Recent trends in tropical cyclones over the Arabian Sea and the vulnerability of India’s west coast

The IPCC’s sixth assessment report emphasizes the increase in the frequency of stronger tropical cyclones (TC, categories 3–5) over the last four decades. The Northern Indian Ocean (NIO) exhibits identical trends. The Arabian Sea (AS) has become a breeding ground for TCs that have wreaked havoc on Oman, UAE, Iran, Muscat, and India’s west coast in recent years. Coastal India is extremely vulnerable to TCs due to its unique coastal topography and population density. Most of the research on the North Indian Ocean is focused on the Bay of Bengal (BoB) TCs, which necessitates rigorous studies in the AS. The focus of this study is to improve our understanding of TC activity over the AS by analyzing the frequency of TCs and energy metrics of TC accumulated cyclone energy (ACE) and Power Dissipative Index (PDI) over the last 39 years. Energy metric analysis confirms the increasing intensity of TCs in the AS. Annual ACE and PDI have been increasing in recent years primarily due to monsoon and post-monsoon TCs, while pre-monsoon TCs show a decreasing trend. The effect of El Niño Southern Oscillation (ENSO) on TCs in the AS is found to be negligible. The study also attempts to ascertain the vulnerability of western coastal districts to TCs using housing damage risk data from the Vulnerability Atlas of India. Ahmedabad and Vadodara in Gujarat, Thane and Mumbai Suburban in Maharashtra, Dakshin Kannada in Karnataka, and Malappuram, Ernakulam, and Thiruvananthapuram in Kerala are among the most vulnerable districts along India’s western coast.


Introduction
Tropical cyclone (TC) is one of the most frequently occurring hydrometeorological disasters in tropical and subtropical latitudes, wreaking havoc on sea-based regions and the coast (Wahiduzzaman and Yeasmin 2019). Numerous dangers accompany TCs, including wind (Zhou et al. 2018), storm surges (Kohno et al. 2018), and widespread rainfall events (Knutson et al. 2019). TCs are the most destructive natural disasters on the North Indian Ocean's (NIO) rim, claiming thousands of lives and causing billions of dollars worth of property damage (Alam and Collins 2010;Girishkumar and Ravichandran 2012). Around 40% of the world's population lives within 100 km of the coast (Barua et al. 2021) making the coastal region extremely vulnerable to TC events. India's coastline is approximately 7516 km long and forms part of the coastal zone adjacent to the NIO (encompassing the Arabian Sea (AS) and the Bay of Bengal (BoB)). According to the national disaster management authority's broad-scale assessment, approximately 32 crore people (onethird of the total population) is vulnerable to cyclone-related hazards in India. The NIO is the least active cyclone region (~ 6%) among the ocean basins yet these TCs cause the largest casualties (Neetu et al. 2017). Due to India's narrow continental shelf, flat coastal topography, high population density, geographical location, and physiological characteristics (large deltas, lagoons, and mangroves on the east coast; sandy beaches, rocky headland, and large saline marshlands on the west coast), it is extremely vulnerable to cyclones and related dangers (NDMA 2017;Mukhopadhyay and Karisiddaiah 2014).
In developing countries such as India, the coastline population is growing primarily due to the urbanization of rural areas. Zhang et al. (2018) state that urbanization can aggravate the risk of the coastal population to extreme events. In a climate change scenario, TCs and their associated storm surges pose significant challenges particularly affecting the coastal community (Rao et al. 2020). According to recent IPCC reports, the proportion of stronger TCs (categories 3-5) has increased over the last four decades, and the average maximum wind speed of a tropical storm in the Atlantic and Pacific Oceans is projected to increase by 2-11% over the next century (IPCC 2021(IPCC , 2013. Numerous studies have revealed a global increase in the frequency and strength of TCs in recent times. The North Atlantic Ocean is observed to have the greatest increase in TC frequency and intensity among the ocean basins (Deo et al. 2011;Elsner et al. 2008;Webster et al. 2005;Walsh et al. 2016). Other basins, such as the north and south Indian oceans, and the southwestern Pacific oceans, have also seen an increase in the frequency of TCs (Deo et al. 2011). Recent research on TCs in the NIO corroborates the IPCC's findings. Deo and Ganer (2014) find that the intensity of tropical cyclones in the NIO has increased over the last 15 years. Sahoo and Bhaskaran (2016) determined that the accumulation of cyclone energy in the BoB has increased significantly since 1980. Singh et al. (2000) report an increase in the prevalence of TCs in the BoB, but not in the AS. Balaguru et al. (2014) find that the intensity of post-monsoon TCs in the BoB has increased in conjunction with rising sea surface temperature and upper ocean heat content. Many more studies have shown an increase in TCs in the NIO (Balaji et al. 2018;Deshpande et al. 2021;Bhardwaj et al. 2019;Sattar and Cheung 2019). Singh et al (2020) discovered the unusual tracks of TCs over the AS and rapid intensification are a result of a warmerthan-average ocean.
