T  h  e    P  a  n  g  e  a    M  o  d  e  l

Glossary of terms and concepts

• ArcGIS geoprocessor: ESRI/ArcGIS is distributed with a Python library (class) called geoprocessor in ArcGIS 9.x and arcpy in ArcGIS 10.x. It essentially provides all the tools from the toolbox (nested red folders in e.g. ArcMap), available as methods. This provided Pangea v1.x with all the tools that a GIS analyst would use to e.g. build grids and summarize data, allowing Pangea to automatize these processes. Pangea v2.x is progressively implementing all relevant tools in MATLAB.
• Cell (grid -) : In Pangea, a grid cell is a geometric entity that is a component of a grid. It has a geographical "reality", often irregular boundaries, complex connections to other grid cells, and delinates a region of the space with hetergeneous content (in terms of environmental media, in the general case).
• Compartment : In Pangea, a compartment is an abstract component of the virtual system. It has a simple geometry, simple connections to other compartments, and it represents one homogeneous component of one grid cell (e.g. medium fresh water of some grid cell south of Chicago).
• Computation engine : Set of tools for solving large systems of equations (ODEs, linear systems, etc). In Pangea v1.x, the computation engine was implemented in Python with an external module implemented in MATLAB (for the computational part involving large sparse matrices). In Pangea v2.x it is fully implemented in MATLAB.
• Environmental Model (EM) : Environmental Models (EMs, for lack of a better terminology) are models and data sets that parameterize/characterize environmental media (e.g. fresh water hydrology) or types of regions (e.g. land cover). They are substance-independent. The atmospheric model, for example, provides tools for creating 3D atmospheric grids, projecting wind speeds, computing horizontal and vertical flows, and processing rain data sets. EMs are explicitly spatial; they are based on the geometry and geography of the features that they cover. On the contrary, EPMs (see below) describe substance-related processes, e.g. diffusion or degradation. They are spatial in an abstract way: the atmospheric advection EPM for example receives flows, compartments dimensions, etc. from the atmospheric EM, but it uses them without knowing e.g. their "true" location or geometry (which do no exist as compartments are abstract/virtual). EMs characterize therefore part of the geometric system, whereas EPMs work in the virtual system.

Pangea provides the following set of EMs by default:
• PAM: Pangea Atmospheric Model, parameterized using a reference year (2005) of GEOS-Chem (GEOS-4, 2°$$\times$$2.5° global 55-layers grid) wind fields, or higher resolution years (2013 on, supporting 72-layers 0.25°$$\times$$0.3125° continental grids nested in a global 2°$$\times$$2.5° grid) defined by GEOS-FP.
• PHM: Pangea Hydrological Model. Two versions are available, the first/historical is based on the WWDRII 0.5°$$\times$$0.5° fresh water model, and the second (in development) is based on the HydroBASINS model/database.
• PTM: Pangea Terrestrial Model, parameterized using the GlobCover data set. It uses the hydrological grid as defined by PHM. At this stage, it defines the sediments grid as well.
• POM: Pangea Oceans Model. In development. Not the focus currently.
• Environmental Process Model (EPM) : Environmental Processes Models (EPMs, for lack of a better terminology) are models that characterize substance-specific environmental processes such as advection and degradation. The difference between EMs and EPMs is discribed in the EMs section. Pangea comes with several sets of EPMs: a first/historical set based on IMPACT2002 and USEtox 1.x, and second based on USEtox 2.0 (concensus model endorsed by UNEP/SETAC), and a third (default) that updates the USEtox set with EPMs related tos sediments from SimpleBox.
• GIS engine : Set of tools and resources relavant for performing all GIS tasks necessary for the functioning of Pangea, mainly the creation of global 3D multi-scale grids, and to projection of geo-referenced data sets. The GIS engine is a cascade of functions that use the MATLAB Mapping toolbox, ArcGIS, and Quantum GIS depending their availability. Currently, MATLAB Mapping and ArcGIS and mandatory, but the objective is to build an engine that can work with any library and taken advantage of e.g. ArcGIS when available.
• Local to global : The model works with grids whose cells surfaces areas range typically from a few square kilometersto the size of a continent. This allows to design grids with a high/local resolution at locations of interest while keeping the ability to obtain results globally, which allows e.g. to compare local versus global results. This is made possible by the GIS engine which allows to build project-specific multi-scale grids and to project spatial data onto them at run time.
• Medium : Environmental medium, e.g. air, fresh water, or agricultural soil.
• Multi-pathways exposure : Six or seven pathways are currently considered important for studying an environmentally mediated multi-pathway exposure: inhalation and ingestion, with the latter through drinking water, fish, beef, eggs, above-ground vegetation (e.g. cereals, fruits, and vegetables) and below-ground vegetation (e.g. carrots and potatoes).
• Multi-scale (spatial) : Multi-scale grids are grids composed of cells whose sizes span multiple orders of magnitude, e.g. 10km×10km for the smallest (highest resolution) and 1000km×1000km for the largest (lowest resolution). Multi-scale approaches are relevant in contexts where the spatial extent of the modeling domain is large (here global), yet we need a high-enough spatial resolution for capturing e.g. the specifics of the direct vicinity of emission sources (of pollutants) and receptors (e.g. people). The basis of the multi-scale approach implemented in Pangea is what we call a refinement potential (RP): a scalar field which defines at each location the "need for having a high resolution". The RP is integrated in an iterative procedure that refines a low resolution background grid until the integral over each cell is below a given threshold (or until a given refinement depth is reached). The outcome of this process is a multi-scale grid called the results grid, the grid onto which all results are projected ultimately. Other grids are built based on the results grid and/or the refinement potential.
• Multimedia fate (elimination) and transport : Transport and elimination of substances in the environment that involve multiple media, e.g. accounting for exchanges between air, various soil types and water, and for the degradation within each one of these media.
• Spatial : Pangea is spatially explicit, which means that it uses real geo-referenced spatial data. It creates grids that cover real geographic spaces, uses watersheds delineations that correspond to real streams and watersheds, is parameterized using geo-referenced data (e.g. longitude and latitude of emission sources, and rasters of population densities), etc. This is opposed to (non-spatial) generic models (like SimpleBox) or to abstract spatial models (like USEtox™) that use abstract[1] nested regions whose features emulate local, continental, and global extents. [1] USEtox uses a continental scale which is parameterized based on the IMPACTWorld model. This scale corresponds therefore to specific geographic regions.