Merge pull request #55 from PoisotLab/tp/sdt

Vignette update

.github/workflows/CompatHelper.yml

@@ -9,7 +9,7 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        julia-version: [1.6.0]
+        julia-version: [1.8.0]
         julia-arch: [x86]
         os: [ubuntu-latest]
     steps:

Project.toml

@@ -1,17 +1,17 @@
 name = "SpatialBoundaries"
 uuid = "8d2ba62a-3d23-4a2b-b692-6b63e9267be2"
 authors = ["Tanya Strydom <[email protected]>", "Timothée Poisot <[email protected]>"]
-version = "0.0.4"
+version = "0.1.0"
 
 [deps]
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
-Requires = "ae029012-a4dd-5104-9daa-d747884805df"
+SpeciesDistributionToolkit = "72b53823-5c0b-4575-ad0e-8e97227ad13b"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 VoronoiDelaunay = "72f80fcb-8c52-57d9-aff0-40c1a3526986"
 
 [compat]
-Requires = "1"
+SpeciesDistributionToolkit = "0.0.1"
 StatsBase = "0.33"
 VoronoiDelaunay = "0.4"
-julia = "1.6"
+julia = "1.8"

docs/Project.toml

@@ -3,8 +3,9 @@ Documenter = "e30172f5-a6a5-5a46-863b-614d45cd2de4"
 Literate = "98b081ad-f1c9-55d3-8b20-4c87d4299306"
 NeutralLandscapes = "71847384-8354-4223-ac08-659a5128069f"
 Plots = "91a5bcdd-55d7-5caf-9e0b-520d859cae80"
-SimpleSDMLayers = "2c645270-77db-11e9-22c3-0f302a89c64c"
+CairoMakie = "13f3f980-e62b-5c42-98c6-ff1f3baf88f0"
+SpeciesDistributionToolkit = "72b53823-5c0b-4575-ad0e-8e97227ad13b"
 StatsPlots = "f3b207a7-027a-5e70-b257-86293d7955fd"
 
 [compat]
-SimpleSDMLayers = "0.8"
+SpeciesDistributionToolkit = "0.0"

docs/src/vignettes/simplesdmlayers.jl

@@ -20,35 +20,39 @@
 # $y$, $z$) and the output data would be typed as `TriangulationWomble` objects.
 
 using SpatialBoundaries
-using SimpleSDMLayers
-using Plots
+using SpeciesDistributionToolkit
+using CairoMakie
+import Plots # Required for radial histograms
 
 # First we can start by defining the extent of the Southwestern islands of
 # Hawaii, which can be used to restrict the extraction of the various landcover
 # layers from the EarthEnv database. We do the actual layers querying using
 # `SimpleSDMLayers`:
 
 hawaii = (left = -160.2, right = -154.5, bottom = 18.6, top = 22.5)
-landcover = SimpleSDMPredictor(EarthEnv, LandCover, 1:12; hawaii...)
+dataprovider = RasterData(EarthEnv, LandCover)
+landcover_classes = SimpleSDMDatasets.layers(dataprovider)
+landcover = [SimpleSDMPredictor(dataprovider; layer=class, full=true, hawaii...) for class in landcover_classes]
 
 # We can remove all the areas that contain 100% water from the landcover data as
 # our question of interest is restricted to the terrestrial realm. We do this by
 # using the "Open Water" layer to mask over each of the landcover layers
 # individually:
 
-ow_index = findfirst(isequal("Open Water"), layernames(EarthEnv, LandCover))
-not_water = broadcast(!isequal(100), landcover[ow_index])
+ow_index = findfirst(isequal("Open Water"), landcover_classes)
+not_water = landcover[ow_index] .!== 0x64
 lc = [mask(not_water, layer) for layer in landcover]
 
 # As layers one through four of the EarthEnv data are concerned with data on
 # woody cover (*i.e.* "Evergreen/Deciduous Needleleaf Trees", "Evergreen
 # Broadleaf Trees", "Deciduous Broadleaf Trees", and "Mixed/Other Trees") we
 # will work with only these layers. For a quick idea we of what the raw
 # landcover data looks like we can sum these four layers and plot the total
-# woody cover for the Southwestern islands.
+# woody cover:
 
-tree_lc = convert(Float32, sum(lc[1:4]))
-plot(tree_lc; c = :Greens, frame = :box, leg = false)
+classes_with_trees = findall(contains.(landcover_classes, "Trees"))
+tree_lc = convert(Float32, reduce(+, lc[classes_with_trees]))
+heatmap(tree_lc; colormap=:linear_kbgyw_5_98_c62_n256)
 
 # Although we have previously summed the four landcover layers for the actual
 # wombling part we will apply the wombling function to each layer before we
@@ -57,15 +61,15 @@ plot(tree_lc; c = :Greens, frame = :box, leg = false)
 # woody cover layers. This will give as a vector containing four `LatticeWomble`
 # objects (since the input data was in the form of a matrix).
 
