DAP4: DAP4 Grids Proposal: Difference between revisions

From OPeNDAP Documentation
⧼opendap2-jumptonavigation⧽
 
(11 intermediate revisions by 3 users not shown)
Line 1: Line 1:
[[OPULS_Development| <<back to OPULS Development]]
[[Category:Development|Development]][[Category:DAP4|DAP4]]
[[OPULS_Development| << Back to OPULS Development]]


== Background: Grids Delenda Est ==
== Background ==
(with apologies to Cato the Elder)


The grid construct as originally
The grid construct as originally
Line 71: Line 71:


== Proposal: Grid as mapping ==
== Proposal: Grid as mapping ==
NB: The current Data Model page in the straw man design already does this. Unfortunately, I used XML for the examples (which muddies the idea of an abstract information model with one particular representation of that model) but I think what is presented there is the same as this proposal.[[User:Jimg|Jimg]]
Rather than making grids be scope containers, grids need to be simple relationship instances between an array and its coordinate variables. This would be done by associating the coordinate variables with an array variable.


Rather than making grids be scope containers, grids need to be
Specifically:
simple relationship instances between an array and its coordinate
# The Grid data type in DAP4 should shed the enclosing lexical scope
variables. This would be done by associating the coordinate variables
# Grid is a relation that binds one or more coordinate variables (aka maps) to one Array.
with an array variable. For example, the first case above (D1,D2,D3)
 
might be represented as:
Using OGC coverage terminology, we have this.
<font size="2">
# The maps specify the ''Domain''
<pre>
# The array specifies the ''Range''
<variable name="D1"...>
# The Grid itself is a ''Coverage''
  <map coordinate="x"/>
# The Domain and Range are sampled functions
  <map coordinates="y"/>
</variable>
...
</pre>
</font>


The case of point data would be represented as follows:
There are a number of constraints on the form of maps and their relationship to the Array.
<font size="2">
<pre>
<variable name="temp"...>
  <map coordinate="lat"/>
  <map coordinates="lon"/>
  <map coordinates="depth"/>
</variable>
</pre>
</font>
Note that the dimensions can be inferred from the specified coordinate variables.


Assume we have a (grid) array of the form
<pre>float32 A[d1...,dn]</pre>
and associated maps
<pre>float32 M1[d1,d2];
float32 M2[d3,d4];</pre>
Assume the following definitions:
* Let {A} be the set of dimensions of A, namely {d1,...dn}.
* Let {D} be the set of dimensions mentioned in any of the map variables, so in our case above {D} = {M1} union {M2}.
* Let |A| be the rank of A (n in this case).
* Let |{...}| be the number of elements in a set.
* Let {Mi} be the set of maps (= {M1,M2} in this case).


''-Dennis Heimbigner''
Using these notations, the contraints are as follows.
# |Mi| <= |A| : i.e. each map var has a dimension no more than that of the grid array.
# {Mi} has no fixed upper bound : i.e. there can be as many maps as desired.
# {D} = {A} or {D} is a subset of {A} : i.e. every named dimension mentioned in the map variables must appear in the set of dimensions of A.
# |A| = |{A}| : i.e. the dimensions of A may not contain duplicates so A[x,x] is disallowed.
# {Mi} is in fact a set, which means that any duplicates are ignored and the order is irrelevant. So {Mi} = {v1,v1,v2} is the same as {m1,m2} is the same as {m2,m1}.


== Discussion ==
== Discussion ==
[[User:Jimg|Jimg]] 16:05, 27 March 2012 (PDT) I moved the existing ''Discussion'' section to ''Old Discussion''; I think our group has converged on the following, with one topic to be resolved:
The remaining issue seems to be how 'point data' are represented. There are two candidate representations for point data:
# The Grid data type in DAP4 should shed the enclosing lexical scope
# Grid is a relational data type that binds one or more coordinate variables (aka maps) to one Array.
# In a Grid, we can define the following terms/concepts:
#* The maps specify the ''Domain''
#* The array specifies the ''Range''
#* The Grid itself is a ''Coverage''
#* The Domain and Range are sampled functions
#* We are leaning heavily on the OGC abstract coverages specification when we use these terms...
# If the Grid array has rank ''N'', each of its ''M'' maps can have at most rank ''N'' (but may have less).
# All maps have at lest one dimension ''(we didn't talk about this, but I think it's implied by all of our discussions so far)''
# That maps match the array in size, for the dimensions they share. ''(There is certainly a better way of saying this...)''
# There can be any non-zero number of maps. That is, ''M'' can be 1 <=  ''M'' < ''infinity''. ''This point is contentious; discussed below''
 
