Global assessment of storm disaster-prone areas.

BACKGROUND: Advances in climate change research contribute to improved forecasts of hydrological extremes with potentially severe impacts on human societies and natural landscapes. Rainfall erosivity density (RED), i.e. rainfall erosivity (MJ mm hm-2 h-1 yr-1) per rainfall unit (mm), is a measure of rainstorm aggressiveness and a proxy indicator of damaging hydrological events. METHODS AND
FINDINGS: Here, using downscaled RED data from 3,625 raingauges worldwide and log-normal ordinary kriging with probability mapping, we identify damaging hydrological hazard-prone areas that exceed warning and alert thresholds (1.5 and 3.0 MJ hm-2 h-1, respectively). Applying exceedance probabilities in a geographical information system shows that, under current climate conditions, hazard-prone areas exceeding a 50% probability cover ~31% and ~19% of the world's land at warning and alert states, respectively.
CONCLUSION: RED is identified as a key driver behind the spatial growth of environmental disruption worldwide (with tropical Latin America, South Africa, India and the Indian Archipelago most affected).

Introduction

Although there is a growing need to assess ecosystem responses to climate change-induced disaster risk reduction, there is a lack of research on sensitive areas and on coastal zones, drylands and watersheds, particularly in Global South low-income countries [1, 2]. In order to advance global climate change studies, climate impact indicators are needed to help develop guidelines for landscape conservation planning [3, 4] and support decision-making based on hydrological damage data [5, 6]. Rainfall erosivity density (RED in MJ hm-2 h-1, equivalent to the most common MJ ha-1 h-1), i.e. rainfall erosivity per unit of rainfall, is an important climatic indicator of floods and soil erosion [7, 8]. RED effects are apparent in catastrophic weather events, due to the intensification of daily storms in the northern Hemisphere () and erosive precipitation in other parts of the world (). The effects of more intense extreme rainfall worldwide [9-12] include intensified sub-daily precipitations [13, 14], the local occurrence of flash floods [15] and the incidence of the erosive force of rainfall [16-19].

Emerging catastrophic weather events over the last four decades.

a) Annual mean daily maximum storm depth (mm d-1) in the Northern Hemisphere (mean data of areal maxima from NCEP/NCAR Reanalysis [20]), and b) Global evolution of the erosive force of rainfall (rainfall erosivity) in MJ mm hm-2 h-1 yr-1 (arranged from Bezak et al. [19]). Graph a) covers the period 1981–2018, and graph b) covers the period 1981–2020.

Damaging hydrological events are extreme phenomena, the source of multiple hazardous events with potentially serious impacts on human societies [21]. Among these, floods and events causing flooding are identified as hydrological disasters, which also include meteorological disasters like thunderstorms. Like other extreme phenomena, hydrological disasters typically leave behind socio-economic damage, the severity of which depends on the resilience of the affected population and the available infrastructures [22]. Potentially severe natural events are typically not classified as natural disasters if they occur in areas without a vulnerable population, e.g. deserts [23]. Information on the spatial distribution of RED can help delineate areas prone to multiple damaging hydrological events [24, 25] and, in turn, can transfer important prescriptions for disaster response planning. However, geoinformation and hazard mapping have been and remain critical in modern visual communication science practice [26], particularly in landscape decision-making, where the uncertainty component must be included in the mapping [27, 28]. In particular, it is difficult to represent disaster-prone areas by RED, as different parts of terrestrial ecosystems respond differently to uneven, often nonlinear and universal forcing agents, with threshold-like features [29, 30]. Landscapes are, in fact, highly responsive and non-linear systems to both external dynamics, such as climatic and non-climatic factors, with a combination of gradual changes coupled with infrequent high-magnitude events has led to dramatic landscape responses throughout Earth’s history [31]. Then, the effects of these historical landscape responses to climate extremes and threshold processes are key parameters affecting the geomorphological impacts of extreme hydrological events [32, 33].

Threshold mapping, and the processes of developing probability mapping, is a challenge in geographic information science and spatial downscaling [34]. Scaling and integrating the relative uncertainty of these thresholds, for which storms drive surface flows, including geomorphic effectiveness and erosional soil degradation, affect the spatial pattern of damaging hydrological events, and ultimately improve our understanding of ecological responses to climate extremes and thresholds [35].

The apparent naturalness of a landscape measures the degree to which it is free from storm damage (). Under erosive climatic conditions, the strength, intensity and frequency of a given rainfall influence hydrological processes, even though they may still maintain the shape of the land and the equilibrium of the environmental system in a resilient landscape (). On the contrary, a change in the hydrological regime, especially when the thresholds of an acceptable level of hydrological disturbance are exceeded, can have harmful consequences for the landscape ().

Landscape under different weather regimes with changing RED thresholds for a given level of hydrological disturbance in: a) Storm-free landscape, b) Landscape under storms not exceeding thresholds, and c) Stressed landscape where thresholds are exceeded resulting in damaging hydrological events (image arranged from MeteoBlue, https://static.meteoblue.com/assets/images/crosslinks/yearcomparison.svg).