-wombled_layers = broadcast(wombling, (lc[1:4]))
+wombled_layers = wombling.(lc[classes_with_trees]);
 
 # As we are interested in overall woody cover for Southwestern islands we can
 # take the `wombled_layers` vector and use them with the `mean` function to get
 # the overall mean wombling value of the rate and direction of change for woody
 # cover. This will 'flatten' the four wombled layers into a single
 # `LatticeWomble` object.
 
-wombled_mean = mean(wombled_layers)
+wombled_mean = mean(wombled_layers);
 
 # From the `wombled_mean` object we can 'extract' the layers for both the mean
 # rate and direction of change. For ease of plotting we will also convert these
@@ -81,21 +85,14 @@ rate, direction = SimpleSDMPredictor(wombled_mean)
 # and this is simply for ease of plotting.
 
 b = similar(rate)
+b.grid[boundaries(wombled_mean, 0.1; ignorezero = true)] .= 1.0
 
-for t in 0.1
-    b.grid[boundaries(wombled_mean, t; ignorezero = true)] .= t
-end
+# We will overlay the identified boundaries (in green) over the rate of change
+# (in levels of grey):
 
-# In addition to being used to help find candidate boundary cells we can also
-# use this object (`b`) as masking layer when visualising wombling outputs. In
-# this case we can view the `rate` layer  in a similar fashion to the original
-# landcover layer but by masking it with `b` we only plot the candidate
-# boundaries *i.e.* the cells with the top 10% of highest rate of change values.
-
-plot(
-    plot(rate; c = :grey85, frame = :box),
-    mask(b, rate);
-    c = :black, frame = :box, colorbar = false)
+heatmap(rate, colormap=[:grey95, :grey5])
+heatmap!(b, colormap=[:transparent, :green])
+current_figure()
 
 # For this example we will plot the direction of change as radial plots to get
 # an idea of the prominent direction of change. Here we will plot *all* the
@@ -120,8 +117,8 @@ direction_candidate = mask(b, direction)
 # collect the cells and convert them from degrees to radians. Then we can start
 # by plotting the direction of change of *all* cells.
 
-stephist(
-         deg2rad.(collect(direction_all));
+Plots.stephist(
+         deg2rad.(values(direction_all));
          proj=:polar,
          lab="",
          c=:teal,
@@ -132,8 +129,8 @@ stephist(
 # Followed by plotting the direction of change only for cells that are
 # considered as candidate boundary cells.
 
-stephist(
-        deg2rad.(collect(direction_candidate));
+Plots.stephist(
+        deg2rad.(values(direction_candidate));
         proj=:polar,
         lab="",
         c=:red,

src/SpatialBoundaries.jl

@@ -5,14 +5,15 @@ using VoronoiDelaunay
 using LinearAlgebra
 using Statistics
 using StatsBase
-using Requires
+using SpeciesDistributionToolkit
 
 include(joinpath("types", "Womble.jl"))
 export Womble, TriangulationWomble, LatticeWomble
 
 include(joinpath("lib", "rateofchange.jl"))
 
 include(joinpath("lib", "wombling.jl"))
+include(joinpath("extensions", "SpeciesDistributionToolkit.jl"))
 export wombling
 
 include(joinpath("lib", "boundaries.jl"))
@@ -21,10 +22,4 @@ export boundaries
 include(joinpath("lib", "overallmean.jl"))
 export mean
 
-function __init__()
-    @require SimpleSDMLayers = "2c645270-77db-11e9-22c3-0f302a89c64c" include(
-        joinpath("extensions", "SimpleSDMLayers.jl"),
-    )
-end
-
 end # module

src/extensions/SimpleSDMLayers.jl → src/extensions/SpeciesDistributionToolkit.jl

@@ -3,7 +3,7 @@
 
 Performs a lattice wombling on a `SimpleSDMLayer`.
 """
-function wombling(layer::T; convert_to::Type = Float64) where {T <: SimpleSDMLayers.SimpleSDMLayer}
+function wombling(layer::T; convert_to::Type = Float64) where {T <: SimpleSDMLayer}
     try
         global nan = convert(convert_to, NaN)
     catch

test/Project.toml

@@ -1,4 +1,4 @@
 [deps]
 NeutralLandscapes = "71847384-8354-4223-ac08-659a5128069f"
-SimpleSDMLayers = "2c645270-77db-11e9-22c3-0f302a89c64c"
+SpeciesDistributionToolkit = "72b53823-5c0b-4575-ad0e-8e97227ad13b"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"

test/units/R1_simplesdmlayers.jl

@@ -1,13 +1,12 @@
 module SBTestSimpleSDMLayers
 
 using SpatialBoundaries
-using SimpleSDMLayers
+using SpeciesDistributionToolkit
 using Test
 
 precipitation = SimpleSDMPredictor(
-    WorldClim,
-    BioClim,
-    12;
+    RasterData(WorldClim2, BioClim),
+    layer=12;
     left = -80.0,
     right = -56.0,
     bottom = 44.0,