clarification of area type definitions needed

[taylor13]

26 January 2025

Many of the CF area type options have no "description" or need clarification, and I'm pretty sure that in some cases, interpretation of the "types" is not obvious. In requesting model output for CMIP, we want all models to interpret the area types in the same way, so I have come up with the the following interpretations, and wonder if we might formally adopt these in the conventions:

sea: the description currently reads:

The area occupied by type "sea" is equal to the sum of the areas of types "ice_free_sea" and "sea_ice".

Proposed new description:

This area type can either refer to the area occupied by the upper surface of the sea or refer to the area occupied by sea at some depth below its surface, as in the following:

1. When a variable represents some property of the sea reported as a function of a vertical coordinate, then the area designated as "sea" may depend on depth, with smaller area included at greater depths if parts of the column of water bottom out.

2. Similarly, if a sea property is integrated in the vertical and the upper limit of the integral is below the seafloor, the location is not considered "sea" (even though sea water might be found above the integrated portion of the column).

3. For all other variables, area type "sea" will identify the locations where the surface of the earth is "sea" (or where "sea" is overlain by sea ice, icebergs, ice shelves and the like). For example,

Not all water bodies are identified as "sea". "Sea" is defined as a region occupied by a body of water with its upper surface at sea level (or near sea level if there is ice floating on it) that directly exchanges mass, heat, and momentum (horizontally) as a physically interactive part of an ocean. This excludes, for example, rivers with mass transport dominated by gravitational force and typically much fresher than the ocean. Some familiar seas, identified as such on world maps, do not meet the above definition for the area type "sea" (e.g., the Caspian Sea, Dead Sea, and Salton Sea).

Every location on Earth is either designated as being "sea" or "land". Any location on Earth not considered to be "sea", as defined here, is considered to be "land". Together these two area types cover the entire globe. The land_area_fraction and the sea_area_fraction sum to 1 (but the portion assigned to sea can decrease below the Earth's surface).

An area type designated as being "sea" may also more specifically be characterized as ice_free_sea or ``floating_ice_shelf```, or it may belong to one of the various sea ice area types. This is consistent with the general rule that an area underlain by "sea" should also be assigned to the "sea" area type.

There are some area types that may overlap with sea but are not exclusively associated with "sea" . For example, the area type "snow" references areas of surface snow cover, which may be found both on land and on sea ice and floating ice shelves (which may sometimes be considered "sea").

In climate models, the locations considered to be of area type "sea" can depend on model resolution and the choices made in treating marginal seas. For example, a large bay with a rather narrow opening to the ocean might in a model be closed off dynamically from the ocean and would not be considered "sea" for the purposes of that model. The area type "sea" in global climate models is reserved for water bodies that are connected forming the world ocean and for which mass, momentum, and energy are exchanged using a common set of governing equations. Isolated marginal seas, lakes, and other inland bodies of water are classified as belonging to the comprehensive "land" area type, but may also be more specifically referenced as being an area type such as "lake" or "fresh_free_water".

Categorizing "floating_ice_shelves" as "sea" or "land" depends on the context. The area type "land" includes regions of floating ice shelf when a variable describes: 1) a property of a bulk floating ice shelf itself (e.g., its density), 2) a characteristic of its upper surface (e.g., the upwelling longwave radiation), a substance lying on its surface (e.g., snow), or processes in the atmosphere above (e.g. cloud area fraction "where land" would include the clouds above floating ice shelves). On the other hand, the area type "sea" would apply to variables describing the sea below an ice shelf and the exchanges that take place at the interface between the ice shelf and the sea below. In climate models, if the water below an ice shelf is excluded from directly interacting with the world oceans, that water is considered to belong to the land area type.

[Note that the current CF description of "sea" leaves out "floating_ice_shelves."]

land: the description currently reads:

No help available.

Proposed new description:

Any location not designated as belonging to area type "sea" belongs to area type "land". Thus, any location on Earth is considered to be either "sea" or "land".

An area designated as "land" may more specifically be characterized by some property of its surface. There are specific area types for referencing land covered with various types of vegetation (e.g., "tree", "crops", "shrubs") or other surface or subsurface properties (e.g., "urban", "fire", "unfrozen_soil", "bare_ground", "wetland", "lake").

Note that inland bodies of water and some marginal seas are classified as being area type "land", as described more fully in the definition of the "sea" area type.