The prerequisite for RED modelling is that the environmental system adapts to changes in the natural hydrological regime. Usually, the prediction of precipitation and its extremes is performed using physical models, mainly due to the high spatial variability and nonlinearity of the problem [36]. The limitations of physical models (which are mainly computational at the global level) are encountered with very extreme rainfall, when the prediction has to be performed at time-scales less than 30 minutes over many years and for a spatial resolution of a few kilometres. Recently improved convection models can predict extreme rainfall, but simulations are not currently within reach due to their computational cost and degree of uncertainty [37]. Retrieving extreme rainfall from satellite data would be an option to improve the estimation of rainfall erosivity [38], but satellite data require corrections and need further assessment, especially on a global scale [39]. With the advancement of global atmospheric reanalysis data, numerical weather prediction techniques are becoming an encouraging mean of estimating rainfall erosivity [40]. In particular, the production of global RED maps is a challenge due to two conflicting conditions, namely that the analysis requires precipitation data with both high temporal resolution and global coverage. With a large volume of data and uneven distribution of stations, geostatistical methods provide reasonable estimates of what the variable of interest would be at intermediate locations. Geostatistics offers different approaches to deal with this issue and provides attracting results when experimentally determined rainfall erosivity data are available at both regional [41, 42], continental [43, 44], or even global scales [5]. However, geostatistical estimation of RED [17, 45, 46], and its spatial patterns above given thresholds [47], have generally received limited attention [48].

The value of geostatistical probability mapping for geographic information systems (GIS) lies in providing reliable interpolation methods in error assessment and scaling, which can be used in environmental modelling and decision-making [49]. The spatial uncertainty associated with RED hazard over a range of scales is, in any case, an open issue, coupled with uncertainties in downscaling methods and a lack of primary information data in many areas. The uncertainty of RED thresholds actually poses challenges for geospatial assessment, as the worst storms can fall at locations not well-covered by weather recording stations [50]. Global quantification of storms at fine time scales also remains challenging, as hourly or sub-hourly rainfall data of sufficient length are poor, especially for critical and vulnerable regions such as the tropics [51].

Here, we present for the first time the use of the Global Rainfall Erosivity Database (GLoREDa) to estimate RED worldwide [5]. For this study, this database was updated as GLoREDa-V2 and contains erosivity values estimated as RED from 3,625 stations in 63 countries with time resolutions from 1 to 60 minutes. The results obtained are based on climate data and a probabilistic approach to proceed under a soft geovisualisation in order to mitigate the uncertainty involved in downscaling and geocomputational tracking. For this purpose, we used a parametric kriging technique, which provides great flexibility for modelling environmental data [52]. In particular, GIS and log-normal ordinary kriging coupled with output probability mapping (LNOKpm) were used to continuously delineate the spatial uncertainty of RED thresholds and predict areas prone to damaging hydrological events on a global scale.

Methods

Rainfall erosivity data

We refer to annual rainfall erosivity data from the Global Rainfall Erosivity Database (GloREDa, here updated to GloREDa-V2), which covers 3,625 precipitation stations from 63 countries with temporal resolutions of 1 to 60 minutes. It is the result of an extensive collection of high temporal resolution rainfall data from as many countries as possible in order to have a representative sample across climatic and geographical gradients ().

The number of stations varies greatly from continent to continent, with no station-data available above 70° North and below 47° South. However, the latter is not an issue. In fact, the heaviest rains have a low probability of occurring over the northern limit, where rainfall erosivity values are close to zero, while large areas below the southern limit are open sea water or ice-covered (Antarctic ice cap). Precipitation time-series range from a minimum of 5 years to maximum of 52 years (on average, 16.8 years).

GloREDa contains the best available global datasets of annual rainfall erosivity (RE, MJ mm hm-2 h-1 yr-1), in the form of (R)USLE-R factor [53], calculated on a monthly (j = 1, …, 12) basis (REm, MJ mm hm-2 h-1 month-1), from which we obtained rainfall erosivity density (RED, MJ hm-2 h-1) as the ratio between rainfall erosivity and precipitation amount [54]: where n is the number of years recorded; m is the number of erosive events during a given month j; k is the index of a k single event; v is the volume of rainfall (mm) during the r period of a storm, which splits into m parts; I is the maximum 30-minute rainfall intensity (mm h-1); i is the rainfall intensity during the time interval (mm h-1); P is the amount of rain (mm) during a given year y. The numerator in Eq (3) represents the long-term mean rainfall erosivity (MJ mm hm-2 h-1 yr-1), while the denominator is the relative mean amount of precipitation (mm) over the same number of years.