==== There can be any non-zero number of maps ====
 
This is the part that is '''Under Review'''
 
The essential issue seems to be how 'point data' are represented. There are two candidate representations for point data:
* CDM/CF-1.6: where a grid/coverage is used where the array has one dimension and there are several maps (example: temperature data - one array with one dimension - has two maps, one for lat and one for lon - each map is one dimension).
* CDM/CF-1.6: where a grid/coverage is used where the array has one dimension and there are several maps (example: temperature data - one array with one dimension - has two maps, one for lat and one for lon - each map is one dimension).
* Using Sequence: The same data can be represented using a sequence with three columns (one for temperature and one each for lat and lon).
* Using Sequence: The same data can be represented using a sequence with three columns (one for temperature and one each for lat and lon).
Line 128: Line 111:
There's no debate about the suitability of each of the above to represent 'point data'.  
There's no debate about the suitability of each of the above to represent 'point data'.  


However, there is debate about how best to ''transport'' and/or ''represent'' this information. That is, given that many systems will store point data in a relational database while many others will adopt CF-1.6 and use arrays to store the data, does adopting this mean that servers (in the aggregate) will provide two different representations for the same kind of data? My ([[User:Jimg|Jimg]]) prediction, based on the past, is that servers will have to be modified to provide the kind of responses different clients expect (rather than the case where clients are written to process each of the representations). This is a function of users/clients tending to cluster around different application areas. It may not matter for within-domain access, but it will hinder cross-domain access.
However, there is debate about how best to ''transport'' and/or ''represent'' this information. That is, given that many systems will store point data in a relational database while many others will adopt CF-1.6 and use arrays to store the data, does adopting this mean that servers (in the aggregate) will provide two different representations for the same kind of data? My ([[User:Jimg|Jimg]]) prediction, based on the past, is that servers will have to be modified to provide the kind of responses different clients expect (rather than the case where clients are written to process each of the representations). This is a function of users/clients tending to cluster around different application areas. It may not matter for within-domain access, but it will hinder cross-domain access.
 
===== Some ideas we talked about at the meeting =====
 
# We just go with it
# We limit ''grid'' to being a coverage where the number of maps ''M'' is <= the rank of the grid array and We admit a new type - ''point'' - to hold point data that does not have this limitation. Both are ''Coverages''
# We force point data to be encoded using Sequence and ensure that clients which expect CDM variable organization will see that.
 
== Old Discussion ==
The four points listed above were/are addressed on the [[DAP4: Data Model| Data Model]] but in doing so we created a new set of issues with constraints (and there were already some issues left over from DAP2). John's comment #2 immediately below seems to address one of these new problems. I think that restricting the way grids can be subset along with differentiating between subsetting Grids and parts of Grids addresses the problems. [[User:Jimg|Jimg]] 14:59, 6 March 2012 (PST)
 
=== John's comments ... ===
1) The CDM uses this object model for coordinate systems:
 
<font size="2">
  [http://www.unidata.ucar.edu/software/netcdf-java/CDM/index.html#CoordSys CDM CoordSys]
</font>
 
When translating things like GRIB into CDM, we usually also add the CF attributes, which simplifies things since now the coordsys info is encoded at the data access layer. This is very simple, in CDL:
 
<font size="2">
float Temp(z,y,x);
  :coordinates = "lat, lon, depth";
</font>
 
2) It appears that a Variable that contains map elements is a "grid", and that when you make a data request for a grid, you get back the corresponding values of the maps. Correct?
 
