Estimating Cyclone Vulnerability and Its Linkages With Child Mortality Trajectories Along the Bay of Bengal Coast in India: a Geospatial Approach

coastal population of variety of and


Introduction
The global sea surface temperature (SST) has increased throughout the past half-century due to the global greenhouse gas emission, leading to the intensi cation of cyclones over time [1]. Globally, 40% of the world population resides within 60 km from the coast, a driver for the coastal vulnerability [2]. The Intergovernmental Panel on Climate Change (IPCC) has identi ed several hotspots of climate change sensitive region where livelihood depends on subsistence agriculture, shing, aquaculture [3]. Vulnerable places experience damage of property, lives, and displacements. Aside from the direct losses, insecurity of food, lacking opportunity of employment, and health impairments are exacerbated by coastal hazards. Additional socio-economic impact is mainly accrued by developing nations due to low challenges to adaptation and high mitigations strategies [4]. Over the decades, South Asia has experienced recurrent hazards related to tropical cyclones. The area is identi ed as a hotspot characterized by high population density, poverty, low development level, and exclusion [5]. In the north Indian Ocean region, Bay of Bengal (BoB) alone accounts for 7% of all global cyclones of the world [6]. Historically, coastal zones of India have attracted people due to port facilities, prosperity in agriculture, tourism, industries, trades, transport bene ts, and habitable ecology [7,8]. However, these regions are vulnerable zones due to the coincidence of low elevation, storm surges, oods, saline intrusion, coastal erosion [9], and other environmental challenges leading to infectious diseases and mortality [10,11].
The coastal population of Indian subcontinent experience variety of coastal hazards of which cyclone is the most destructive one. With nearly 137 million populations [12] residing along with the BoB coastal districts and higher frequencies of depression-cyclonic storms, the coast is quite vulnerable as compared to the Arabian Sea coast in India due to higher frequency of cyclone (4:1 ratio of cyclones in Bay of Bengal than the Arabian Sea cyclone), low at coastal terrain, high population density, poor knowledge of community, inadequate response and preparedness and absence of any hedging mechanism [13]. Most recently, on 20th May 2020, the super cyclone Amphan lashed with 155 km/h, as tall as a two-story building on the Eastern coast and state of West Bengal in India [14]. The devastation was ampli ed with strong winds, tidal waves, and heavy rains that caused ooding across the deltaic regions, with an estimated economic loss of US$13.5 billion [15]. In the following year 2021, another cyclonic storm, named Yaas, intensi ed in the Bay of Bengal in May, and it has ravaged many parts of the apex of the BoB coast. India has witnessed severe to very severe cyclones: Laila, Helen, Phailin, Hudhud, Vardah, Ockhi, Titli, Gaja, Amphan and Yaas from 2010-2021 [16] As a result of cyclone hazard proneness, the Indian Coastal Zone (CRZ) has demarcated 500 m from the shoreline as coastal vulnerability zones. However, a buffer of 100 km perpendicular from the shoreline is measured for coastal vulnerability assessments considering the coastal geomorphology [17]. Literature suggests that, to escape from subsistence livelihood and poverty, aquaculture and industrialization have been introduced in these coastal regions, which had been the prime loop to coastal vulnerability [18].
Natural hazard vulnerability has a detrimental impact on human health, especially at conception and ending at the start of the third postnatal year ( rst 1000 days of life) [19][20][21][22][23]. Progressively, with the availability of satellite images and radar data, coastal hazard vulnerability assessments have soared [9,17,[24][25][26][27]. However, a handful of scientists has conducted vulnerability assessment as a function of geomorphological, socio-economic, infrastructures, and other vectors [28,29] The use of convoluted indices to understand the vulnerability has been adopted by many scholars like Cutter et al., 2006, who measured Social Vulnerability Index (SVI) using socioeconomic and infrastructural indicators. Kim and Gim (2020) measured ood vulnerability and adaption along the Java coast, integrating the spatial regression model [31]. In India, Mazumdar and Paul (2016) and Sharma and Patwardhan (2008) measured the SVI due to cyclone with the principal component method. Scientists have studied the extreme effect of climate variability on health and mortality trajectories [34,35]. Maternal exposure to cyclone increases the risk of having preterm [36-38] birth and consequently plays critical role in neonatal mortality [39,40]. Past study observed that tropical strom increases the risk of illness, injuries and health needs [41].The risk of disease transmission tends to heighten in developing countries due to population density, inadequate sanitation, and poor health facilities. The outbreak of gastroenteritis and diarrhoeal disease has been documented in the parts of West Bengal, India, after Cyclone Aila, 2008 [42,43]. Other morbidities like acute respiratory diseases, leptospirosis have been reported following a cyclone in Orissa, India, in 1991 [44]. The extreme weather events have a signi cant impact on the health of a child [45,46]. Female infant mortality was 15.1 times higher than males in a natural disaster as observed in Philippines [47]. Furthermore, a cross-country analysis of 12 developing countries indicated an increase in infant mortality due to the cyclone's long-term effect [48]. As cyclones have the potential to cause infant and under 5 mortality, empirical studies suggest that natural hazard-induced child mortality positively responds to higher fertility rates [49,50].
The rich body of literature in this eld mainly focused on different natural hazard vulnerability assessments or aftermath of extreme events. Studies in India are rare in this eld, especially highlighting aggregate measurement of vulnerability based on public data and revealing the associations of vulnerability and child health. The present study, unique in its approach, aims to estimate vulnerability at coastal districts of the Bay of Bengal, consisting of 45 districts within a 100 km buffer zone along the coastline. We made progress on integrating the vulnerability and its effects on health by considering two crucial health indicators, i.e. neonatal and under-5 mortality. The study adds to the literature by providing additional evidence of socio-economic vulnerability as a function of exposure, adaptation, and sensitivity due to cyclones and its linkages to child mortality in India.