Warning and alert thresholds

We use concepts of a “warning state” and an “alert state” [55] for erosivity density, as regions with high RED are at risk of flooding and water scarcity owing to their infrequent but very intense and erosive rainstorm [19, 56]. As such, RED reflects not only the component of climate forcing reproduced in the aggressiveness of the storm (rainfall erosivity), but also the damaging hazard and its associated hydrological risk [18]. In particular, RED values > 3.0 MJ ha-1 h-1 indicate an increased risk of erosive rainstorms, soil erosion and flooding [57]. This critical threshold value was popularised in storm geomorphology by Dabney et al. [58] as the runoff increases when RED exceeds 3 MJ hm-2 h-1, thus leading to an increasing erosive hazard as storm erosivity represents a large proportion of the rainfall amount in an intense event. Mostly set regionally on a monthly basis [44], this hydrological threshold (alert) could help detecting areas prone to erosion- and overland flow [59, 60]. Abstractions such as thresholds of change and strength are all essential as landscapes may be able to counter or assimilate pulses of change as a form of sensitivity or stability [61]. The use of this abstraction for the way geomorphological systems react to climate variability and change is still an important topic since its conception by Allison and Thomas [62] and Phillips [63]. Given the global scale and annual resolution of this study, a second, smaller RED threshold value (warning) was also used to better capture interactions between changing hydrologic variability and ecological responses. For instance, major tropical forest regions like Amazon and African regions are less vulnerable to hydrological hazards than Southeast Asia [64]. When RED values are > 1 MJ ha-1 h-1, only a certain amount of precipitation can cause relatively high erosivity [57]. Therefore, we adopted two threshold values as follows: RED exceeds the threshold of 1.5 MJ hm-2 h-1 in the warning state and exceeds 3.0 MJ hm-2 h-1 in the alert state. The two values correspond, respectively, to the median and the 3rd quartile of the distribution of the RED data (Fig 4A). These values also mark the range of critical RED values identified by Diodato et al. [8] with a 50-year return period in north-western Italy. The advantage of using RED threshold values compared to other rainfall aggressiveness indicators are [54]: a) greater propensity to classify geomorphological hazards; b) greater stability, obtained with a shorter and heterogeneous period of recording; and c) easier mapping over large areas due to its independence on altitude up to 3000 m a.s.l. [65].

Fig 4

Exploratory data analysis and verification of normality of rainfall erosivity density (RED) data.

a) Distributional frequency on the original RED data and b) after log-normal transformation; c) QQ-Plot of the theoretical and estimated distribution of the original RED data and d) after log-normal transformation. Warning and alert thresholds (MJ hm-2 h-1) are shown in a).

Log-normal ordinary kriging probability map

Kriging is a generic name covering a range of spatial least-square prediction methods [66]. Most kriging algorithms with references to hydrogeological applications were reviewed in Kitanidis [67]. For some kriging models, only one estimate per cell is required; for others, as in decision-making, it is necessary to know the local uncertainty associated with the estimates [68, 69]. The ordinary lognormal kriging algorithm has the potential to improve maps because such techniques can be combined with data transformation and detrending options, and can take different forms, such as maps of probability outputs, quantiles and standard errors of prediction, when the normality of the distribution is verified [52]. In contrast, indicator kriging cannot perform data-transform and detrending, when soft information is required within the probability mapping.

If the prediction in the unknown locations is normally distributed, then the mean and median of the predictions will be positioned centrally in the probability density distribution of each location.

The area under the distribution curve to the right of the threshold line predicts the probability that the value is greater than a threshold value. The distribution of predictions changes for each location as the mean and standard error of the predictions change. Thus, by keeping the threshold value constant, a probability map is produced for the whole area. Since the RED data show a skewed distribution, we used a lognormal ordinary kriging in the form of a probability output map (LNOKpm). The assumption of normality of the distribution was checked after the data were log-transformed. A straightforward approach is to classify as hazardous all locations where the probability of exceeding a critical threshold value, z, is greater than a critical RED value (1.5 MJ hm-2 h-1 for the warning state and 3.0 MJ hm-2 h-1 for the alert state). The ordinary kriging model assumes that the data are a realisation of an auto-correlated process plus an indipendent random error. For a complete analytical procedure of LNOKpm, refer to Krivoruchko et al. [70].

Results and discussion

Exploratory data analysis and transformation

Exploratory data analysis is important to inspect and explore data statistically before deciding whether and how to transform them for analysis and to illustrate what can be achieved by transforming data (e.g. into logarithms) of single variates and calculating principal component analysis of multivariate data [71]. The first step in the spatial analysis is to check the raw data for drifts and outliers [72], and finally for normal distribution. Outliers and spatially-skewed data can be detected by the frequency distribution and the third standardised moment or skewness (g), whose value is 0 for a normal distribution and any other symmetric distribution with finite third moment.

A lognormal transformation was used to approximate the skewed distribution of the RED data (), with g = 2.244 to a normal distribution (), with g = 0.011.

The QQ-plot shows the theoretical and estimated distribution deviating from the normal distribution as the RED values increase (), while a normalisation of the distribution is evident after the logarithmic transformation (). With the normality of the distribution restored, it is possible to extend the log-normal ordinary kriging to probability mapping [73].

The drift analysis demonstrated the existence of a non-random (deterministic) component in spatial distribution of the data: the moderate gradient of RED data occurs from north to south regions of the world. However, we considered that the stationarity hypothesis does not hold for the global, but only locally. Thus, in order to make a robust assumption of homogeneity of variances, the concept of process stationarity has been replaced by a stationarity of governing influence regarding local hydrological processes and nearby local anisotropy. In such a situation, ordinary kriging is recommended for interpolation [74].

Spatial structural modelling

To instruct the kriging interpolation, a regionalisation model was fitted using an iterative procedure developed by Johnston et al. [73], which consists of two steps. Step 1 assumes an isotropic model, and performs an initial run of the experimental spatial structure on the standard deviation-scaled data , where z(s0) is used to denote the jth measurement of a variable at the αth spatial locations s0, and σ is the sample standard deviation. With step 2 any parameter is calibrated interactively, such as: lag number (assumed equal to 7), lag size h (assumed equal to 10 km for the warning state and 20 km for the alert state), range representing the limit of spatial dependence (equal to 112 km for the warning state and 70 km for the alert state, which are comparable to the ranges of spatial dependence of extreme precipitation on a global scale, varied between 54 and 265 km [75].