One problem with that is when you have 2D coordinates, as in:
 
<font size="2">
float lon(y,x);
float lat(y,x);
float Temp(z,y,x);
  :coordinates = "lat, lon, depth";
</font>
then you get back 3X more data, which you may not want.
 
--[[User:JohnCaron|JohnCaron]] 15:29, 2 March 2012 (PST)
 
From above: "...when you make a data request for a grid, you get back the corresponding values of the maps. Correct?" Ans: Maybe. Slightly longer answer: If a client requests that a Grid be subset, then it receives all of the information that a Grid requires in the response. If a client asks for a subset of one of a Grid's components, then it will receive just that component (which will no longer be a Grid). I'm defining a Grid to be a set with three things: The Grid (a specialization of an Array), one or more Maps (each also a specialization of an Array) and one or more Dimensions).  In your example, the set of ''lon'', ''lat'' and ''Temp'' is a Grid. A client can ask for the Grid to be subset and it will receive parts of all three of those pieces, which would also be a Grid. If it asks for just part of ''Temp'', it will receive just that, which is not a Grid. [[User:Jimg|Jimg]] 13:57, 6 March 2012 (PST)
 
=== Basic features of Grids in DAP4 ===
We're still discussing just how constraining "grids" works. I think that the model we choose needs to support:
* N-dimensional coordinate variables (aka maps)
* Shared dimensions
* subsetting that returns a valid grid
* subsetting that returns parts that make up the grid
 
I think optimizing transfers should be secondary to proper semantics.
 
NB: This is already present in the [[DAP4: Data Model]]
 
[[User:Jimg|Jimg]] 17:11, 2 March 2012 (PST); Updated: [[User:Jimg|Jimg]]
 
=== Counter Proposal ===
 
I think I misunderstood the original proposal... I thought it was about subsetting. I've removed my comments about that topic from this page.
 
I think the gist of this proposal is correct: "...grids need to be simple relationship instances between an array and its coordinate variables." However, the proposal also includes: "Note that the dimensions can be inferred from the specified coordinate variables." which is not correct. In the past we have talked this over and it's come to light that some grids do not have coordinate variables for all of their axis (this was news to me, but an example, I believe, is that a model run might have maps/coordinates for lat and lon and a third dimension for run number where various parameters are varied for different runs).
 
I propose that a grid be an array with explicit maps/coordinate variables for some or all of its dimensions. I also propose that an array must have explicit dimensions, which I know is obvious, but I wan to draw attention to the idea that a grid has ''both'' dimensions and maps/coordinates.
 
[[User:Jimg|Jimg]] 11:30, 7 March 2012 (PST)
 
 
=== Coordinate Systems non est Grid ===
 
So heres another aspect of this. In the above, you are (I think) describing CDM Coordinate Systems. A CDM Grid is a different animal altogether. Here is the classic example thats not a CDM grid:
 
  float data(sample);
    data:coordinates="lat lon time";
  float lat(sample);
  float lon(sample);
  float time(sample);
 
So my first thought is maybe to use a different name than Grid. Its really an "Array with CoordSys". But CDM CoordSys is more general than an Array, it happens with, say a sequence:
 
  float lat;
  float lon
  Sequence {
    data1;
    data2;
    time;
  } buoyData;
  buoyData:coordinates = "lat lon time";
 
this data has a coordinate system with a constant lat lon, and a time coordinate in each sequence.
 
So, are you intending to capture full CDM CoordSys semantics, or maybe just partial?
 
[[User:JohnCaron|JohnCaron]] 11:55, 7 March 2012 (PST)
 
I am not trying to capture the CDM coordinate system semantics. But I am trying to capture the semantics of DAP2 Grids, and then extend that to include three ideas:
# more than one dependent variable for a given set of independent variables (also can be read: more than one array for a single set of maps)
# maps that are no longer limited to vectors
# dependent variables (arrays) where not every dimension has an associated independent variable (map) (because that seems to be a reality in many cases - the array structure is used to store metadata - unless I'm mistaken on that point).
 
We have many requests for these features.
 
Is it that case that the coordinate systems in CDM have domain-specific semantics bound to them (e.g., Latitude, Mercator projection, etc.)? If so, that would be a clear difference between this idea - Grid as a mapping - and CDM coordinate systems.
 