Most of the NIO research is focused on the BoB TCs, which are known for their cyclone activity and damage to the Indian coast. In comparison to the AS, TCs are approximately four to five times more abundant in the BoB (Bhardwaj et al. 2019) approximately 10-20% with a mean frequency of 1.33TCs each year (Soltanpour et al. 2021). Following the history of the AS TCs, the first cyclone in the modern record to make landfall on the Somalia coast was a very severe cyclonic storm (VSCS) in the AS in 1994. Though the possibility of a cyclone making landfall over the AS coast is low, several strong TCs have recently approached these coastlines, including Gonu (2007) (Bandyopadhyay 2010). A few TCs had a significant impact on the low-lying coastlines of Oman, the United Arab Emirates (UAE), and Iran (Vieira et al. 2021). SuCS Kyarr (2019), though it dissipated over the sea, posed damage to the agriculture sector in Maharashtra owing to the un-seasonal extremely heavy rainfall (IMD 2020). The track of all TCs detected in the AS during the study period is depicted in supplemental figures (Fig. S1). Researchers (e.g., Deshpande et al. 2010;Mukhopadhyay et al. 2011;and Osuri et al. 2012) have built atmospheric models to aid in forecasting TCs over the AS. Some publications have recently identified an increase in the intensity of TCs in the AS basin during the last few decades. Murakami et al. (2013) predicted a significant increase in TC frequency over the AS (by 46%) based on Meteorological Research Institute Atmospheric General Circulation Model (MRI-AGCM) version 3.1 (MRI-AGCM3.1) and 3.2 (MRI-AGCM3.2). According to Murakami et al. (2017), anthropogenic forcing accelerated the growth of extremely severe cyclonic storms (ESCS) over the AS. Studies on the AS (Evan and Camargo 2011;Murakami et al. 2017;Chowdhury et al. 2020) indicate that TCs have become more intense and frequent in recent years, implying that the AS should be given increased attention due to its potential to affect many countries along the coast. Regardless of the magnitude of the AS TCs' impact, the basin has received significantly less attention than the BoB basin. While the AS TCs are investigated in conjunction with the BoB as a component of the NIO TCs, only a few studies have examined the AS as a source of TCs on its own (Zhang and Villarini 2019).
The present study examines trends in the frequency of TCs across the AS from 1982 to 2020 and attempts to gain a better understanding of the distribution and trends of the intensity of TCs (accumulated cyclone energy, ACE, and power dissipative index, PDI). We further classified districts on India's west coast according to their housing vulnerabilities to TCs. This paper is divided into five sections. Following "Introduction," "Data and methodology" describes the data sources and methodologies used to analyze the cyclone parameters and classify cyclone-vulnerable areas. The analysis result and trends in TCs over the AS are described in "Results" and discussed in line with the existing literature in "Discussion." "Conclusion" summarizes the study findings.

Data and methodology
The AS TCs are examined in this study over 39 years from 1982 to 2020. This is a significant and stable period, as it is the post-geostationary satellite era (Deshpande et al. 2021). The frequencies of tropical cyclones are derived from the Indian Meteorological Department's Cyclone e-Atlas (IMD). The storm's intensity (maximum sustained wind speed) and details about the cyclone's origin and landfall are derived from IMD's best-track data for cyclonic storms (version 2.0/2011 version) (https:// rsmcn ew. delhi. imd. gov. in). As this study is concerned with the disaster potential of TCs, systems with a maximum sustained wind speed (3 min) of less than 34 knots are excluded. The categories used in this study are based on those used by the IMD (IMD classification is given in Table S1 of supplementary information). To calculate the roof-type vulnerability and wall-type vulnerability, housing risk data is collected from the Vulnerability Atlas of India (BMTPC 2019).