There are some area types that may overlap with land but are not exclusively associated with "land" . For example, the area type "snow" references areas of surface snow cover, which may be found not only on "land", but also on sea ice and floating ice shelves.

sea_ice: the description currently reads:

The area occupied by type "sea_ice" within a grid cell is the sum of the areas of types "sea_ice_melt_pond" and "melt_pond_free_sea_ice". Melt ponds occur on top of the existing sea ice. The area occupied by type "sea" is equal to the sum of the areas of types "ice_free_sea" and "sea_ice".

Proposed new description (eliminate last sentence because type "sea" is described elsewhere and "sea" may include "floating_ice_shelf":

The area occupied by type "sea_ice" within a grid cell is the sum of the areas of types "sea_ice_melt_pond" and "melt_pond_free_sea_ice". Melt ponds occur on top of the existing sea ice. Any location identified as type "sea_ice" is also defined to be area type "sea". Some other area types can include sea ice or a portion of sea ice, along with locations that are not sea ice. The "snow" area type, for example, includes areas of sea ice covered by snow, but this area type also identifies land areas that are snow covered.

floating_ice_shelf: the description currently reads:

An area type of "floating ice shelf" indicates where ice shelves are present. Ice shelves are the component of ice sheets that flow over the ocean.

Proposed new description:

An area type of "floating ice shelf" indicates where ice shelves are present. Ice shelves are the component of ice sheets that flow over the ocean. As discussed in the description of "sea", "floating_ice_shelves" are included in area type "sea" or area type "land", depending on the context.

[github-actions]

Thank you for your proposal. These terms will be added to the cfeditor (http://cfeditor.ceda.ac.uk/proposals/1) shortly. Your proposal will then be reviewed and commented on by the community and Standard Names moderator.

[JonathanGregory]

Dear Karl

Thanks for drafting these new definitions. It's surprisingly difficult when you consider it in detail! I agree that the distinction between sea and land can depend on the vertical level. I'm not convinced about making the distinction depend on the model. I think we ought to tie it to reality, if we can.

The Mediterranean Sea might be the biggest example. Because the Straits of Gibraltar are narrow, not all models connect the Mediterranean dynamically to the world ocean. However, I don't think anyone would describe the Mediterranean as a "lake", or the area occupied by the Mediterranean as "land", even in a model, would they? In contrast, although the Caspian Sea is not usually called a lake, it is more natural to describe it as such. Wikipedia says it can be described as the world's largest lake, for instance.

Therefore, instead of the paragraph "In climate models, ...", I feel it would be better to define the classification of sea as referring to reality. I agree with

"Sea" is defined as a region occupied by a body of water with its upper surface at sea level (or near sea level if there is ice floating on it) that directly exchanges mass, heat, and momentum (horizontally) as a physically interactive part of an ocean.

Possibly it might sound circular to define sea in terms of sea level, but I think it's reasonable. More precisely, you might say

"Sea" is the area occupied by a body of water whose time-mean upper surface (in the real world) deviates from the geoid only because its time-mean circulation is not zero.

Is that too obscure? The global ocean is a dynamically connected body of water, whose volume is equal to the volume contained between the geoid and the sea floor. If it was at rest, which is impossible because it contains density contrasts and experiences windstress, mean sea level everywhere in the sea would coincide with the geoid, which is a geopotential surface i.e. "flat" in a geophysical sense. This includes the Mediterranean, but it does not include the Caspian Sea, or any inland water bodies (lakes, rivers, etc.), whose time-mean surface level is unrelated to the geoid. It depends on the altitude of their bed and the amount of water they contain, as well as their dynamics.

I would say that "marginal seas" are part of the global sea, if they're dynamically connected to it in reality. Which areas do you have in mind?

I agree that ice shelves are land if you consider their bulk properties (because they're land ice, like glaciers) or when you are looking from above, at the nature of their surface and surface fluxes, but sea if you consider their interaction with the ocean. I'm unsure about "In climate models, if the water below an ice shelf is excluded from directly interacting with the world oceans, that water is considered to belong to the land area type." Considering the water below an ice shelf is a view from below, so the sub-ice-shelf area should be sea, I feel. Couldn't we say that, regarding interaction of the ice with the ocean below, the whole ice shelf area is sea? We can't apply the sea level criterion in any case for water which doesn't have a free surface.

Best wishes

Jonathan