The [[ method works for all oce objects, i.e. objects inheriting from oce. The purpose is to insulate users from the internal details of oce objects, by looking for items within the various storage slots of the object. Items that are not actually stored in the object can also be extracted, including derived data such as potential temperature, the units of measurement for the data, data-quality flags, etc.

# S4 method for argo
[[(x, i, j, ...)



an argo object.


Character string indicating the name of an item to extract.


Optional additional information on the i item.


Optional additional information (ignored).


A two-step process is used to try to find the requested information. First, a class-specific function tries to find it, but if that fails, then a general function is used (see next section).

Details of the specialized argo method

There are several possibilities, depending on the nature of i. Note that argo data may contain both unadjusted data and adjusted data. By default, this extraction function refers to the former, but a preference for the latter may be set with preferAdjusted(), the documentation of which explains (fairly complex) details.

  • If i is "profile" and j is an integer vector, then an argo object is returned, as specified by j. For example, argo[["profile", 2:5]] is equivalent to subset(argo, profile \%in\% 2:5).

  • If i is "CT", then Conservative Temperature is returned, as computed with gsw::gsw_CT_from_t(SA,t,p), where first SA is computed as explained in the next item, t is in-situ temperature, and p is pressure.

  • If i is "N2", then the square of buoyancy is returned, as computed with swN2().

  • If i is "SA", then Absolute Salinity is returned, as computed with gsw::gsw_SA_from_SP().

  • If i is "sigmaTheta", then potential density anomaly (referenced to zero pressure) is computed, with swSigmaTheta(), where the equation of state is taken to be getOption("oceEOS",default="gsw").

  • If i is "theta", then potential temperature (referenced to zero pressure) is computed, with swTheta(), where the equation of state is taken to be getOption("oceEOS",default="gsw").

  • If i is "depth", then a matrix of depths is returned.

  • If i is in the data slot of x, then it is returned, otherwise if it is in the metadata slot, then that is returned, otherwise NULL is returned.

Details of the general method

If the specialized method produces no matches, the following generalized method is applied. As with the specialized method, the procedure hinges first on the values of i and, optionally, j. The work proceeds in steps, by testing a sequence of possible conditions in sequence.

  1. A check is made as to whether i names one of the standard oce slots. If so, [[ returns the slot contents of that slot. Thus, x[["metadata"]] will retrieve the metadata slot, while x[["data"]] and x[["processingLog"]] return those slots.

  2. If i is a string ending in the "Unit", then the characters preceding that string are taken to be the name of an item in the data object, and a list containing the unit is returned. This list consists of an item named unit, which is an expression(), and an item named scale, which is a string describing the measurement scale. If the string ends in " unit", e.g. x[["temperature unit"]] (note the space), then just the expression is returned, and if it ends in " scale", then just the scale is returned.

  3. If i is a string ending in "Flag", then the corresponding data-quality flag is returned (or NULL if there is no such flag). For example, x[["salinityFlag"]] returns a vector of salinity flags if x is a ctd object.

  4. If i is "sigmaTheta", then the value of swSigmaTheta() is called with x as the sole argument, and the results are returned. Similarly, swSigma0() is used if i="sigma0", and swSpice() is used if i="spice". Of course, these actions only make sense for objects that contain the relevant items within their data slot.

  5. After these possibilities are eliminated, the action depends on whether j has been provided. If j is not provided, or is the string "", then i is sought in the metadata slot, and then in the data slot, returning whichever is found first. In other words, if j is not provided, the metadata slot takes preference over the data slot. However, if j is provided, then it must be either the string "metadata" or "data", and it directs where to look.

If none of the above-listed conditions holds, then NULL is returned, without the issuance of a warning or error message. (This silent operation is employed so that [[ will behave like the normal R version.)

See also


data(argo) # 1. show that dataset has 223 profiles, each with 56 levels dim(argo[['temperature']])
#> [1] 56 223
# 2. show importance of focussing on data flagged 'good' fivenum(argo[["salinity"]],na.rm=TRUE)
#> [1] 0.000 34.890 34.926 35.042 47.899
#> [1] 32.850 34.890 34.926 35.042 35.842