Study area
Districts less than 100 km from the coast line were considered coastal districts according to the de nition of MDI [51]. A total of 45 districts were selected for this study (Fig. 1). The temperature of the study area exceeds 30º C and the region experience a high level of humidity. Annual rainfall ranges from 1,000 to 3,000 mm. Most of the cyclones here are developed either in the month of April to May (pre-monsoon season in the apex of BoB coast) or October to November (pre-winter season in the southern part of BoB coast.

Operational de nition and analysis
Vulnerability is de ned as "A function of the character, magnitude, and rate of climate variation to which a system is exposed, its sensitivity, and its adaptive capacity" [52]. Total cyclone frequency, severe cyclone frequency, wind speed and shoreline length were considered in the calculation of cyclone exposure. Additionally, we included three components: sensitivity, demographics, agriculture and health. The components that help to adjust with different shocks like economic capacity and skills & infrastructure were considered to calculate the adaptive capacity as better-off households are able to recover the damages of the cyclones in a short period of time [53,54] (Table 1). The vulnerability index was calculated using the universal normalization technique-Where refers to the normalized index value of the indicator, i represents indicators, j is the coastal district and and indicates the maximum and minimum value of the ith indicator among all the coastal districts. The value of the index varies from 0 to 1. Then, the composite vulnerability index was calculated using the equal weighting approach [55]. After normalizing all the indicators, the vulnerability index was derived as: Where Mj is the vulnerability index Xij is the index value of the ith indicator for district j, and n is the number of indicators considered to represent the index. The fourth round of the National Family Health Survey (NFHS-4), the landmark Demographic Health Survey (DHS) in India, was used to estimate neonatal and under-5 mortality at the district level, considering the information of 69,971children in the study area [56]. The mortality rate was calculated using the information of the date of birth of the child, their survival status and age at death of the deceased child. The synthetic cohort probability approach was applied to estimate the neonatal mortality rate (NMR) and under-5 mortality rate (U5MR) for the ten years preceding the survey using the full birth history information of women aged 15-49. The total population of the districts were taken from the Census of India, 2011.
Scatter plots were used to establish the association between coastal vulnerability and neonatal and under-5 mortality in coastal districts of BoB. Finally, linear regressions, Spatial Lag Model (SLM) and Spatial Error Model (SEM) were used to nd out the empirical association between coastal vulnerability and neonatal and under-5 mortality. SLM and SEM helped to adjust the spatial endogeneity and provide a re ned measure.

