Documentation: Methods for constructing the inventories presented on this site, and their spatial distribution
More information regarding the application of methods for the construction of a specific global inventory can be found in the publications and reports relating to that inventory. Further details of the geospatial distribution methods can be found in Wilson et al., 2006.
The ancillary datasets available on this site include basic grids including attributes for converting from grid values to per unit area values, population datasets, distributed masks, etc.
Compiling Global Mercury Emission Inventories: Basic Methodology
The basic methodology employed to construct the global inventories of mercury emissions to air available on this site is as follows:
Emission estimates are calculated for specified source sectors (e.g. specific industrial sectors or activities giving rise to mercury emissions). These estimates are produced for individual countries using activity data (i.e., amounts of fuels or raw materials used or products manuafactured) for the country concerned in combination with emission factors (EFs: factors that represent the amount of mercury emitted per unit of the ‘activity’).
Early inventories (1990-2005) employed generic EFs that were intended to represent the abated (i.e. 'end-of-pipe') emissions. For a given sector, the same EFs were applied to all countries.
In recent inventories (2010), the concept of emission factors has been developed to apply ‘unabated’ emission factors (which determine the amount of mercury ‘liberated’ by the processes concerned), and then factor in 'Technology Profiles' reflecting the technology employed to abate mercury emissions and releases, in order to derive abated emission estimates. Databases have been developed that allow a more flexible application of these EFs and technology profiles so that differences between situations in differnt countries can be better represented.
Early global emission inventories (1990-1995) covered mainly the unintentional release sectors (i.e. sectors where mercury is released as a 'by-product' of the primary activity; examples are emissions and releases from coal burning, or primary production of cement and ferrous and non-ferrous metals, etc. They also included emisisons associated with intentional-use of mercury in the chlor-alklai industry.
Later global inventories (2000-) introduced additional intentional use sectors such as emissions from use of mercury in artisanal and small-scale gold mining (ASGM) and dentistry (from human cremation), and emissions and releases resulting from disposal of products containing mercury (e.g. lamps, batteries, etc.) in waste streams.
Inventory results comprise sector specific emission estimates for each country.
Most inventories are associated with a ‘nominal year’ – that is the target year for which the inventory applies. It should be recognized however that available data sources may not always include data for that year, so actual data used for different countries and sectors tends to bracket a period of 2-3 years around the target year.
The above calculation methods – especially at the global scale - involve a number of assumptions, and hence it is important to consider the resulting emission estimates in relation to their uncertainties.
Emission Source Sectors
Emission inventories produced at different times have included different emission source sectors, and different degrees of specificity concerning the activity data quantified under a given source sector. The table below provides an overview of the source sectors included in global mercury emission inventories prepared at differnt times. (x = sector included in inventory; grey shading = first inventory including that sector). (table entries for 1990-2000 not yet available)
|Stationary combustion (SC)||coal combustion in power plants||SC-PP-coal||x||x|
|coal combustion in industry||SC-IND-coal||x|
|oil combustion in power plants||SC-PP-oil||x|
|oil combustion in industry||SC-IND-oil||x|
|coal combustion in domestic residential and other uses||SC-DR-coal||x||x|
|oil combustion in domestic residential and other uses||SC-DR-oil||x|
|natural gas combustion in power plants||SC-PP-gas||x|
|natural gas combustion in industry||SC-IND-gas||x|
|natural gas combustion in domestic residential and other uses||SC-DR-gas||x|
|Pig iron and (primary) steel production (PISP)||PISP||x|
|Non-ferrous metal production (NFMP)||copper production (primary)||NFMP-Cu||x|
|lead production (primary)||NFMP-Pb||x|
|zinc production (primary)||NFMP-Zn||x||x|
|aluminium production (primary)||NFMP-Al||x|
|large-scale gold production||NFMP-Au||x||x|
|mercury production (from mining)||NFMP-Hg||x||x|
|Artisanl and small-scale gold mining||ASGM||x||x|
|Waste disposal||Incineration of (product) waste in large incinerators||WI||x||x|
|Other disposal of (product) waste||WASOTH||x||x|
|Secondary steel (recycled scrap)||SSC||x|
Geospatial distribution: Basic Methodology
For modelling purposes, it is desirable not only to know the amounts of mercury emitted/released, but also where these releases take place, and also the form in which the mercury is released; ideally this information would also include better temporal resolution than the annual estimates currently available.
Sector-based global emission estimates are currently ‘speciated’ according to generic assumptions regarding emissions characteristics of the sector concerned – with the total mercury emissions (HgT) split between elemental mercury (Hg0), divalent mercury species (Hg2) and particulate associated mercury (HgP). These assumptions are reflected in emission 'speciation schemes’. In the case of emissions to air, these 'speciation schemes' also include assumptions regarding the proportions of the mercury species emitted within different (geometrical) emission height classes.
The term ‘geospatial distribution’ relates to locating emissions geographically.
Some part of the national emissions estimates can generally be associated with point sources. Information regarding point source emissions is gradually improving, for example through national pollution release and transfer registers. However, at present many identified facilities that have associated mercury emissions or releases (or have the potential for mercury emissions and releases) lack associated emission estimates. Where they are available, such plant specific estimates are typically based on a limited number of measurements that may or may not be representative of actual emissions over longer periods. For example, changes in fuels and raw materials used or other plant operating paramaters can have a significant impact on emissions. In addition, pollution release registers typically employ an emissions threshold below which reporting is not mandatory. Ultimately, assigning national (sector-based) emissions to individual point sources makes use of available data but also involves making further assumptions.
Other emission components are associated with diffuse sources, and (when it comes to geospatially distributing emissions estimates) these components also include that part of the (national) emissions that are associated with point sources for which there is no information regarding the locations of the point sources involved.