Isotropic semivariograms were then modelled as a combination of two distinct spatial structures, the nugget variance and a spherical structure, as shown in Eq (4):

The nugget effect 0, equal to 0.066 for the warning state and 0.072 for the alert state, is simply the sum of the measurement error, and the microscale variation of RED, which remains unknown due to the spatial variability associated with the distance between raingauge stations. The value that the semivariogram model reaches at the range (the value on the h-axis) is called the sill (partial sill + nugget), and counts 0.182 for the warning state and 0.189 for the alert state, while h is the distance between the unknown point (h = 0) and a generic point-station.

In this way, shows the experimental unidirectional semivariogram computed from the 3,625 data of RED, with spherical admissible models fitted for threshold-values z (RED > 1.5, and > 3.0 MJ hm-2 h-1).

Modelling of spatial dependence to instructing kriging interpolation.

Experimental semivariogram (dots) with permissible spherical model estimates (violet curve) at a) threshold z > 1.5 MJ hm-2 h-1 and b) at threshold z > 3.0 MJ hm-2 h-1 (b). Units of the semivariance γ are multiplied by 10.

Semivariogram values increase with separation distance, reflecting the assumption that RED data that are close tend to be more similar than data that are farther away. In particular, the spherical semivariogram model fluctuates around the sill value at 1.05° (~112 km) for the warning state (, violet curve) and at 0.70° (~70 km) for the alert state (, violet curve), suggesting that the phenomenon recorded at alert state operates on a smaller spatial scale than in the warning state. This is physically correct as RED events affect a smaller area as they become more extreme, and the physical mechanisms of extreme storm events are size-dependent [76].

The errors involved in the transfer of information from the point to the landscapes via LNOKpm were assessed by means of quantitative standard error estimation and cross-validation [77], re-estimating the RED data at raingauge locations after removing one RED value at a time from the datase [78]. The difference between the estimated and the corresponding actual indicator value is the experimental error. Thus, repeating this estimation for the number of the experimental data n = 3,625, the cross-validation statistics were calculated, as mean error and root mean square errors (RMSE).

Cross-validation estimates the proportion of neighbourhood values that are below or above the threshold value. The high proportions of low values not exceeding the threshold and high values exceeding the threshold are a measure of the success of the kriging estimates Fig 6.

Fig 6

Cross-validation for warning and alert states.

Scatterplots between actual rainfall erosivity density (RED) values above the given threshold and LNOKpm probability for the thresholds a) z (RED) > 1.5 MJ hm-2 h-1 and b) z (RED) > 3.0 MJ hm-2 h-1. The white vertical lines in both graphs represent the respective RED thresholds (a, warning state; b, alert state). The cross-validation scatter diagrams (a and b) show that the actual RED values below and above the given thresholds at the warning and alert states are in agreement with the respective kriged probability.

This is an estimate of the proportion of the values in the neighbourhood that are above the threshold value. The mean error values of -0.005 (warning state) and -0.019 (alert state), and RMSE = 0.262 (warning state) and RMSE = 0.215 (alert state), are close to zero, making it clear that there are no systematic errors.

Global spatial pattern of RED-kriged probability estimation

presents the map of the hydrological hazard-prone areas at the warning state, highlighting the fact that about 31% of the world’s land area has a greater than 50% probability of exceeding the z threshold (RED > 1.5 MJ hm-2 h-1). The map indicates that the phenomenon taken into account by LNOKpm is not smooth (i.e. RED values change strongly with distance). The most affected hydrological hazard-prone areas are Africa and the southern Asian continents, southern Saudi Arabia, Australia, almost all of the USA with an offshoot to western Canada. These figures are consistent with statistics indicating that about 10 million hectares of cropland are lost each year due to soil erosion worldwide [79], particularly in Asia, Africa and South America, where erosion is more severe [80, 81].

Global spatial patterns of kriged-probability map over the period 2002–2011.

Exceedance of the rainfall erosivity density (RED) threshold-value at a) warning state: z (RED) > 1.5 MJ hm-2 h-1, and b) alert state: z (RED) > 3.0 MJ hm-2 h-1.

China’s far southeast, where the probability of exceeding RED at warning state is high, has experienced significant upward trends in rainfall erosivity over the period 1950–2010 [82]. Similarly, in many parts of Africa, where the warning state is expected to be exceeded over a wide area, soil erosion is becoming a major problem due to the high sediment production in tropical mountain streams [83]. The Mediterranean also has a high probability of reaching a warning state, where the aggressiveness of rainfall [8], in tandem with RED, seems to show a propensity to increase in recent decades [18]. However, in addition to the frequent high erosivity of precipitation regimes and human disturbances, it can be argued that much of the Mediterranean landscape is naturally vulnerable to soil erosion processes [84, 85]. , on the other hand, shows the map of the hydrological hazard-prone areas at alert state, highlighting the fact that ~19% of the world’s land area has a greater than 50% chance of exceeding the z threshold (RED > 3.0 MJ hm-2 h-1).