Accumulated cyclone energy and Power Dissipative Index
ACE, introduced by Bell et al. (2000), is a measure of TC energy and is considered a measure of storm kinetic energy. It accounts for the combined duration and strength of the TC. It is defined as the sum of the squares of the 6-hourly maximum sustained wind speed (knots) of the storm. In the present study, it is calculated by adding up squares of the six-hourly maximum sustained surface wind speed (knot) and dividing by 10 4 to make it an index for easier interpretation (Camargo and Sobel 2005).
where V max is the maximum sustained wind squared at all 6-hourly periods obtained from IMD best-track data.
To characterize the destructive potential of TCs, PDI (Emanuel 2005) is used in this study and is calculated as, This index is approximately proportional to the damage potential of a TC. Both energy metrics consider the collective effect of frequency, intensity, and duration of all the TCs for a particular year. ACE ( 10 4 kt 2 ) accounts for both the strength and duration while PDI ( 10 6 kt 3 ) is more closely related to more on TC intensity.

Mann-Kendall test and Sen's slope estimator
The non-parametric Mann-Kendall test (Mann 1945;Hoeffding 1957) is employed to detect a monotonic upward or downward trend in the time series of the AS cyclone frequency. The detail of the analysis is given in supplementary information. The value of Z (normalized MK statistics) determines an upward or downward trend. If Z yields a positive value, then there is an increasing trend in the time series, and if Z yields a negative value, it means there is a downward trend. Sen's slope estimator reveals the steepness of the trend. This study uses the vulnerability atlas to identify the number of houses with lightweight sloping roofs (L), heavyweight sloping roofs (H), and flat roofs (F), and a weighted average is calculated for each district to calculate roof-type vulnerability (RTV). Because the level of resistance varies for distinct types of roofs under different wind conditions, the study assigns different weights to each roof type when along with the trendline which shows an increasing trend in the frequency of TC over the period evaluating RTV. A level roof is given a 0.5 weight, a lightweight sloping roof is given a 1.5 weight, and a heavyweight sloping roof is given a weight of 1.

Calculation of the vulnerability of a location
where R1, R2, and R3 are the percentages of houses with L, H, and F roof types respectively. L, H, and F here denote the number of houses with respective roof types. The RTV for each district is calculated and divided by 10 5 to construct an index for easy comprehension. According to the RTV values, the districts are classified as high risk, medium risk, or low risk.
To calculate wall-type vulnerability, the number of houses with mud and unburnt brick walls (A1), wood walls (C2), and other materials (X) is identified from the data. The weighted average of this data gives the wall-type vulnerability of a district.
where W1, W2, and W3 are the percentage of houses with A1, C2, and X wall types respectively. A1, C2, and X here denote the number of houses with respective wall types. To produce an index for easier comprehension, the WTV for each district is calculated and divided by 10 4 . According to the WTV values, the districts are classified as high risk, medium risk, or low risk. Table 1 shows the risk levels for RTV and WTV. The index values were used to group districts on India's west coast. Separate plots for roof-type vulnerability and wall-type vulnerability were created using QGIS 3.16 for the analysis.