Generally, diffuse emissions (and emissions associated with point sources for which their location is unknown) are geospatially distributed by employing ‘surrogate’ datasets for which a geospatial distribution is available. A typical example is the use of population distribution – where for some sectors, in the absence of better information, it might be assumed that (within a given country) the pattern of mercury emissions generally follows that of population density. This may be the case for mercury emissions associated with e.g. wastes from mercury-containing products, however, for other sectors (e.g. mining-related sectors), population distribution would likely be a poor surrogate to use. Therefore, a number of other ‘distribution masks’ have been constructed based on different surrogate parameters. These ‘distribution masks’ are essentially a set of weighting factors in a geospatial grid that reflect the proportion of the national total for that (surrogate) parameter (e.g. national population total) that is present in a given grid cell. These factors then become multipliers for national emissions or some part of the national emissions for a given sector. Different distribution masks are applied to differnt sectors.
Geospatially distributed emissions inventories are the result of applying the above geospatial distribution model (through computer applications that capture the various data sources and assumptions) to the national sector based emissions estimates – with output comprising aggregated gridded estimates of the emissions of the various mercury ‘species’ in different height classes.
Currently the available geospatially distributed global emissions inventories are distributed to a 0.5 by 0.5 degree latitude/longitude grid.
Two grids have been employed in distributing global mercy emissions datasets - both cover the entire globe. Current inventories are distributed within the Z05 grid comprises 259200 (720 x 360) grid cells with a cell size of 0.5 x 0.5 degrees. Older inventories were also prepared according to the 1 x 1 degree GEIA grid comprising 64800 (360 x 180) grid cells.
The grids are defined as follows:
Z05 (0.5 x 0.5 degree) grid definition:
259200 (720 x 360) cells
Z05 Cellcode = (j*1000) + i
j = row number starting at 1 for 90S to 89.5S latitude, to 360 for 89.5N to 90N latitude
i = column number starting at 1 for 180W to 179.5W longitude, to 720 for 179.5E to 180E longitude.
(coordinates represent the center of the gridcell)
The latitude and longitude of the center of a gridcell is given by:
latitude = ((j-181)/2) + 0.25
longitude = ((i-361)/2) + 0.25
To calculate the Z05 cell code from latitude/longitude coordinates in Excel use the following formula:
C1 = (((INT(A1*2))+181)*1000)+((INT(B1*2))+361)
where A1 is the worksheet cell containing latitude (in decimal degrees, negative if south), B1 is the worksheet cell containing longitude (in decimal degrees, negative if west) and C1 is the worksheet cell containing the resulting Z05CellCode.
To calculate the Z05 cell code in Python:
grid cell = int((((math.floor(lat * 2)) + 181) *1000) + ((math.floor(lon * 2)) + 361))
Example: Reykjavik 64° 9’ 0”N -> 64.15° ; 21° 56’ 0”W -> -21.93°
>>> print str ( int((((math.floor(64.15 * 2)) + 181) *1000) + ((math.floor(21.93 * 2)) + 361)) )
GEIA (1 x 1 degree ) grid definition
64800 (360 x 180) gridcells
GEIA Cellcode = (j*1000) + i
j = row number starting at 1 for 90S to 89S latitude, to 180 for 89N to 90N latitude
i = column number starting at 1 for 180W to 179W longitude, to 360 for 179E to 180E longitude
(coordinates represent the center of the gridcell)
The latitude and longitude of the center of the grid is given by:
latitude = (j-91) + 0.5
longitude = (i-181) + 0.5
As mentioned, population distribution is widely used as a surrogate parameter for the spatial distribution of emissions from anthropogenic sources (for some sectors) when the exact location of the emissions is unknown. This is based on the logic that emissions derived from human activities are co-located with areas of highest population density.
The dataset required for this consists of values reflecting the proportion of a countries total population in a given gridcell - this value being used as a multiplier for the total emissions from 'distributed sources' (for a relevant source sector) from that country.
In the past, much of the work on spatial distribution of global (and regional) emissions inventories employed a (1990) population distribution dataset available from GEIA (Li, 1996). This dataset has a resolution of 1 x 1 degree and is based on 1990 global population distribution.
In order to (a) improve the spatial resolution of the distribution on 'distributed source' emissions, and (b) use contemporary population distribution with respect to the emissions datasets, new population datasets are constructed as follows:
Data on global population distribution in 1995 and 2000 (Gridded Population of the World (GPW v3)) were obtained from the CIESIN (2004). These datasets have a spatial resolution of 2.5 arc-minutes.
To assess the absolute number of people living each 0.5 degree cell, an intersect was made by overlaying the 0.5 degree grid and a GPW v3 country dataset for the year concerned. Cells that contained land area from more than one country were split into respective sub-cells (polygons).
The image below shows the 0.5 x 0.5 degree grid superimposed on the CIESIN GPW population distribution dataset. Cell boundaries and national borders define the gridded population 'polygons'. In the area where France, Belgium and Luxembourg border each other, one grid cell has been split into three polygons.
A zonal statistics procedure was used to calculate the total population within each grid cell or (in cases or two or more countries occupying part of a given cell) sub-cell (polygon)). During this procedure, a small loss in the population totals occurs due to the fact that the country data and GPW raster cannot be completely spatially matched. To overcome this problem, working values for country total populations were calculated by summing the population values of all polygons assigned to that country. Based on these working totals and the absolute population numbers within each polygon, the proportion of a contries population in that cell was calculated.
The 1995 and 2000 population datasets used in this work are available in both 1 x 1 degree and 0.5 x 0.5 degree grids.
Following use of the proportion of population in a cell as a multiplier for national emission values, emissions within each cell were aggregated to combine the emissions in cells shared by more than one country.
list currently being updated