With respect to the alert threshold (), regions that have become free hazard-areas include the Mediterranean lands, almost all of the USA, Japan, Pakistan, northern and central Africa and south-central Australia. Canada, Greenland and Eurasia still remain below the z threshold in warning and alert states, because these regions have a climate less exposed to the critical level of RED, and where rainfall erosivity is also lower with a decrease in rainfall intensity as latitude increases northwards. For the areas of the USA exposed to hydrological hazards at the alert stage, the map in roughly overlaps with the map of the country highlighting the hurricane-prone areas [86].

The southernmost part of China, and parts of the southern USA, central and southern Africa, Latin America and India have a very high probability of exceeding the warning threshold. These regions reach soil erosion hotspots of > 20 Mg hm-2 yr-1, and are among that most intensely eroded areas in the world [87]. In particular, the observation in India follows the small convective systems that dominate throughout the Western Ghats region [88], but large events are also more intense [89].

These findings are consistent with the results of Medeiros et al. [90], who showed that the most perturbed areas in the tropical region are associated with convective storms that can have smaller radii (~10 km), as suggested by the smallest kriged-range obtained with the semivariogram function exceeding the z threshold (RED > 3.0 MJ hm-2 h-1) at alert state. These are the highest RED values found in most of South America and central and southern Africa, which are characterised by a complex property in the transfer of erosive energy to land. In these regions, sub-grid scale convection and intensification of rainfall generation processes are very hazardous, as long-lived mesoscale convective systems are well organised at these latitudes and contribute disproportionally to extreme tropical precipitation, with ~40% of days with more than 250 mm of rain over land being associated with convective systems lasting more than 24 hours [91].

Kriged-probability mapping validation at continental spatial scale

The geostatistical approach used is robust to outlier effects even when the number of experimental data is relatively small and irregularly distributed [92], as the few network stations over some regions (e.g. Russia and the African continent). Then, a main drawback that can occur is when rainfall decreases significantly after heavy rainstorms [93]. However, our approach has proven to give satisfactory results when comparing LNOK-based probability maps with the hydrological disasters recorded by the Munich Reinsurance Company (Munich Re, https://www.munichre.com, on a continental scale (data not shown). Thus, for the Asian continent, it was possible to compare the kriged-probability map, for RED > 1.5 MJ hm-2 h-1, with the damaging hydrological events (storm + erosional soil degradation) that occurred between 1981 and 2018 ().

Comparison of kriged probability maps with damaging hydrological events (Asia) and flood events (Africa) over the last four-five decades.

a) Rainfall erosivity density (RED) > 1.5 MJ hm-2 h-1 with b) with trend of disasters and c) related percentage; d) Rainfall erosivity density (RED) > 3.0 MJ hm-2 h-1 with e) trend of disasters and f) related percentage. The graphs a) and d) refer to the period 2002–2011, graphs b), c), e) and f) to 1971–2019.

The orange and red colours of the kriged map covering Saudi Arabia, Yemen, Oman, Pakistan, Nepal, Bangladesh, Burma, Vietnam, Thailand, Cambodia, the Indonesian Archipelago, southern Japan and south-eastern China (, red areas) are roughly overlapping with the hazards in multiple locations with the aggregate hydrological impact of the Munich Re dataset (not shown).

The average number of weather, climate and water hazards per decade has increased over the period 1970–2019 in Asia (). They have increased from, on average, one disaster every 15 days to one every three days [94], with a higher proportion of floods and storm events (). On the central and northern Asian continent, the low probability of high RED values (), generally associated with a low probability of damaging hydrological events (), is consistent with the rainfall minima of desert areas (e.g. in Mongolia and in northwestern China) and Siberia [95].

For the African continent, spatial dependence may be more difficult to detect because RED data are scarce. Although broad climatic patterns can be identified across the African continent, there are many local variations from place to place, with the most important differentiating climatic factor being rainfall [96]. The continent is most affected by both a continental tropical air mass to the north and by maritime tropical and equatorial air masses to the south (meeting in the Inter-Tropical Convergence Zone) [97]. Essentially, the equatorial maritime air mass is unstable and brings rain, while the tropical maritime air mass, when fully developed, is stable and generally does not bring rain unless forced over a high mountain [98].

However, with the exception of some northern coastal countries and an overestimation for some southern countries, it is still possible to roughly identify the areas most susceptible to damaging hydrological events for African lands below 15° N (, red areas), compared to the flood pattern of the Munich Re dataset (not shown).

This is also in line with increased flooding and storm-driven erosion in Africa (), and nutrient loss due to increased extreme diurnal rainfall observed in tropical eastern and south-eastern Africa in the late 20th century [99], calling into question the sustainability of food security for the ~300 million people currently living in Africa south of the equator [100]. Floods account for 60% of recorded disasters, while storms account for 17% ().

