RegionGrids.jl

Appearance

Unstructured Grids for Data Extraction ​

There are also RegionGrid types without an actual grid, maybe there are a set of coordinates and geometries that define the corners or the centres of a mesh, such as:

Basically, for each of these datasets, the data is given in such a way that the coordinates of the grid can be expressed via:

  • A Vector of Point2 types, with each Point2 type containing (lon,lat)
julia
using GeoRegions
using RegionGrids
using CairoMakie

Creating Unstructured Grids ​

A Unstructured Grid can be created as follows:

ggrd = RegionGrid(geo,Point2.(lon,lat))

where geo is a GeoRegion of interest that is found within the domain defined by the longitude and latitude grid vectors.

julia
lon = collect(10:20:360); nlon = length(lon)
lat = collect(-80:20:90); nlat = length(lat)

glon = zeros(nlon,nlat); glon .= lon;  glon = glon[:]
glat = zeros(nlon,nlat); glat .= lat'; glat = glat[:]

plon = glon .+ 14rand(nlon*nlat) .- 7
plat = glat .+ 14rand(nlon*nlat) .- 7

geo = GeoRegion([10,100,-80,10],[50,10,-40,50])

iggrd = RegionGrid(geo,Point2.(glon,glat))
pggrd = RegionGrid(geo,Point2.(plon,plat))
The VectorMask Grid type has the following properties:
    Indices             (ipoint) : [51, 55, 70, 71, 72, 73, 74, 75, 90, 91, 92, 93, 94, 108, 109]
    Longitude Points       (lon) : [-68.24406787366178, 4.001122214122667, -43.73345988981572, -28.284722405991374, -13.368821578482311, 4.036629078661573, 36.63176496594365, 47.22430306714027, -10.93308206883654, 10.504098092238973, 30.437873868054226, 46.762004631320664, 65.34644562751859, -13.009610940093523, 3.045101789724331]
    Latitude Points        (lat) : [-34.92730103496109, -14.863442634580185, -14.944662109300994, -16.44153594814545, -21.31073796876577, 0.7579686004540349, -5.524369726875455, -1.29296558853156, -1.4052908007288423, 26.56362646132839, 22.693828007190596, 17.034476179242695, 13.69831758318783, 24.172641407868177, 37.76480707736063]
    Rotated X Coordinates    (X)
    Rotated Y Coordinates    (Y)
    Rotation (°)             (θ) : 0.0
    RegionGrid Weights (weights)
    RegionGrid Size 			  : (15,) points

The API for creating a Unstructured Grid can be found here

What is in a Unstructured Grid? ​

RegionGrids.UnstructuredGrid Type
julia
UnstructuredGrid <: RegionGrid

A UnstructuredGrid is a RegionGrid that is created based on an unstructured grid often used in cubed-sphere or unstructured-mesh grids.

All UnstructuredGrid type will contain the following fields:

  • lon - A Vector of Floats, defining the longitudes for each point in the RegionGrid that describe the region.

  • lat - A Vector of Floats, defining the latitude for each point in the RegionGrid that describe the region.

  • ipoint - A Vector of Ints, defining the indices of the valid points from the original unstructured grid that were extracted into the RegionGrid.

  • weights - A Vector of Floats, defining the latitude-weights of each valid point in the grid. Will be NaN if outside the bounds of the GeoRegion used to define this RectilinearGrid.

  • X - A Vector of Floats, defining the X-coordinates (in meters) of each point in the "derotated" RegionGrid about the centroid for the shape of the GeoRegion.

  • Y - A Vector of Floats, defining the Y-coordinates (in meters) of each point in the "derotated" RegionGrid about the centroid for the shape of the GeoRegion.

  • θ - A Float storing the information on the angle (in degrees) about which the data was rotated in the anti-clockwise direction. Mathematically, it is rotation - geo.θ.

source

We see that in a UnstructuredGrid type, we have the lon and lat vectors that defined the perturbed longitude and latitude points that are within the GeoRegion.

julia
pggrd.lon
15-element Vector{Float64}:
 -68.24406787366178
   4.001122214122667
 -43.73345988981572
 -28.284722405991374
 -13.368821578482311
   4.036629078661573
  36.63176496594365
  47.22430306714027
 -10.93308206883654
  10.504098092238973
  30.437873868054226
  46.762004631320664
  65.34644562751859
 -13.009610940093523
   3.045101789724331
julia
pggrd.lat
15-element Vector{Float64}:
 -34.92730103496109
 -14.863442634580185
 -14.944662109300994
 -16.44153594814545
 -21.31073796876577
   0.7579686004540349
  -5.524369726875455
  -1.29296558853156
  -1.4052908007288423
  26.56362646132839
  22.693828007190596
  17.034476179242695
  13.69831758318783
  24.172641407868177
  37.76480707736063

An example of using Unstructured Grids ​

Say we have some sample data, here randomly generated.

julia
data = rand(nlon,nlat)[:]
162-element Vector{Float64}:
 0.3478415286387807
 0.7409595510339312
 0.7036897096177958
 0.2455304335123396
 0.946295027233593
 0.792616639538769
 0.3694604456085562
 0.9920417685517259
 0.6599081036754361
 0.6652015968995164
 ⋮
 0.16342440816846215
 0.3107232389489485
 0.028283301903656244
 0.4074328874905664
 0.6289720684155027
 0.19764277289284793
 0.49539285902055974
 0.19126118407592252
 0.9533035799896278

We extract the valid data within the GeoRegion of interest that we defined above:

julia
ndata = extract(data,iggrd)
pdata = extract(data,pggrd)
15-element Vector{Float64}:
 0.9065563425649826
 0.20896651049501969
 0.10107399443318055
 0.03420930081188911
 0.4643736510089548
 0.25617903329721536
 0.3586287285390325
 0.5032041275424867
 0.28367820889492734
 0.41604663563403455
 0.3623993460508096
 0.8677671111619495
 0.8747664293850191
 0.8091671700594689
 0.10565052424380283

And now let us visualize the results.

julia
slon,slat = coordinates(geo) # extract the coordinates
fig = Figure()

ax1 = Axis(
    fig[1,1],width=450,height=150,
    limits=(-180,360,-90,90)
)
scatter!(ax1,glon,glat,color=:lightgrey)
scatter!(ax1,plon,plat,color=data)
lines!(ax1,slon,slat,color=:black,linewidth=2)
lines!(ax1,slon.+360,slat,color=:black,linewidth=2,linestyle=:dash)

hidexdecorations!(ax1,ticks=false,grid=false)

ax2 = Axis(
    fig[2,1],width=450,height=150,
    limits=(-180,360,-90,90)
)
scatter!(ax2,iggrd.lon,iggrd.lat,color=:lightgrey)
scatter!(ax2,pggrd.lon,pggrd.lat,color=pdata)
lines!(ax2,slon,slat,color=:black,linewidth=2)

Label(fig[3,:],"Longitude / º")
Label(fig[:,0],"Latitude / º",rotation=pi/2)

resize_to_layout!(fig)
fig