[[User:Jimg|Jimg]] 12:39, 7 March 2012 (PST)
 
=== Coordinate Systems est divida in tres partes ===
 
CDM coordinate systems are totally general, and very simple, and include your 3 desired features. They describe both grids and points, as well as swaths, radial data, and a whole lot more. So its a shame not to use them.
 
1) Essentially one associates with each dependent variable (aka data variable) a list of independent variables (aka coordinate variables). The only restriction is that the set of dimensions (always shared) used by any coordinate variable is a subset of the dimensions used by the data variable. Then, for each data point, you get a set of coordinate values, one from each coordinate variable; eg at data(i,j,k) you get {lat(i), lon(j), alt(k)} [thats the grid case]. the point case looks like: eg at data(i) you get {lat(i), lon(i), alt(i)}.
 
Thats it, thats the general case: an association of independent and dependent variables with a restriction on the shared dimensions used.
 
There are two more parts, each a bit more specific to earth referenced data:
 
2) one can classify the coordinate variables, which is domain specific, and needed for a lot of practical things. The classification in CDM is Lat, Lon, Time, etc.
 
3) one can associate transforms that convert between coordinate systems (aka CRS). These are in practice a controlled vocabulary which we (CF/ CDM) has defined for horizontal CRS and vertical CRS. This needs 2) to work.


This part could be skipped, but you need it if you want to map into OGC protocols.  
We have moved this discussion to the proposal about [[DAP4:_VLens_(and_Sequences) | Sequences and VLens]].


None of this is needed by DAP <--> CDM, because we already do it with attributes. So one could just leave it out and/or implement it above the "transport data model". IMO adding it explicitly to the protocol would really add a lot of semantics at very little complexity. But you have to understand what these things mean or it wont be used correctly.  
[[User:dmh|Dennis Heimbigner]](5/17/2012)
The above proposal allows for map variables that are fields of a dimensioned structure. This means that map names have to be prepared to deal with names like this.
<pre>/g/S[0].f</pre>


I think this is undesirable and I propose that we do two things.
# Distinguish variables from fields; a variable is a top-level decl in a group, a field is a decl within a structure or sequence.
# Require that all map variables and all variables with maps be variables and not be allowed to be fields.


--[[User:JohnCaron|JohnCaron]] 18:11, 27 March 2012 (PDT)
[[User:Jimg|Jimg]] 13:34, 17 May 2012 (PDT)
I agree. Allowing Maps and 'Variables with Maps' to be fields is a needless mess. If there is some dataset that is really like that, the DAP layer, which is necessarily an abstraction, can hide those details. My preference in this case is for a bit more complexity on the server side to simplify the logic of clients.

Latest revision as of 19:54, 31 August 2012

<< Back to OPULS Development

Background

The grid construct as originally established in the DAP2 protocol has been a source of problems from its inception. The evolution of the notion of coordinate variables makes its use in its current form (or even closely similar forms) untenable.

Problems Addressed

Grid as scoping/lexical container

This means that properly sharing coordinate variables is not possible without duplication, which is highly undesirable.

Consider the following situation.

    Arrays: D1(x,y), D2(y,z), D3(x,z).
    coord vars: x(x), y(y), z(z)

No grid, as currently defined can represent this because the three coordinate variables x(x), y(y), and z(z), cannot be properly distributed across needed three grids without duplication. The only way this can work is if all the Arrays and all the coordinate variables reside in a single grid; not, I maintain, a useful solution. Further, the Grid must change if new arrays are defined that use any of the coordinate variables, D4(x,w), for example.

Grid projections

When a projection is applied to a grid, the result cannot be a grid. This has been an ongoing source of problems in DAP2 where projecting the array component of a grid results in a structure. From the point of view of semantics, this is a really bad idea.

Multi-dimensional coordinate variables

When representing point data, it is desirable to have coordinate variables distinguished using more than a single dimension. Consider the following:

    array: temp(x,y,z)
    coordinate vars: lat(x,y,z), lon(x,y,z), and depth(x,y,z).