Vulnerability
In India, the number of cyclones, especially severe cyclones, increased considerably over the last decade (Fig.  2b). Meanwhile, a large number of people were displaced as a result of these disasters, notably in 2012 (Fig. 2a), when more than 9 million people were displaced. More than 5 million people in India were displaced by natural catastrophes in 2019, accounting for the highest number of new internal movements in the world [57].
The districts located in the apex part of BoB were more prone to coastal cyclone vulnerability than the southern part ( Table 2). The highest degree of exposure was observed in districts of West Bengal. A very high degree of sensitivity was noted in Kendrapara, followed by Gajapati, Thiruvarur and Baleswar because of the higher degree of demographic, agricultural and health sensitivity. Adaptability was higher in the districts of Chennai and Calcutta. This study demonstrated a high vulnerability index in South 24 Parganas district, followed by Baleswar, Purba Medinipur, Kendrapara, Srikakulam and Bhadrak. Chennai had the lowest vulnerability index due to its higher adaptive capacity.  Figure 4 depicts the result of the scatter plot between the coastal vulnerability with neonatal and under-5 mortality rates. The radius of the circles represents the volume of total population of the district. Districts with higher levels of coastal vulnerability had higher rates of neonatal mortality and under-ve mortality. The study found a positive association though the relationship was weaker in the NMR relative to U5MR.

Association of NMR and U5MR with vulnerability
OLS regression was applied to examine the effects of coastal vulnerability on child survival after adjusting other socio-economic components ( Table 3  The result of SEM showed a spatial association between coastal vulnerability and NMR -U5MR. Of the two spatially estimated models, the study found a lower AIC value for the SEM model. With respect to neonatal mortality, the coe cient was largest for coastal vulnerability (β = 44.86, p value = 0.00). Further, the coastal vulnerability was signi cantly associated with under-5 mortality (β = 52.34, p value = 0.01). In the spatial lag model, a 10-point increase in unimproved sources of drinking water across the districts was associated with a 3point increase in U5MR. In addition, diarrhoea among children was signi cantly associated with NMR (β = 1.05, p value = 0.03) and U5MR (β = 1.41, p value = 0.00). Also, unimproved sanitation facilities (β = 0.28, p value = 0.04) was a strong predictor of NMR (β = 0.25, p value = 0.00) and U5MR (β = 0.35, p value = 0.00) in these spatial models. Similarly, a 10-point increase in coverage of child vaccination was associated with a 3-point decrease in NMR and U5MR. We observed that the U5MR model showed a greater AIC value with a pseudo R square value of 0.64 than the NMR model across 45 coastal districts of BoB.

