for the second time; the maximum wind speed reached 46.1 m/s, and the minimum pressure in the center was 950 hPa. It passed through Wanning, Qiongzhong and Danzhou, and other counties and then moved from Danzhou to the northern sea surface of Beibu Gulf at midnight on October 19. Finally, at around 02:10 pm on October 19, it landed again in the coastal area of Dongxing City, Fangchenggang, Guangxi Province, with the level of a severe tropical storm (Fig. 1).
(2) ΔNDVI = NDVI P − NDVI b Therefore, Sarika first upgraded from a tropical storm to a super typhoon, then weakened to a typhoon, again strengthened to a strong typhoon, and finally gradually weakened into a typhoon and a strong tropical storm. The intensity increased and changed rapidly. Sarika was the strongest typhoon that hit Hainan Island in October since 1971.
The coastal land around Hainan Island generally had gales with a maximum speed of 35 m/s from 08:00 am on October 17 to 03:00 pm on October 19. Five villages and towns experienced gales with a wind speed of more than 45 m/s. The wind speed was maximum (46.1 m/s) in Dahuajiao, Wanning City ( Table 2). The gale area is distributed in the coastal land of Hainan Island. The average wind speed of the whole island reached 12.8 m/s. The maximum wind speed was mainly distributed in the coastal region of Hainan Island, especially in the coastal region of the east island. The average maximum wind speed reached 28 m/s, and the wind speed in the coastal region of the west island was 20 m/s. The longer the distance from the sea and the higher the altitude, the lower the wind speed. The small wind area is distributed in the middle island with a high altitude (Fig. 3). Figure 4 illustrates the NDVI images acquired in November 2016, images acquired after the typhoon event, and control images acquired in September 2016. Before the typhoon, the average NDVI of Hainan Province was 0.77, most areas were dark green, and the vegetation was in good condition (Fig. 4A). After the typhoon, the average NDVI of Hainan Province reduced to 0.66, especially in the eastern part of Hainan and the coastal area around Hainan Island, which changed from dark green to orange and yellow, indicating the NDVI decreased of 15% (Fig. 4B). NDVI has a great impact on the eastern part of Hainan Island and the coastal area around the island. A comparative analysis of the impact of the two phases indicated that super typhoon Sarika impacted the vegetation of Hainan Province; however, it had a greater impact on vegetation in Wanning, Qiongzhou, Qionghai, Wenchang, Dingan, Haikou, Changjiang, and the coastal areas around the island. In the central and western regions of Hainan, including Chengmai, Baisha, Tongshi, Dongfang, and Sanya, vegetation damage was relatively less (Fig. 4C). Overall, the NDVI was reduced in an area of 23,256 km 2 , accounting for 79% of the total affected area. Among them, NDVI decreased in the range of [− 1, − 0.5). The severely affected area was 1481 km 2 , accounting for 5% of the total area, in which NDVI decreased in the range of [− 0.5, − 0.1). The more severely affected area was 9944 km 2 , accounting for 33% of the total area, in which NDVI decreased in the range of [− 0.1, 0). The area has a general impact of 11,832 km 2 , accounting for 40.2% of the total area. NDVI did not decrease during the typhoon, showing that it increased by 6193 km 2 , accounting for 21% of the total vegetation area of the affected area (Fig. 5).