Frequency of TCs (1982-2020)
The frequency of all TCs from 1982 to 2020 is depicted in Fig. 1 for five distinct storm types: cyclonic storms (CS), severe cyclonic storms (SCS), very severe cyclonic storms (VSCS), extremely severe cyclonic storms (ESCS), and super cyclonic storms (SuCS). (This classification is based on the IMD cyclone classification system, as shown in Table S1 in the supplemental information.) A total of 45 TCs formed in the AS during the period. The AS is found to be active in most years with at least one TC and up to five TCs per year while it did not experience TCs in the years 1983, 1984, 1986-1991, 1997, 2000, 2005, 2008, 2013, 2016, and 2017. The BoB is known to have the highest concentration of TCs in the NIO basins. However, there have been some recent years in which the AS TCs outnumbered the BoB TCs (2001,2004,2014,2015, and 2019) as well as years in which both the AS and BoB TCs had an equal share of the NIO TCs (1993, 1994, 1998, 2007, 2011, and 2012). During this period, 17 (38%) of the 45 TCs that formed are CS, 10 (22%) are SCS, 7 (16%) are VSCS, 9 (20%) are ESCS, and 2 (4%) are SuCS. The trend line in Fig. 1 shows that the number of TCs in the AS has been increasing over time. At the significance level of 0.05, the total frequency of TCs shows a positive trend with a slope of 0.0447 ± 0.016. To ascertain the recent trend in TC frequency, we divide the 39-year period into two segments: the former epoch (FE), 1982-2000, and the modern epoch (ME), 2001-2020. Figure 2a illustrates the frequency of TCs in FE and ME. ME has 31 total TCs, compared to 14 in FE, a 54.8% increase. It is concerning to note that ME had two SuCS, while FE had none. From FE to ME, VSCS indicates a maximum rise of 60%. From FE to ME, ESCS exhibits a 50% growth. In comparison to FE, ME had a 58% rise in CS. SCS indicates a minimum of a 33% increase from FE to ME when compared to other categories. As observed from Fig. 2a, the number of cyclones intensifying to VSCS or higher has increased. Figure 2b illustrates the seasonal distribution of TCs over the AS. The occurrence of TCs is demonstrated to be negligible in January and February. The cyclone season in the AS begins in the pre-monsoon season (March-April-May); May (9) has the highest occurrence of tropical cyclones during the pre-monsoon season. March and April are devoid of TC incidents. The monsoon season (June-September) is responsible for a sizable portion of the annual TC frequency. Cyclones occur most frequently during the monsoon season in June (10) and September (4). July and August have no TC events.
Post-monsoon season TCs contribute the most to the annual frequency of TCs (October to December). TCs are most prevalent in October (12), followed by November (8) and December (2). Thus, the AS is most active during the post-monsoon and monsoon seasons, particularly in May, June, October, and November.
To establish the trend in the frequency of TCs over the AS, the Mann-Kendall (MK) test is used. The results of the MK test to determine the frequency of cyclones over the AS are reported in Table 2. The MK test validates the existence of a nonlinear increasing trend in the annual frequency of TCs in AS with a 95% confidence level. The annual frequency of TCs has a Z value of 2.4004 and a p-value of 0.016. The considerable annual increase in the frequency of higher-category TCs is due to an increase in the frequency of ECSC (Z = 2.0987 and p = 0.035). This conclusion is consistent with recent ESCS events such as Nilofar (2014), Chapla (2015), Megh (2015), Mekunu (2018), and Maha (2019). The MK test did not reveal any significant trends in the frequency of CS, SCS, VSCS, or SuCS because the p-values for any of these categories are more than 0.05. Thus, the MK test indicates that the annual frequency and frequency of ESCS have a substantial upward trend, whereas the other four categories exhibit no monotonic trend. Figure 3 shows the frequency of TCs by category in the AS and the BoB from 1982 to 2020. Figure 3a illustrates the frequency of CS. Both basins are trending down in the frequency of CS, with the BoB experiencing a greater decline. The frequency of SCS is depicted in Fig. 3b; a sharp decline in the frequency of the BoB SCS can be observed. The AS SCS also exhibits a downward trend. Figure 3c illustrates VSCS, which depicts the frequency of higher-intensity TCs. Surprisingly, while the frequency of VSCS appears to be decreasing for the BoB, it appears to be increasing for the AS. After 2010, the AS VSCS slope exhibits a more pronounced increase. In terms of ESCS frequency, the AS has a significantly increasing trend, whereas the BoB TCs do not (Fig. 3d). SuCS frequency is shown in Fig. 3e, which is stationary in both the AS and BoB. Super cyclones were confined to the BoB; there were no super cyclones in the AS until after 2007. In the 6 years from 2014 to 2020, five ESCS and four VSCS occurred in the AS, out of a total of nine ESCS and seven VSCS over 39 years. During this period, the AS was also struck by two SuCS, transforming it into a hotspot for strong cyclonic activity. The declining trend in CS and SCS in the BoB and the AS suggests an increased proclivity for TCs intensification, which increases the frequency of VSCS, ESCS, and SuCS. The MK test performed for BoB found a statistically significant negative trend for the frequency of SCS (Z = − 2.8622). The trend for other categories was not statistically significant.    Fig. 4. Both the ACE and PDI trendlines in Fig. 4a demonstrate an increasing trend in the annual values of the parameters. The rising trend can be confirmed with 95% confidence, and the ACE slope is 0.368 ± 0.13 and that of PDI is 0.305 ± 0.12. This conclusion is comparable to the rising trend in the frequency of higher-category TCs reported in "Frequency  of TCs (1982TCs ( -2020." To examine variations over time, the ACE and PDI values for FE and ME are plotted. The values of ACE and PDI are observed to increase during ME as shown in Fig. 4a. The ACE plots for FE and ME are shown in Fig. 4b and c, and the PDI plot is given in the supplementary figures (Fig. S2). It is determined that the maximum value of ACE (PDI) during FE is 11.29 × 10 4 kt 2 (11.11 × 10 6 kt 3 ) whereas the maximum value of ACE (PDI) during ME is 51.22 × 10 4 kt 2 (44.56 × 10 6 kt 3 ). These findings indicate that the energy of TCs has recently increased, as has their destructive potential. The slope of the pre-monsoon ACE (PDI) is decreasing in ME, whereas the post-monsoon ACE (PDI) exhibits a significant upward slope of 0.757 ± 0.35 (0.689 ± 0.33). The annual increase in ACE (PDI) levels in AS is attributed to post-monsoon ACE (PDI) values.

