The first step in creating a multihazard surface is to isolate those areas that are the most at-risk for mortality (deciles 8-10) for each of the hazards considered in the study (cyclones, droughts, earthquakes, floods, landslides, and volcanoes). Three categorical hazard surfaces representing the highest mortality risks are created from combinations of the single hazard surfaces. The hazard categories are as follows: Drought - drought; Seismic - earthquakes and volcanoes; Hydro - cyclones, floods, and landslides.
Each of the hazard category surfaces are binary. To produce the multihazard suface, each of the categorical surfaces are combined into a single surface. Each grid cell of the multihazard surface is attributed with information regarding its value in each of the hazard categories (items Drought3, Seismic3, and Hydro3). This sequence of values forms a unique class of multihazardness.
The population per grid cell is based on GPW v3.0 (beta). The area per grid cell, in square kilometers, is calculated by subtracting from the area of the grid cell those portions that are permenantly inundated. Identification of permanently inundated areas utilizes VMAP(0) data. VMAP(0) data is also used to identify major roads and railroads.
Building upon a methodology proposed by Sachs et al. (2003), a gross domestic product (GDP) value (US$, 2000, purchase power parity adjusted (PPP)) is estimated on a per grid cell basis. First, the contribution of subnational units to national GDP utilized data of varied origin. The ratio of the subnational production to the national GDP is the contribution rate. To standardize across countries, these contribution rates are applied to published World Bank GDP estimates to determine GDP values of the subnational units.
The standardized, subnational GDP estimate is divided by the total population within the subnational unit to produce a spatially variable per capita GDP estimate. The GDP per grid cell is calculated by multiplying the subnational, per capita GDP estimate by the grid cell population density.
Agricultural GDP (US$, 2000, PPP) is also based on a process of spatial reallocation. At the level of the subnational unit, estimates are made of the agricultural GDP and of the total agricultural area within the subnational unit. Agricultural area is identified using a reinterpretation of USGS Global Land Cover Characterization data. Dividing the estimate of agricultural GDP at the subnational level by the total amount of agricultural land within the subnational unit produces a value of agricultural GDP per unit agricultural area. The agricultural value of a grid cell is determined by multiplying the agricultural area within the grid cell by the agricultural GDP per unit agricultural area value.