Damages on various land cover types
To evaluate the impact of typhoon disasters on different vegetation types, we divided vegetation types into 6 types: forest; spare forest; shrubs; plantation and orchard; grassland; dryland crop; and paddy field crop. At the extremely stressed level, NDVI reduction was in the range of [− 1, − 0.5); the forest was the most affected area (699 km 2 ), accounting for Fig. 3 The distribution of wind speed for the super typhoon Sarika 48.8% of the total damaged area, followed by dry land crops (285 km 2 ), plantations and orchards (214 km 2 ), and paddy fields crops (119 km 2 ), accounting for 20.1, 15.1, and 8.4% of the total damaged area, respectively. In addition, the damaged area of grassland, spare forest, and shrub was low (106 km 2 ); the damage proportion was low, accounting for only 7.5% of the total damaged area (Fig. 6).
At the severely stressed level, NDVI reduction was in the range of [− 0.5, − 0.1); forests had the largest damaged area (3122 km 2 ), accounting for 34.0% of the total damaged area, The percentage of areas affected by the super typhoon Sarika at various levels followed by dry land crops (2158 km 2 ), plantations and orchards (1545 km 2 ), and paddy crops (1152 km 2 ), accounting for 23.5, 16.9, and 12.6% of the total damaged area, respectively. In addition, the damaged area of grassland, spare forest, and shrubs was small (1197 km 2 ), and the damage proportion was low, accounting for only 13%. At the slightly stressed level, NDVI reduction was in the range of [− 0.1, 0); forests had the largest damaged area (4591 km 2 ), accounting for 40.6% of the total damaged area, followed by plantations and orchards (1909 km 2 ), dryland crops (1814 km 2 ), shrubs (1096 km 2 ), and paddy fields (1076 km 2 ), accounting for 17.0, 16.2, 9.8, and 9.6% of the total damaged area, respectively (Fig. 6).
Overall, the super typhoon Sarika impacted 68.1% of the forests in Hainan Island, of which the damage at extremely stressed, severely stressed, and slightly stressed levels accounted for 5.6, 25.3, and 37.2%, respectively. The damaged area of sparse forest and shrubs accounted for 72.8 and 75.2% of the total area, respectively. The damaged area of plantation and orchard accounted for 75.1% of the total area of plantation and orchard; that of grassland accounted for 70.0% of the total area of grassland, and that of dryland agriculture accounted for 80.4% of the total area of dryland agriculture. The damaged area of paddy field agriculture accounted for 79.9% of the total area of paddy field agriculture (Table 3). Thus, the average percentage of vegetation damage caused by the super typhoon Sarika is 73.4%.

Damages according to different elevations
To evaluate the impact of the super typhoon Sarika on vegetation at different altitudes, we divided the altitude of Hainan Island into 12 levels: 50,100,200,300,400,500,600,700,800,900,1000, and 1100-1800 m. The super typhoon Sarika affected the vegetation concentrated mainly in the areas at an altitude of < 300 m. Owing to the topography of Hainan Island, the elevation area below 300 m accounts for 78% of the total area of Hainan Island. Therefore, most areas with decreased NDVI are mainly concentrated below 300 m. At the extremely stressed level, 47.4, 16.9, 11.8, 6.7, 5.8%, and 4.3% of the areas were within   , 50-100, 100-200, 200-300, 300-400, and 400-500 m elevation, respectively. The area above 500 m elevation accounted for only 6.9% of the seriously affected area (Fig. 7). Therefore, from the perspective of altitude, the impact on the vegetation in medium-and high-altitude areas was relatively less. In addition, the super typhoon Sarika mainly passed through Hainan's central and northern parts, which are relatively flat with low altitudes.

Damages as per different slopes
We further analyzed the impact of super typhoon Sarika on the vegetation along different degrees of slope. Based on the topographic characteristics of Hainan Island, we divided the slope of Hainan Island into seven degrees, namely, 0, 1, 2, 3-7, 8-20, 21-30, and 31-73 degrees. At the extremely stressed level, the affected area with a slope of 8-20 degrees accounted for the highest proportion (211 km 2 ), accounting for 36.2%, followed by a slope of 3-7 degrees (144 km 2 ), accounting for 24.6%. The area affected by the gradient of 2 degrees was 106 km 2 , accounting for 18.1%. Similarly, at the severely stressed level, the vegetation at the slope of 8-20, 3-7, and 2 degrees had a relatively large affected area (988, 771, and 744 km 2 , respectively), accounting for 31.4, 24.5, and 23.6%, respectively. At the slightly stressed level, the affected vegetation areas at 8-20, 3-7, and 2 degrees were relatively large (1334, 1409, and 1154 km 2 , respectively), accounting for 33.4, 23.6, and 19.3%, respectively (Fig. 8).
Therefore, the impact of the super typhoon Sarika is greater not only on the flat land but also on the areas with a slope of 8-12 degrees, followed by 3-7 and 2 degrees. However, the impact was concentrated in the range of 2-20 degrees.