TC energy metrics
The non-parametric MK test and Sen's slope estimator are used to analyze ACE and PDI trends in the data. The MK test results are summarized in Table 4. The MK trend test demonstrates a 95% statistically significant increase in annual ACE and PDI. Annual ACE and PDI both exhibit a significant upward trend, with magnitudes of 0.0436 and 0.017 (Sen's slope estimator values), respectively. The significant increase in annual ACE (PDI) is associated with an increase in VSCS and ESCS in AS, which is evident in ACE from 2014 to 2020. Pre-monsoon and post-monsoon seasons also exhibit a rising trend, but this is not statistically significant. The MK test, on the other hand, indicates a positive trend for both ACE and PDI during the monsoon season with a 95% confidence interval.
As a result of an increase in post-monsoon and monsoon season TC activity, the energy metric of TCs in the AS has been increasing over the study period. Increases in these characteristics indicate that the wind speed, duration, and severity of TCs in the AS, as well as their destructive potential, have increased over the study period. Sen's slope Fig. 4 a. Representation of ACE and PDI from 1982 to 2020. The area under the graph for both ACE and PDI shows an increase in the energy metrics over the time period indicating the increase in the devastation potential of TCs over the period. b. Depiction of FE annual ACE, pre-monsoon ACE, post-monsoon ACE, and monsoon ACE with trendlines. c. Depiction of ME annual ACE, pre-monsoon ACE, post-monsoon ACE, and monsoon ACE with trend lines. The figure clearly shows that the ACE values for all seasons and annual values increase many folds in ME compared to FE. Even within ME, the values are extremely high after 2014

Monsoon and TCs
The Indian summer monsoon (June to September) is wellknown for depressions and deep depressions. The present study finds that there are more TCs in the AS during the monsoon (JJAS), particularly in June, during ME than FE. Supplementary figures (Fig S3) explain the increase in ACE and PDI values in the AS during the monsoon season. The number of TCs (6), ACE (slope = 0.127 ± 0.16), and PDI (slope = 0.068 ± 0.13) increases more rapidly during ME than during FE. Also, the MK test for seasons showed a significant positive trend in monsoon values of ACE and PDI. As discussed in "Frequency of TCs (1982-2020)" and "The Arabian Sea and the Bay of Bengal comparison," monsoon TCs contributed a significant amount to annual TC frequency, ACE, and PDI during this period, as illustrated in Fig. 5. ACE and PDI are depicted in Fig. 5a and c in FE, respectively, while Fig. 5b
The Niño 3.4 values were then correlated with the frequency of TCs, ACE, and PDI to determine if there is a correlation between Niño 3.4 sea surface temperature anomaly Fig. 5 Illustration of monsoon ACE and PDI in FE and ME. a. FE ACE with an increasing trend. b. ME ACE with a steep slope indicating the increasing trend. c. FE PDI with a slightly increasing trend. d. ME PDI with a steep increasing trend and AS TCs. There was no correlation between Niño 3.4 values and TC parameters that was statistically significant (at the 95% significance level). Pearson's correlation coefficient of TC frequency (0.075), ACE (0.142), and PDI (0.131) is not statistically significant. This concludes that the ENSO events have no discernible effect on the frequency of TCs, ACE, or AS PDI. Examination of ENSO demonstrates that we cannot attribute the occurrence of active TC in AS solely to this large-scale climate variability. Additionally, the distribution of TCs is random. For example, the most active TC year in AS was 2019 with five TCs, a neutral year; 2018 was another active year in AS with three TCs, an El Niño year; and 1998 was a La Niña year in AS with three TCs, also a neutral year. The characteristics of SSTAs and TCs are depicted in Fig. 6.