Conclusion

Detailed rainfall datasets and appropriate modelling approaches are necessary to establish the mechanisms underlying the complex set of physical processes that govern the response of hydrological cycles to a changing climate. Indeed, hydrological hazard responses to climate change remain difficult to quantify with existing modelling frameworks. Although in some areas of the world (such as the African continent) the network of weather stations is still insufficient, with the widespread availability of high-temporal resolution rainfall records for large areas and modern advances in climate modelling, new opportunities are opening up for the use of geostatistical methods for large-scale planning, hazard management and risk prevention. This paper presents a geostatistical modelling framework for the proper interpretation of spatially-explicit hydrological hazard, which assumes a set of quantitative data for the location of interest (precipitation extremes and rainfall erosivity) and probabilities associated with ranges of rainfall erosivity density above critical values. They include the best available precipitation data from a set of stations (3,625) worldwide (63 countries) and rainfall erosivity data, as provided by an updated version of the Global Rainfall Erosivity Database.

In this study, we have analysed for the first time the spatial pattern of hydrological hazard associated with rainfall erosivity in a global-scale visualisation. The results indicated that about 31% and 19% of the world’s land area have a greater than 50% probability of exceeding the warning and alert thresholds of 1.5 and 3.0 MJ hm-2 h-1, respectively, with the most affected regions being tropical Latin America, South Africa, India and the Indian Archipelago. The geostatistical modelling, designed for a spatial resolution of ~100 km, is compatible with the vast majority of countries in the world (167 out of 234 having an area >10,000 km2, https://www.worldometers.info/geography/largest-countries-in-the-world) and within the aggregation range of most environmental and biodiversity models [101]. Highlighting the potential of probabilistic geostatistical modelling, our results suggest the possibility of using geostatistical spatial modelling to determine the probability of exceeding thresholds of high erosivity density and to generate probability maps to delineate the most sensitive areas, which may lead to catastrophic regime shifts related to the occurrence of damaging hydrological events. This soft-computing modelling represents a paradigm shift on how to provide timely, accurate and actionable information on hydrological hazards [102-104]. Without giving the value at each point but returning a probability map, this approach offers the possibility to obtain information also where no data or measurements exist, as it identifies the hydrological hazard associated with the probability of exceeding an erosivity density threshold. In this way, the approach can support decision-making. We thus offer these results as a springboard to support policy-makers, local authorities and civil protection in planning medium- and long-term actions to reduce hydrological disasters [105-106]. As rainfall erosivity is projected to increase by at least 35% globally by 2070 [107], the probability of hydrological hazards will also show similar trends.

A more careful validation of the global map of hydrological disaster-prone areas is certainly needed. However, this is a promising first step, and the global probability map was well suited to hydrological disasters in regions where data coverage was substantial. Quantifying the probability of exceeding threshold values of erosivity density in a way that enables meaningful comparisons with hydrological records is an important topic of study, and our article is a step forward towards this goal. In fact, geostatistical methods can be practically implemented to create spatially explicit probabilistic maps at the country level, and can be useful in the study of erosive hazards, which, however, depend on several interacting factors, such as complex orography, large-scale air flow and teleconnection patterns. Then, purely geostatistical findings do not produce an explicit mechanistic modelling of rainfall erosivity density, which, however, is data demanding and can be accompanied by a large amount of uncertainties in the estimates. Future studies should assess the results in diverse physical geographic conditions and socio-economic situations, taking into account that population density, infrastructures, plant density and other factors also influence the occurrence of damage. In addition, probability calculations have also to take into account particularly long periods of low rainfall intensity, which are not erosive but can lead to deadly flooding and landslides.

4 Apr 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Chun Liu

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following financial disclosure:

"NO funding was received. The Open access will be paid by the Corresponding author institution"

At this time, please address the following queries:

a) Please clarify the sources of funding (financial or material support) for your study. List the grants or organizations that supported your study, including funding received from your institution.

b) State what role the funders took in the study. If the funders had no role in your study, please state: “The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

c) If any authors received a salary from any of your funders, please state which authors and which funders.

d) If you did not receive any funding for this study, please state: “The authors received no specific funding for this work.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

3. We note that Figures 3, 4 and 5 in your submission contain [map/satellite] images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

a. You may seek permission from the original copyright holder of Figures 3, 4 and 5 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

b. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Dear authors

The Manuscript is valuable and can be considered by different researchers of the world.

They have investigated Rainfall erosivity density (RED), i.e. rainfall erosivity (MJ mm hm -2 h -1 yr -1 ) per rainfall unit (mm) in worldwide using log-normal ordinary kriging with probability mapping. Then they identified damaging hydrological hazard prone areas that exceed warning and alert thresholds.

congratulations.

Reviewer #2: This is a well-written paper discussing the global outlook of the erosivity density (RED), as an indicator of hydrology-related disaster-prone areas. I provide some specific comments below for the authors to address:

1) I believe the current title is too broad to be covered by the RED-based analysis. The term "hydrological disaster" is too general in my opinion; not only the quantity, but also the quality point of view can be assessed. Even if we focus on only quantity, not only too much water (e.g., flood) but also too little water (e.g., drought) can lead to a disaster. The authors should revise the title to better reflect what they really present and discuss in the present study. By the way, it should be disaster-prone areas instead of disaster prone-areas.

2) Similar to the issue in the title, we can also find some descriptions that are not so accurate in the text. For instance, Line 31, I don't think RED is a straightforward indicator of storm surges as surges depend (more) on other factors such as sea level and tidal activity. The authors should correct this kind of inaccuracy throughout the text.