Damages as per different aspects
We analyzed the vegetation condition with slope direction and divided the terrain into 8 directions: north, northeast, east, southeast, south, southwest, west, and northwest. At the extremely stressed level, the areas of affected vegetation in the southeast, south, and east directions were 236, 153, and 153 km 2 , respectively. At the severely stressed level, Fig. 7 Distribution of loss or gain in NDVI along different altitudes the affected vegetation areas in the east, northeast, and north directions were relatively large (1059, 983, and 905 km 2 , respectively). At the slightly stressed level, the affected vegetation areas in the northwest, southeast, and northeast directions were larger (1249, 1099, and 1069 km 2 , respectively). Additionally, many affected areas were observed in the north and west directions. Therefore, at the extremely stressed level, the effect of slope direction on vegetation damage was not pronounced.
Overall, the super typhoon Sarika greatly impacted vegetation in the southeast, south, and east. This is mainly because the typhoon landed in Wanning, Hainan Island, which exhibits a relationship with the direction of the movement of typhoon. In addition to the aforementioned 8 directions, the typhoon had a great impact on flat land vegetation. The impact proportion of the typhoon on the flat land vegetation at extremely stressed, severely stressed, and slightly stressed levels was 20.8, 25.6, and 22.6%, respectively (Fig. 9). The flat land is mainly planted with crops and thus are more fragile than forests and vulnerable to typhoons.

Discussion
Ecosystems are complex and undergo adaptive changes in response to continued physical and biological damages (Elmqvist et al. 2003;Fath et al. 2004). Natural calamities such as typhoons, mudslides, and floods play an essential role in disrupting tropical vegetation. Many assumptions concerning the relative importance of natural disasters warrant reexamination (Ayala and Yaw 2004). After the occurrence of the super typhoon Sarika, the field investigation revealed that the forestry and plantation were seriously damaged, many plants were uprooted, and fallen trees, broken branches, and fallen leaves overwhelmed other plants. The survey resulted indicated that the super typhoon Sarika caused severe disasters and affected 228 towns from 19 counties in Hainan Province, affecting a population of 3 million. Overall, 6,59,500 people were moved, 837 houses collapsed, and 3,80,700 ha crops were damaged. Additionally, the direct economic loss of 4.6 billion RMB yuan was reported. More than 30% of the trees were uprooted from one-third of the tree height, more than 15% of the trees were knocked down, and more than 35% of the trees were broken at the side branches and top (NDRCC 2016;NCC 2017;Liu and Huang 2017;Li et al. 2020), thereby confirming our remote sensing assessment results that the typhoon Sarika had a serious impact on vegetation. Disturbance promotes the succession of forest systems; eliminating some tree species due to such damages accelerates degradation and promotes the development of new tree species (Chen and Fu 2000;Lugo 2008;Jing et al. 2014). Additionally, such damages result in changes in the light, temperature, nutrients, and moisture conditions (Lin et al. 2003;Ostertag et al. 2003;Ueda and Shibata 2005).
We analyzed the relationship between wind speed and the local ΔNDVI. The correlation coefficient between the wind speed and ΔNDVI was − 0.208, indicating a significant negative correlation at the level of 0.01. This result indicated that wind speed was significantly and positively correlated with the affected vegetation index and that the typhoon Sarika had a great influence on vegetation in Hainan Island. The typhoon's impact can be attributed to the strong cyclonic wind causing defoliation, instantaneous litterfall, and uprooting of trees, contributing to an adverse change in the NDVI. The affected forest area and degree drastically increased when wind speed exceeded a certain level, below which forest damages were minimal and independent of wind speed (Mandal and Hosaka 2020). Cyclones with maximum sustained wind speeds of approximately 100 km/h or more can cause largescale forest disturbance (Frances and Gillespi 1993;Tanaka 2008). In this study, the maximum wind speed of Sarika exceeded 32.1 m/s (115.2 km/h), and the affected area reached 79% (Table 2), indicating that our research results are reasonable.
The magnitude of ecological damage from a typhoon depends on wind speed, prevailing wind direction, topographic position in the landscape, and the successional structure of vegetation (Sawaid et al. 2020). We assessed the impact of the typhoon Sarika on vegetation, mainly by considering the impact of strong winds, while ignoring the impacts of heavy rains and flash floods, mudslides, and landslides caused by a rainstorm. The impact of typhoons on vegetation is mainly related to the influence of wind speed, altitude, slope, aspect, and other factors, all of which affect the wind speed, indicating that there is a relationship between high wind speed and vegetation damage. Our study also revealed that the extent of damage is greater at lower topographic positions than at higher elevations. The elevation pattern is consistent across spatial scales, including variation in slope across the landscape, possibly because the intensity of typhoons declines quickly in rough topography as they move upward. Different aspects have been reported to exhibit different disturbance frequencies (Chi et al. 2015;Peerman et al. 2020).

