Merge pull request #41 from JuliaClimate/improvs

Minor various improvements

docs/src/index.md

@@ -74,10 +74,11 @@ using ClimateBase # hide
 ClimateBase.COMMONNAMES
 ```
 
-Also, two convenience functions are provided for examining the content of on-disk `.nc` files without loading all data on memory.
+Also, the following convenience functions are provided for examining the content of on-disk `.nc` files without loading all data on memory.
 ```@docs
 nckeys
 ncdetails
+globalattr
 ```
 
 ### Write
@@ -114,8 +115,7 @@ function extract_dimension_values_xarray(xa, dnames = collect(xa.dims))
         if d ≠ "time"
             push!(dim_values, x)
         else
-            # This date specification assumes up to day sampling (hence the 1:10)
-            dates = [np.datetime_as_string(y)[1:10] for y in x]
+            dates = [np.datetime_as_string(y)[1:19] for y in x]
             dates = DateTime.(dates)
             push!(dim_values, dates)
         end

src/core/coredefs.jl

@@ -11,10 +11,10 @@ Time = DimensionalData.Ti
 AbDimArray = DimensionalData.AbstractDimArray
 
 export At, Between, Near # Selectors from DimensionalArrays.jl
-export hasdim, DimensionalArray
-export get_var_as_dimarray
+export hasdim, DimensionalArray, dimnum
 export Time, Lon, Lat, dims, Coord, Hei, Pre
 export EqArea, Grid, spacestructure
+export DimensionalData # for accessing its functions
 
 @dim Lon IndependentDim "Longitude"
 @dim Lat IndependentDim "Latitude"
@@ -70,7 +70,7 @@ struct ClimArray{T,N,D<:Tuple,R<:Tuple,A<:AbstractArray{T,N},Me} <: AbstractDime
 end
 ClimArray(A::DimensionalArray) = ClimArray(A.data, A.dims, A.refdims, A.name, A.metadata)
 ClimArray(A::ClimArray, dims::Tuple = A.dims; name = A.name, attrib = A.attrib) =
-ClimArray(A.data, DimensionalData.formatdims(A.data, dims), A.refdims, name, attrib)
+ClimArray(A.data, DimensionalData.formatdims(A.data, dims), A.refdims, Symbol(name), attrib)
 
 """
     ClimArray(A::Array, dims::Tuple; name = "", attrib = nothing)

src/core/nc_io.jl

@@ -5,7 +5,7 @@ https://github.com/rafaqz/GeoData.jl
 =#
 using NCDatasets
 export NCDataset
-export nckeys, ncdetails
+export nckeys, ncdetails, globalattr
 export climarrays_to_nc
 
 dim_to_commonname(::Lat) = "lat"
@@ -39,6 +39,16 @@ function ncdetails(file::String, io = stdout)
 end
 ncdetails(ds::NCDataset, io = stdout) = show(io, MIME"text/plain"(), ds)
 
+"""
+    globalattr(file::String) → Dict
+Return the global attributions of the .nc file.
+"""
+function globalattr(file::String)
+    NCDataset(file) do ds
+        return Dict(ds.attrib)
+    end
+end
+
 """
     ClimArray(file::Union{String,NCDataset}, var::String, name = var) -> A
 Load the variable `var` from the `file` and convert it
@@ -198,7 +208,6 @@ function climarrays_to_nc(file::String, Xs; globalattr = Dict())
                 @warn "$i-th ClimArray has no name, naming it $(n) instead."
             end
             println("processing variable $(n)...")
-            println("writing dimensions...")
             add_dims_to_ncfile!(ds, dims(X))
             println("writing the CF-variable...")
             attrib = X.attrib
@@ -215,6 +224,7 @@ function add_dims_to_ncfile!(ds::NCDatasets.AbstractDataset, dimensions::Tuple)
     dnames = dim_to_commonname.(dimensions)
     for (i, d) ∈ enumerate(dnames)
         haskey(ds, d) && continue
+        println("writing dimension $d...")
         v = dimensions[i].val
         # this conversion to DateTime is necessary because CFTime.jl doesn't support Date
         eltype(v) == Date && (v = DateTime.(v))

src/physical_dimensions/spatial_equalarea.jl

@@ -35,6 +35,9 @@ function ClimArray_eqarea(ds::NCDatasets.AbstractDataset, var::String, name = va
 
         X = ClimArray(A, Tuple(actualdims))
         X = X[Coord(si)]
+        if !any(ismissing, X)
+            X = nomissing(X)
+        end
         return ClimArray(X; name = Symbol(name), attrib)
     else
         error("Don't know how to handle this equal area grid!")

src/physical_dimensions/temporal.jl

@@ -299,15 +299,15 @@ end
 # Monthly/yearly/daily/seasonal means
 #########################################################################
 """
-    monthlyagg(A::ClimArray, f = mean) -> B
+    monthlyagg(A::ClimArray, f = mean; mday = 15) -> B
 Create a new array where the temporal daily information has been aggregated to months
 using the function `f`.
-By convention, the dates of the new array always have day number of `15`.
+The dates of the new array always have day number of `mday`.
 """
-function monthlyagg(A::ClimArray, f = mean)
+function monthlyagg(A::ClimArray, f = mean; mday = 15)
     t0 = dims(A, Time).val
-    startdate = Date(year(t0[1]), month(t0[1]), 15)
-    finaldate = Date(year(t0[end]), month(t0[end]), 16)
+    startdate = Date(year(t0[1]), month(t0[1]), mday)
+    finaldate = Date(year(t0[end]), month(t0[end]), mday+1)
     t = startdate:Month(1):finaldate
     tranges = temporalrange(t0, Dates.month)
     return timegroup(A, f, t, tranges, "monthly")

src/tsa/decomposition.jl

@@ -10,18 +10,16 @@ periodic parts of `A`, with frequencies given in `fs`, and residual contains wha
 `fs` is measured in 1/year. This function works even for non-equispaced time axis (e.g.
 monthly averages) and uses LPVSpectral.jl and SignalDecomposition.jl.
 """
-seasonal_decomposition(A::AbDimArray, b) = seasonal_decomposition(dims(A, Time), A, b)
-
-function seasonal_decomposition(t, A::AbDimArray, fs::Vector)
+function seasonal_decomposition(A::AbDimArray, fs::Vector)
     @assert hasdim(A, Time)
-    E = _numbertype(T)
+    E = _numbertype(eltype(A))
     method = Sinusoidal(E.(fs ./ DAYS_IN_ORBIT))
-    seasonal = DimensionalData.basetypeof(A)(copy(Array(A)), dims(A); name = A.name*"seasonal")
-    residual = DimensionalData.basetypeof(A)(copy(Array(A)), dims(A); name = A.name*"residual")
+    seasonal = DimensionalData.basetypeof(A)(copy(Array(A)), dims(A))
+    residual = DimensionalData.basetypeof(A)(copy(Array(A)), dims(A))
 
-    # TODO: Fix this to use "real time"
-    truetime = Float32.(cumsum(daysinmonth.(t)))
-    for i in otheridxs(T, Time)
+    t = dims(A, Time).val
+    truetime = time_in_days(t, E)
+    for i in otheridxs(A, Time)
         y = Array(A[i...])
         sea, res = SignalDecomposition.decompose(truetime, y, method)
         seasonal[i...] .= sea