3) I don't mean to be picky, but based on the main result in Fig. 3, I am not in full agreement with the referred "global" assessment. Apparently there is a big chunk of regions left out because of the short in ground observations (and/or other reasons). In fact, since Africa is blacked out, I don't understand how you could state something like Line 125 that the most affected hydrological hazard-prone areas are Africa, etc.

4) Following my point above, I don't feel right to select Africa for validation, as shown in Fig. 4. There is nothing we can compare. North or South America and Australia should be a better region for the task like this.

5) Another point related to 3), why is the (nearly) entire European territory left out in Fig. 3, but apparently data is not an issue as shown in Fig. 5? A minor point here regarding Fig. 5 is the weird "arc-like" symbol while the caption tells "circles."

6) The authors should discuss how changes in the coefficient in the RUSLE equation affect their assessment. A similar question can be applied to the deterministic RED thresholds (i.e., 1.5 and 3). Some sensitivity analysis is required.

7) What is the spatial resolution of the RED-based assessment? Can the assessment provide fine enough details for regions at a relatively smaller spatial scale (e.g., a country)? The aspect of usefulness (e.g., pertaining to Fig. 3) should also be addressed.

Reviewer #3: Rainfall erosivity is a critical factor that affect human societies and natural landscapes. Global Rainfall Erosivity Database was used, this paper tried to assess the global hydrological disaster prone-areas. The work is important and has its value to enhance our understanding on the spatial characteristics of soil erosion and disasters.

I suggest the authors can further select some references which support these conclusions, Since some of them may go beyond the research content, et al. the explanations of regional intense RED, making the conclusions of this paper weaker. And there was a little mess about the results and discussion. I would recommend clarifying sections of the results and “discussion and look a forward” before publishing this manuscript. It needs major revision before it could be considered for publication.

7. Line 281. Had the data been downscaling?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

13 May 2022

Panos Panagos

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Our Reply: The resubmitted manuscript conforms to the journal’s style requirements.

2. Thank you for stating the following financial disclosure:

"NO funding was received. The Open access will be paid by the Corresponding author institution"

At this time, please address the following queries:

a) Please clarify the sources of funding (financial or material support) for your study. List the grants or organizations that supported your study, including funding received from your institution.

b) State what role the funders took in the study. If the funders had no role in your study, please state: “The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

c) If any authors received a salary from any of your funders, please state which authors and which funders.

d) If you did not receive any funding for this study, please state: “The authors received no specific funding for this work.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

Our Reply: Please consider the following statements by the authors:

"NO funding was received. The Open access will be paid by the Corresponding author institution".

“The authors received no specific funding for this work”.

3. We note that Figures 3, 4 and 5 in your submission contain [map/satellite] images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

Our Reply: We have removed Munich Re’s data and maps (former Fig. 1a, Fig. 4a, b) because they are not available for the time being and the company cannot provide permission for their publication. This implies that we refer to Munich Re’s data as “not shown” (line 337, line 347).

Reviewer #1: Dear authors

The Manuscript is valuable and can be considered by different researchers of the world.

They have investigated Rainfall erosivity density (RED), i.e. rainfall erosivity (MJ mm hm -2 h -1 yr -1 ) per rainfall unit (mm) in worldwide using log-normal ordinary kriging with probability mapping. Then they identified damaging hydrological hazard prone areas that exceed warning and alert thresholds.

congratulations.

Our Reply: We are grateful for recognising the value of our submission.

Reviewer #2: This is a well-written paper discussing the global outlook of the erosivity density (RED), as an indicator of hydrology-related disaster-prone areas. I provide some specific comments below for the authors to address:

Our Reply: We are grateful for recognising the importance of our submission and for providing comments to improve it.

1) I believe the current title is too broad to be covered by the RED-based analysis. The term "hydrological disaster" is too general in my opinion; not only the quantity, but also the quality point of view can be assessed. Even if we focus on only quantity, not only too much water (e.g., flood) but also too little water (e.g., drought) can lead to a disaster. The authors should revise the title to better reflect what they really present and discuss in the present study. By the way, it should be disaster-prone areas instead of disaster prone-areas.

Our Reply: New title: “Global assessment of storm disaster-prone areas”.

2) Similar to the issue in the title, we can also find some descriptions that are not so accurate in the text. For instance, Line 31, I don't think RED is a straightforward indicator of storm surges as surges depend (more) on other factors such as sea level and tidal activity. The authors should correct this kind of inaccuracy throughout the text.

Our Reply: This is right because "storm surges" is not part of our storm disaster series. Inadvertently reported in the original submission, it was deleted in the revised manuscript.

3) I don't mean to be picky, but based on the main result in Fig. 3, I am not in full agreement with the referred "global" assessment. Apparently there is a big chunk of regions left out because of the short in ground observations (and/or other reasons). In fact, since Africa is blacked out, I don't understand how you could state something like Line 125 that the most affected hydrological hazard-prone areas are Africa, etc.

Our Reply: We can state that “Africa and the southern Asian continents, southern Saudi Arabia, Australia, almost all of the USA with an offshoot to western Canada” (lines 286-288) are the areas most affected by storm hazards, as the red colours in the map of Fig. 7a indicate high values of the kriged probability. Backed out Africa (but also parts of Europe and South America) may be due to the loss of colour when converting the .docx file to .pdf during the initial submission process.

