The records of the Emergency Events Database (EM-DAT), collected over a 20 year period from 1981 to 2000, provide regional, hazard-specific mortality and economic loss rates. A crude estimation of the global drought hazard mortality is developed using the EM-DAT regional mortality rates, population distributions from Gridded Population of the World (GPW) Version 3.0 (beta), and frequency/distribution data from Global Drought Hazard Frequency and Distribution. To better reflect the confidence associated with the result, mortality figures are classified into deciles, 10 classes of an approximately equal number of grid cells of increasing mortality (item Value).
Building upon a methodology developed by Sachs et al. (2003), a gross domestic product (GDP) value (US$, 2000, purchase power parity adjusted (PPP)) is estimated for each grid cell. The process begins by determing the contribution of each subnational unit to national GDP using data of varied origin. The ratio of the subnational production to the national GDP is the contribution ratio. To ensure uniformity between countries, these contribution ratios are utilized with published World Bank estimates of GDP.
Once a standardized version of subnational GDP has been calculated, this value is further divided by the total population within the subnational unit. This subnational, per-person GDP value is multiplied by the grid cell population density to determine a GDP value for the grid cell. The GDP values presented in this dataset (item Gdpvalue) are not projections of impacted GDP, but rather the estimates of GDP that serve as a baseline for estimating hazard impacts. Furthermore, Gdpvalue is indicative of the GDP associated with each of the hazard risk deciles and not the individual grid cell.
Estimating the agricultural GDP (item Agvalue) follows a process similar to GDP. The amount of agricultural GDP is derived at the subnational unit using available data of various origin. Agricultural lands themselves are identified using a reinterpretation of USGS Global Land Cover Characterization data. At the level of the subnational-unit, an agricultural value per unit area is determined based upon the subnational agricultural productivity and the amount of agricultural lands within the subnational unit. This value is then multiplied by the amount of agricultural area within a grid cell to determine the agricultural GDP per grid cell. As with Gdpvalue, Agvalue is not an estimate of the agricultural GDP impacted by the hazard, but rather the estimated agricultural productivity that serves as a baseline. Agvalue is representative of the hazard decile class and not the indivual grid cell.
Areas of "No Data" are not necessarily without risk from hazards, but may rather be an artifact of a mask that excluded from analysis those areas with a population density less than 5 persons per square kilometer and without significant agriculture.