Vulnerability of districts on the west coast of India based on roof and wall types
This study has so far focused on the rising frequency and intensity of TCs across the AS. According to IPCC (2014), a location's vulnerability is directly related to its exposure to a hazard. And given the growing sensitivity of the AS to TC occurrences, it is inevitable to examine the risk in India's western coastal districts. The vulnerability of each district in the western coastal states using the RTV and WTV indices is depicted in Fig. 7. The districts on the west coast that are being considered are listed in Table S2 of supplementary materials. The RTV and WTV indices for each district are calculated, and the districts are classified accordingly.
Gujarat: Gujarat is a large state with cyclone-prone districts. Ahmedabad, Surat, Rajkot, and Vadodara are classified as moderately vulnerable, whereas Junagadh, Kutch, Bhavnagar, Amreli, Jamnagar, Anand, Navsari, Valsad, Bharuch, and Porbandar are classified as lowly vulnerable according to the type of roofing material used. Ahmedabad, Bhavnagar, Anand, and Vadodara, according to WTV, are moderately vulnerable due to the type of wall. Junagadh, Kutch, Amreli, Jamnagar, Surat, Valsad, Bharuch, Porbandar, and Rajkot all have a low-vulnerability level. The proportion of houses with unburned brick walls, wood walls, and other lightweight materials is greater in moderate-risk districts than in low-risk districts. Ahmedabad and Vadodara districts face a moderate risk when compared to the rest of Gujarat's coastal districts.
Maharashtra: the RSMC classifies Maharashtra's coastal districts as P3 districts, which means they are moderately prone to cyclones. This study finds that Thane and Mumbai Suburban are considered high-risk areas by the RTV index. Mumbai Suburban has a remarkably high RTV index value (10.7). Ratnagiri is the next most vulnerable. Other coastal districts such as Sindhudurg and Raigarh exhibit low vulnerability as indicated by the RTV index. Districts are classified differently in the WTV index. According to the WTV index, Thane and Raigarh are highly vulnerable, while Mumbai suburban and Ratnagiri are moderately vulnerable, Goa: the RSMC classifies Goa as cyclone-prone in the P3 category. This study discovers that both North and South Goa fall into the low-vulnerability category according to the RTV and WTV indices. There is only a very slight difference in index values between North and South Goa.
Karnataka: Udupi, Uttar Kannada, and Dakshin Kannada are the three coastal districts in Karnataka. On the RSMC's cyclone proneness map, all these districts are classified as P3. All three districts are classified as low-vulnerability districts by our analysis of the RTV. However, Dakshin Kannada is classified as having moderate vulnerability in the WTV classification, making it the most vulnerable district in Karnataka.
Kerala: Nine of Kerala's fourteen districts are coastal. This fact increases the state's vulnerability to hazards associated with cyclones. However, the RSMC classifies the districts of Kozhikode, Malappuram, Thrissur, Kannur, Kollam, Alappuzha, and Thiruvananthapuram as P3 (moderately prone). Ernakulam and Kasaragod are classified as P4 areas (less prone). This study identifies Malappuram and Ernakulam as moderately vulnerable districts by the RTV index, whereas the remaining districts are classified as less vulnerable. The WTV index classifies Thiruvananthapuram as moderately vulnerable, while the remaining districts are classified as less vulnerable.
Daman and Diu, Lakshadweep: Daman is classified as a P3 vulnerability by the RSMC, while Lakshadweep and Diu are classified as P2 vulnerabilities. These areas have a significantly lower population and housing density than India's western coast states. As a result, both the WTV and Fig. 7 a. Vulnerability of districts on the west coast of India based on roof types. Districts marked red are highly vulnerable, yellow is moderately vulnerable, and green depicts low vulnerability. b. Vul-nerability of districts on the west coast of India based on the wall type. Districts marked red are highly vulnerable, yellow moderately vulnerable, and green low vulnerability RTV indices used in this study classify the vulnerability of these union territories as low risk.