4) Following my point above, I don't feel right to select Africa for validation, as shown in Fig. 4. There is nothing we can compare. North or South America and Australia should be a better region for the task like this.

Our Reply: The answer to this remark is linked to the previous one.

5) Another point related to 3), why is the (nearly) entire European territory left out in Fig. 3, but apparently data is not an issue as shown in Fig. 5? A minor point here regarding Fig. 5 is the weird "arc-like" symbol while the caption tells "circles."

Our Reply: The answer to this remark is linked to the previous one.

6) The authors should discuss how changes in the coefficient in the RUSLE equation affect their assessment. A similar question can be applied to the deterministic RED thresholds (i.e., 1.5 and 3). Some sensitivity analysis is required.

Our Reply: Since the aim of our study is to assess storm hazards from an actual dataset on rainfall and rainfall erosivity around the world, an assessment on the impact of changes in the erosivity factor seems out of place here. In fact, rainfall erosivity is known to be the most important climatic factor in the (R)USLE approach, and how its changes impact on soil erosion and other damaging storm events such as floods and flash-floods is not discussed again. Then, a sensitivity analysis on thresholds would involve checking the probability of exceeding each threshold on a gradient of thresholds. However, the question does not arise here because ours are specific thresholds that we have identified among the n possible ones that are able to highlight the probability of occurrence of given hydrogeological events (e.g. floods, soil erosion). Testing geostatistical modelling with other threshold values becomes a redundant and superfluous task (as a geostatistical model is required for each threshold), as our thresholds are statistically relevant as they “correspond, respectively, to the median and the 3rd quartile of the distribution of the RED data (Fig. 4a)” (lines 169-170) and “also mark the range of critical RED values identified by Diodato et al. [8] with a 50-year return period” (lines 170-171). Representing probability maps with values other than the selected thresholds would not add any relevant information to this study, apart from not revealing the purpose for which the critical thresholds were set.

7) What is the spatial resolution of the RED-based assessment? Can the assessment provide fine enough details for regions at a relatively smaller spatial scale (e.g., a country)? The aspect of usefulness (e.g., pertaining to Fig. 3) should also be addressed.

Our Reply: “The geostatistical modelling, designed for a spatial resolution of ~100 km, is compatible with the vast majority of countries in the world (167 out of 234 having an area >10000 km2, https://www.worldometers.info/geography/largest-countries-in-the-world) and within the aggregation range of most environmental and biodiversity models [101]” (lines 394-397). This can be seen in the semivariograms in Fig. 5a, b, where in fact the semivariance is constant just after a distance (h) of “1.05° (~112 km) for the warning state … and at 0.70° (~70 km) for the alert state” (lines 253-254). Based on the pattern of available stations, it was possible to go down to this spatial detail although the model works on a global scale. This downscaling has a positive effect on the assessment of storm disaster-prone areas around the world and helps us to better understand the geographical aspects of flooding, soil erosion and landscape transformation (e.g. Amatulli et al., 2018, Scientific Data, https://doi.org/10.1038/sdata.2018.40).

Reviewer #3: Rainfall erosivity is a critical factor that affect human s!ocieties and natural landscapes. Global Rainfall Erosivity Database was used, this paper tried to assess the global hydrological disaster prone-areas. The work is important and has its value to enhance our understanding on the spatial characteristics of soil erosion and disasters.

Our Reply: We are grateful for recognising the importance of our submission and for providing comments to improve it.

I suggest the authors can further select some references which support these conclusions, Since some of them may go beyond the research content, et al. the explanations of regional intense RED, making the conclusions of this paper weaker.

Our Reply: The added references 102-106 support our concluding statements. The following supported concluding statement was added: “The geostatistical modelling, designed for a spatial resolution of ~100 km, is compatible with the vast majority of countries in the world (167 out of 234 having an area >10000 km2, https://www.worldometers.info/geography/largest-countries-in-the-world) and within the aggregation range of most environmental and biodiversity models [101]”.

And there was a little mess about the results and discussion. I would recommend clarifying sections of the results and “discussion and look a forward” before publishing this manuscript. It needs major revision before it could be considered for publication.

Our Reply: The manuscript was restructured to conform to the style of the journal. The revised version includes a “Results and Discussion” section and a “Conclusion” section.

7. Line 281. Had the data been downscaling?

Our Reply: This is a downscaling study. Our text informs that we used “downscaled RED data from 3,615 raingauges worldwide and log-normal ordinary kriging with probability mapping” (lines 22-23). We added the following: “The geostatistical modelling, designed for a spatial resolution of ~100 km, is compatible with the vast majority of countries in the world (167 out of 234 having an area >10000 km2, https://www.worldometers.info/geography/largest-countries-in-the-world) and within the aggregation range of most environmental and biodiversity models [101]” (lines 394-397).

Submitted filename: reply-to-reviewers.docx

Click here for additional data file.

14 Jul 2022

Global assessment of hydrological disaster prone-areas

PONE-D-22-06568R1

Dear Dr. Panagos,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Chun Liu

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: (No Response)

Reviewer #2: Thanks the authors for addressing my comments. I think this article is a great piece of work for publication.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

**********

28 Jul 2022

PONE-D-22-06568R1

Global assessment of storm disaster-prone areas

Dear Dr. Panagos:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Chun Liu

Academic Editor

PLOS ONE