Discussions And Conclusion
IPCC has declared that climate change has a very high con dence in disease's global burden [58]. Further climate change, sea level rise to one meter in the present century has intensi ed cyclone affecting the socioeconomic and culture cohesion through displacement of coastal communities [59]. A large proportion of rural population in India lives near the coastline, depends on agriculture, shery and forestry. Almost every year, moderate to severe cyclone hits the coastal districts of BoB, leading to signi cant cause of economic and human loss. In this context, our estimates highlight several important issues related to coastal vulnerability and its linkages with neonatal-child mortality in BoB coast of India. The paper analyses the vulnerability level of the east coastal districts or BoB coast of India stretching for 1400 km and supporting large number of rural populations as well as some important cities and state capitals.
Findings reveal that northern districts located close to the apex of BoB are more prone to exposure of vulnerability of coastal cyclones as compared to the southern part. The reasons are varying; rst, the topography of the northern part of BoB bordering West Bengal and Odisha are atter than the southern part where eastern hills stand along the coastline; second, the funnel shape apex of the BoB exactly collides with the northern coast (along with Bangladesh), that helps intensi cation of wind speed [9]; third, the urban proportion of southern districts are more and thus it has better adaptive capacity, as revealed in the estimate. Poor adaptive capacity attributed to the high degree of vulnerability in selected districts is mostly rural, agrarian, and generally located in West Bengal, Odisha, and parts of Andhra Pradesh.
The study establishes that the districts with a higher degree of coastal vulnerability experience a higher neonatal and under-5 mortality rate. Further, under-5 mortality is more strongly associated with the coastal vulnerability index in the districts of BoB as compared to neonatal mortality rate. Exposure to climate shocks like cyclones is harmful to young children as the shocks lead to reduced food intake and increased infectious diseases. In addition to economic hardship, medical access and utilization are also compromised during such catastrophize affecting the children disproportionately [60][61][62][63]. Therefore, it is essential to understand whether climate change poses a threat to children's health, and if so, where support is most needed. In the coastal districts, water and waste disposal quality in terms of drinking water and sanitation facility are the major determinants of child's survival. The drinking water and sanitation help to improve people's overall health and quality of life [64-66].
The risk of communicable diseases among vulnerable communities is due to the water and unimproved sanitation facilities [67]. The evidence from Amphan, 2020 affected areas of Bangladesh has reported an increase in the number of diarrhoea and skin disorders due to the usage of polluted pond water as their primary source of water for varying use. In addition, people also suffered from dysentery, jaundice, and eye irritation after the said cyclone [68]. Few studies undertaken in India reveals that excessive precipitation is associated with an increased risk of contracting water borne diseases such as diarrhoea among children under ve [69].
Furthermore, Bhattacharjee et al 2010 has reported an increase of Vibrio uvialis (diarrhoea) along the cyclone affected areas after Aila, 2009 cyclone [70]. Indeed, the present study substantiates the ndings of Dimitrova and Bora (2020) by revealing that in these coastal districts of India, the quality of water and waste disposal in terms of drinking water and sanitation facility are the major determinants of child's survival [69]. So, it is evident that due to coastal cyclones, the water-sanitation facilities are not up to the mark that leads to higher mortality.
WHO and UNICEF recommends measles vaccination and Vitamin-A supplements for children during an emergency [71]. In India, after any major disasters, the government organized mass vaccination campaigns to limit the spread of infection and fatality [72]. However, Mallik et al (2011) identi ed major geographical, infrastructural constraints for mass measles vaccination campaigns in the Aila cyclone-affected areas [73]. In general, India's healthcare infrastructure, manpower and accessibility are inadequate, especially in remote areas [74] and this could promote child mortality in vulnerable districts.
The present study has adopted an innovative approach to estimate vulnerability index. The most common weakness of the pre-existing studies of vulnerabilities in India is the use of satellite data. Here, we used existing credible secondary data from well established sources to compute the index. Through this measure, considering sensitivity, exposure and adaptive capacity, we identi ed the problem, quanti ed it, and assessed the index value in formulating development strategies to reduce the risk and vulnerabilities. In the context of increasing frequency and severity of coastal cyclone, the need of the hour to have better planning and protection strategies for Indian coast to safeguard coastal environment, human health, and livelihoods.
The policies related to climate change and coastal vulnerability in India are lacking in implementation. A recent research on climate change strategies, covering 136 coastal cities in 68 countries, observed the lack of effective climate change adaptation policies with no signs of implementation in nearly half the cases. Six major Indian coastal cities were included in the analysis of which three are from the BoB coast, i.e. Chennai, Visakhapatnam and Kolkata. For six Indian cities, only eight policies were identi ed by the team, which explicit inclusion of plans to prepare for the impacts of climate change; a plan at the national level, six at state and only one at local level (for Gujarat in the West Coast of India). This exhibits the lack of concrete measures at the city government stage [75].
Overall, current policies in India are not adequate to deal with the challenges of coastal vulnerabilities. The issue of displacement is addressed mostly as a post-disaster response, while there is a serious need for more comprehensive national and state polices on forestall managed adaptation for at-risk coastal populations. Early response mechanism and safeguarding households especially those with small children in terms of water sanitation, immunization, safety during such climate catastrophe are essential in areas where cyclone vulnerability is high in India.

Declarations Acknowledgement
We are thankful to the Editor and reviewers for their encouraging comments that helped us to revise the manuscript. The authors would like to thank International Institute for Population Sciences (IIPS), O ce of the Registrar General & Census Commissioner and India Meteorological Department (IMD) for providing publicly accessible data.
Subhojit Shaw: writing and analysis