Discussion
Our study on the TCs over a period of 39 years, from 1982 to 2020, reveals an increase in the frequency and intensity of TCs over the Arabian Sea. This study demonstrates that the frequency of TCs is increasing as they are becoming more intense as explained in "Frequency of TCs ." We find that in ME the number of TCs belonging to higher categories (VSCS, ESCS, and SuCS) is remarkably large. The frequency of VSCS is observed to increase by 60% during ME. It is also identified via the MK test that there is a significant upward trend in the frequency of ESCS and this is evident in recent occurrences of TCs in the AS from 2014 to 2020. This result is consistent with a trend towards stronger storms projected by Knutson et al. (2010), who predicted a 2-11 percent increase in storm intensity by 2100. According to Knutson et al. (2013) and Kossin et al. (2013), the frequency of categories 4-5 systems may increase by approximately 40% by 2100. Sahoo and Bhaskaran (2018) find that out of four to five TCs occurring in the NIO one or two are very severe. Many studies (Deshpande et al. 2021;Murakami et al. 2017;Balaguru et al. 2014) reveal an abrupt shift towards intensification of TCs in the NIO, resulting in an increase in the number of VSCS storms in the NIO basins throughout cyclone seasons, especially in the BoB. This study observes a similar trend in the AS indicating that the AS is as significant globally as any other ocean basin.
Our findings show that the AS has a much greater rate of VSCS and ESCS than the BoB. Some numerical model simulations suggest that the frequency of global TCs is expected to decrease, not increase, in the twenty-first century (Knutson et al. 2010). In contrast to this, our study observes a decline in the frequency of CS and SCS but a clear increase in the frequency of higher-category TCs in both the AS and the BoB. We observed that ME witnessed consecutive ESCS incidents in 2015. Chowdhury et al. (2020) observed that the warm ocean surface resulted in the back-to-back occurrence of TCs over the AS which is in line with our observation. Even though this study observes two SuCS in ME, we cannot make any conclusions because the number of SuCS is extremely small in comparison to the number of individual TCs in cyclone seasons.
This study found a significant rise in ACE and PDI in the AS during the period. The increase is more pronounced after 2010, implying that the strength and destructive potential of TCs have increased significantly over the years. This rise in ACE is coherent with the findings of Deshpande et al. (2021) who find a significant increasing trend in the intensity, frequency, and duration of cyclonic storms over the NIO. Additionally, we observe a parallel increase in PDI. Sahoo and Bhaskaran (2016) report that the PDI of TCs in BoB is sixfold that of the past decade. Our study observes a drastic increase in AS PDI. Surprisingly, this study observes a decline in ACE and PDI values during pre-monsoon in ME and a steep increase in post-monsoon values of the parameters. Monsoon values exhibit a similar trend to postmonsoon values, though they are less steep. Thus, the annual increase in ACE and PDI values is attributed to the postmonsoon and monsoon values of these parameters. We also observe that during ME the TCs in monsoon shows abrupt values for ACE, PDI, and frequency. The increase is more gradual during FE. Such unprecedented occurrences amplify the risk associated with TCs and vulnerability. Muni (2009) predicted that the decreasing trend in the tropical easterly jet because of the warming ocean would facilitate the formation of hurricane-intensity TCs in the NIO during the summer monsoon. Our findings validate this prediction. A recent study (Dhavale et al 2022) has pointed out that the poor mean monsoonal wind circulation generated by the southward influx of subtropical westerlies is conducive to cyclogenesis in the AS.
May, June, October, and November are the most active TC months in the AS, according to our results. Evan and Camargo (2011) hypothesized that TCs in May and June in the AS were associated with the early and late onset of the southwest monsoon, respectively. Studies such as that of Sattar and Cheung (2019) cover the pre-monsoon season from June to August; this study adheres to the traditional division of months into pre-monsoon, monsoon, and post-monsoon in the Indian context. Previous studies have identified the increase in TC parameters over the NIO giving special mention to the BoB. This study however emphasizes the AS and shows that the AS needs to be treated as active and hazardous as any other ocean basin.