Here we are trying to represent point data where each point is defined by three dimensions: lat, lon, and depth. Grids are not capable of properly representing this case. I should note that neither is, for example, netcdf-3 or netcdf-4. CDM can do it, by only by encoding the proper relationships as attributes with complex internal structure.

Coordinate Variable Duplication

In examining a large number of DAP2 DDS's, I note that coordinate variables inside grids are almost always duplicated outside the grid. My hypothesis has been that this a result of the fact of problem (1) above. In any case, this proposal below would obviate the need for duplication.

Proposal: Grid as mapping

Rather than making grids be scope containers, grids need to be simple relationship instances between an array and its coordinate variables. This would be done by associating the coordinate variables with an array variable.

Specifically:

  1. The Grid data type in DAP4 should shed the enclosing lexical scope
  2. Grid is a relation that binds one or more coordinate variables (aka maps) to one Array.

Using OGC coverage terminology, we have this.

  1. The maps specify the Domain
  2. The array specifies the Range
  3. The Grid itself is a Coverage
  4. The Domain and Range are sampled functions

There are a number of constraints on the form of maps and their relationship to the Array.

Assume we have a (grid) array of the form

float32 A[d1...,dn]

and associated maps

float32 M1[d1,d2];
float32 M2[d3,d4];

Assume the following definitions:

  • Let {A} be the set of dimensions of A, namely {d1,...dn}.
  • Let {D} be the set of dimensions mentioned in any of the map variables, so in our case above {D} = {M1} union {M2}.
  • Let |A| be the rank of A (n in this case).
  • Let |{...}| be the number of elements in a set.
  • Let {Mi} be the set of maps (= {M1,M2} in this case).

Using these notations, the contraints are as follows.

  1. |Mi| <= |A| : i.e. each map var has a dimension no more than that of the grid array.
  2. {Mi} has no fixed upper bound : i.e. there can be as many maps as desired.
  3. {D} = {A} or {D} is a subset of {A} : i.e. every named dimension mentioned in the map variables must appear in the set of dimensions of A.
  4. |A| = |{A}| : i.e. the dimensions of A may not contain duplicates so A[x,x] is disallowed.
  5. {Mi} is in fact a set, which means that any duplicates are ignored and the order is irrelevant. So {Mi} = {v1,v1,v2} is the same as {m1,m2} is the same as {m2,m1}.

Discussion

The remaining issue seems to be how 'point data' are represented. There are two candidate representations for point data:

  • CDM/CF-1.6: where a grid/coverage is used where the array has one dimension and there are several maps (example: temperature data - one array with one dimension - has two maps, one for lat and one for lon - each map is one dimension).
  • Using Sequence: The same data can be represented using a sequence with three columns (one for temperature and one each for lat and lon).

There's no debate about the suitability of each of the above to represent 'point data'.

However, there is debate about how best to transport and/or represent this information. That is, given that many systems will store point data in a relational database while many others will adopt CF-1.6 and use arrays to store the data, does adopting this mean that servers (in the aggregate) will provide two different representations for the same kind of data? My (Jimg) prediction, based on the past, is that servers will have to be modified to provide the kind of responses different clients expect (rather than the case where clients are written to process each of the representations). This is a function of users/clients tending to cluster around different application areas. It may not matter for within-domain access, but it will hinder cross-domain access.

We have moved this discussion to the proposal about Sequences and VLens.

Dennis Heimbigner(5/17/2012) The above proposal allows for map variables that are fields of a dimensioned structure. This means that map names have to be prepared to deal with names like this.

/g/S[0].f

I think this is undesirable and I propose that we do two things.

  1. Distinguish variables from fields; a variable is a top-level decl in a group, a field is a decl within a structure or sequence.
  2. Require that all map variables and all variables with maps be variables and not be allowed to be fields.

Jimg 13:34, 17 May 2012 (PDT) I agree. Allowing Maps and 'Variables with Maps' to be fields is a needless mess. If there is some dataset that is really like that, the DAP layer, which is necessarily an abstraction, can hide those details. My preference in this case is for a bit more complexity on the server side to simplify the logic of clients.