3
This study provides a quantitative method for assessing the effect of typhoons on vegetation by using remote sensing data. However, this study has some limitations. The most critical limitation is related to phenology. Hainan Island has a tropical marine climate with a slight temperature difference. Plants at Hainan Island are evergreen. NDVI varies slightly between September and November and shows deviation within 2% between November and September. In this study, before-and after-typhoon images were obtained within less than 2 months, thereby minimizing phenological differences.
Furthermore, a general phenology curve of vegetation types in the study area indicated a decline in greenness from late September to early November, further reducing the bias due to phenological changes. A previous study revealed that a window period of 60 days before and after a wind disaster is ideal for monitoring, with complete image coverage and high sensitivity of monitoring indicators. It has been suggested that the maximum value of the pre-disaster time-series images and the median value of the post-disaster images should be assessed to obtain complete cloud-free images of the study area, and the differences between the images before and after the disaster should be compared (Kang et al. 2005;Chen et al. 2021).
Another limitation of the study is the cloud-free, remote sensing image acquisition. The perennial, cloudy, and rainy weather in the tropics seriously affects the acquisition of optical images. Time-series image synthesis is a useful method for rapidly assessing typhoon disasters. The maximum value synthesis is better than the average value synthesis. Typhoons occur in the vegetation growing season; thus, recent images for the maximum value synthesis method can better reflect the growth state of vegetation before the typhoon disaster and provide greater difference values than post-disaster images. Moreover, the maximum synthesis method is superior to the nearest value synthesis method because the maximum synthesis method can effectively reduce the effects of cloud, solar angle, water vapor, aerosol, and directional surface reflectivity (Holben 1986). Although images obtained at a time window longer from the disaster are cloud-free, they are not conducive to disaster monitoring and evaluation. After the disturbance, vegetation growth is rapidly restored and NDVI increases, which offsets a part of the affected information.

Conclusions
Tropical areas experience high-frequency and high-intensity typhoons, and vegetation on Hainan Island is highly vulnerable to typhoon interference. We studied the wind distribution and the relationship among vegetation damage severity, typhoon wind speed, and topographic features of the super typhoon Sarika. Vegetation damage is related to biotic factors, such as vegetation types and vegetation distribution, and abiotic factors such as typhoon intensity, track, topographical characteristics of evolution, slope, and aspect. The key abiotic factors for vegetation damage are typhoon intensity, track, and wind. The use of remote sensing data enhances the speed and accuracy of such assessments over a large area. Our study demonstrates that MODIS images with a 250-m spatial resolution could reflect vegetation damage in Hainan Island at the macro scale. NDVI is one of the most frequently used indices for determining vegetation changes, and multiple NDVI analyses can reveal the variation in vegetation with space and time. This study also shows that remote sensing data can be useful in understanding the impacts of the passage of typhoons. Specifically, the percentage change in the NDVI was significantly correlated with the wind speed. Remote sensing technology for large-scale rapid disaster assessment can be valuable in guiding post-disaster recovery, insurance compensation, and scientific research.