We also observe that ENSO events have a negligible effect on TC frequency, ACE, and PDI values in the AS. This is in contrast with the findings of Koll and Singh (2018) who suggest an ENSO control on the AS pre-monsoon activity using CMIP5 model simulations. Bhardwaj et al. (2019) and Girishkumar and Ravichandran (2012) observe that in BoB Niño 3.4 sea surface temperature anomalies are negatively correlated with TC parameters. Camargo and Sobel (2005) observe a positive correlation between ENSO indices and western north Pacific basin TC parameters. We observe random TC events during various ENSO indices in the AS. This could be because TCs are inexplicable in terms of a single mode of climate variability. Numerous climate variables are at work during a cyclone. Rao et al. (2020) demonstrate that the Indian peninsula is affected by the intensification of TCs. In comparison to India's west coast, almost every state on the east coast is extremely vulnerable to cyclones ).
Many studies have highlighted disaster-prone districts on India's east and west coasts Mohapatra 2015). RSMC has compiled a list of Indian districts that are prone to cyclones. This classification was made solely based on hazard, without regard for the vulnerability of a location. Additionally, studies examined the vulnerability of districts such as Orissa (Dube et al. 2000;Mohanty et al. 2020), Andhra Pradesh (Raghavan and Rajesh 2003), Tamil Nadu (Jeganathan and Andimuthu 2013), and West Bengal (Gayathri and Debabrata 2015). It should be emphasized that these studies focus exclusively on India's eastern coastal states (Rao et al. 2020). Only Gujarat, on the west coast, is known to be particularly prone to cyclones. Our study concentrates on the vulnerability of the western coast of India and observes that Gujarat's Ahmedabad and Vadodara districts, Maharashtra's Thane and Mumbai Suburban districts, Karnataka's Dakshin Kannada district, and Kerala's Malappuram, Ernakulam, and Thiruvananthapuram districts are the most vulnerable districts along the country's western coast. It is worth noting that some districts classified as highly prone in the RSMC list fall into the less vulnerable category according to the RTV and WTV indices. This discrepancy arises because RSMC considered only the hazard aspect of TC events. The vulnerable districts according to our classification are densely populated and rapidly urbanizing, which contributes to a locality's increased risk during TC events.

Conclusion
In recent years, the Arabian Sea has become a hotspot for TCs, with dire socioeconomic consequences for India's west coast. The purpose of this study is to gain a better understanding of tropical cyclone activity over the AS by analyzing the frequency of TCs and explaining observed higherintensity TCs in terms of accumulated cyclone energy and power dissipative index. Our major findings are as follows: • The frequency analysis of TCs reveals an increase in the annual frequency of TCs over the AS from 1982 to 2020. Between FE and ME, VSCS increases by 60% and ESCS by 50%. Additionally, the frequency of CS and SCS has decreased over time, indicating a shift in the AS towards higher-intensity TCs. • Energy metric analysis confirms the increasing intensity of TCs over the AS. It is discovered that the increase in post-monsoon and monsoon TCs contributes to the annual TC parameter increase (though post-monsoon trend is not significant in the MK test). Increases in these characteristics indicate that the wind speed, duration, and severity of storms are increasing. The pre-monsoon parameters of TCs show a decreasing trend during ME, whereas the monsoon values show an increasing trend (June). • Monsoon TCs contributed a significant amount to annual TC frequency, ACE, and PDI during this period, and this trend is expected to continue. The effect of ENSO events on TC activity in the AS and conclude that ENSO has a negligible effect on TC activity in the AS. Also, the TC occurrence shows a random distribution over El Niño, La Niña, and neutral years. • Analysis for vulnerable districts revealed that Gujarat's Ahmedabad and Vadodara, Maharashtra's Thane and Mumbai Suburban, Karnataka's Dakshin Kannada, and Kerala's Malappuram, Ernakulam, and Thiruvananthapuram are among the most vulnerable districts along India's western coast.
The study indicates that the strength and frequency of TCs in the AS are expected to increase in the future. This scenario will exacerbate the vulnerability of countries such as India, the UAE, Oman, Iran, Muscat, and even Somalia. The risk of damage to a location can be mitigated to a certain extent by using proper building materials, which minimize a house's vulnerability, thereby reducing the vulnerability of a district. The long-term goal of the study is to develop a disaster mitigation strategy for India's coastal regions.