Update environment

02-geoviz.qmd

@@ -24,6 +24,12 @@ In this book, we use **CairoMakie.jl**:
 import CairoMakie as Mke
 ```
 
+```{julia}
+#| echo: false
+#| output: false
+Mke.activate!(type = "png")
+```
+
 ::: {.callout-note}
 
 We import the backend as `Mke` to avoid polluting the Julia

03-geoio.qmd

@@ -10,6 +10,12 @@ using GeoStats
 import CairoMakie as Mke
 ```
 
+```{julia}
+#| echo: false
+#| output: false
+Mke.activate!(type = "png")
+```
+
 In order to disrupt existing practices and really develop
 something new in Julia, we had to make some hard decisions
 along the way. One of these decisions relates to how we are

04-geoproc.qmd

@@ -24,6 +24,12 @@ We will also assume that the Makie.jl backend is loaded:
 import CairoMakie as Mke
 ```
 
+```{julia}
+#| echo: false
+#| output: false
+Mke.activate!(type = "png")
+```
+
 ## Geometries
 
 We provide a vast list of geometries, which are organized into two main classes,

05-transforms.qmd

@@ -9,6 +9,12 @@ using GeoStats
 import CairoMakie as Mke
 ```
 
+```{julia}
+#| echo: false
+#| output: false
+Mke.activate!(type = "png")
+```
+
 ## Motivation
 
 In **Part I** of the book, we learned that our `GeoTable` representation

06-projections.qmd

@@ -9,6 +9,12 @@ using GeoStats
 import CairoMakie as Mke
 ```
 
+```{julia}
+#| echo: false
+#| output: false
+Mke.activate!(type = "png")
+```
+
 In this chapter, we detach the concept of `Point` from its **geospatial coordinates**
 in a given [coordinate reference system](https://en.wikipedia.org/wiki/Spatial_reference_system)
 (CRS). We explain how the same point in the physical world can be represented with multiple

08-splitcombine.qmd

@@ -9,6 +9,12 @@ using GeoStats
 import CairoMakie as Mke
 ```
 
+```{julia}
+#| echo: false
+#| output: false
+Mke.activate!(type = "png")
+```
+
 In **Part II** of the book, we introduced **transform pipelines** to
 process the `values` and the `domain` of geotables using high-level
 abstractions that preserve geospatial information. In this chapter, we

09-geojoins.qmd

@@ -9,6 +9,12 @@ using GeoStats
 import CairoMakie as Mke
 ```
 
+```{julia}
+#| echo: false
+#| output: false
+Mke.activate!(type = "png")
+```
+
 Another important tool in geospatial data science is the **geospatial join**.
 We will introduce the concept with a practical example, and will explain how it
 is related to the standard [join](https://en.wikipedia.org/wiki/Join_(SQL)) of

10-correlation.qmd

@@ -10,6 +10,12 @@ using GeoStats
 import CairoMakie as Mke
 ```
 
+```{julia}
+#| echo: false
+#| output: false
+Mke.activate!(type = "png")
+```
+
 In **Part II** and **Part III** of the book, we learned two important
 tools for *efficient* geospatial data science. We learned how **transform
 pipelines** can be used to prepare geospatial data for investigation, and

11-interpolation.qmd

@@ -9,6 +9,12 @@ using GeoStats
 import CairoMakie as Mke
 ```
 
+```{julia}
+#| echo: false
+#| output: false
+Mke.activate!(type = "png")
+```
+
 A very common task in geospatial data science is **geospatial interpolation**,
 i.e., predicting variables on geometries that lie between two or more geometries that
 have measurements. In this chapter, we will exploit **geospatial correlation**
@@ -60,7 +66,7 @@ With the measurements of the variable `z` in the geotable, and the domain of
 interpolation, we can use the `Interpolate` transform with the `IDW` model:
 
 ```{julia}
-interp = data |> Interpolate(grid, IDW())
+interp = data |> Interpolate(grid, model=IDW())
 ```
 
 ```{julia}
@@ -90,7 +96,7 @@ increasing values of the exponent:
 fig = Mke.Figure()
 Mke.Axis(fig[1,1])
 for β in [1,2,3,4,5]
-  interp = data |> Interpolate(grid, IDW(β))
+  interp = data |> Interpolate(grid, model=IDW(β))
   Mke.lines!(interp[seg, "z"], label = "β=$β")
 end
 Mke.axislegend(position = :lb)
@@ -102,7 +108,7 @@ between the two locations. In addition, the IDW solution will converge
 to the nearest neighbor solution as $\beta \to \infty$:
 
 ```{julia}
-data |> Interpolate(grid, IDW(100)) |> viewer
+data |> Interpolate(grid, model=IDW(100)) |> viewer
 ```
 
 Custom distances from [Distances.jl](https://github.com/JuliaStats/Distances.jl)
@@ -152,7 +158,7 @@ to improvements in the estimates:
 ```{julia}
 γ = GaussianVariogram(range=30.0)
 
-data |> Interpolate(grid, Kriging(γ)) |> viewer
+data |> Interpolate(grid, model=Kriging(γ)) |> viewer
 ```
 
 In the previous chapter, we learned how the **range** of the variogram determines
@@ -164,7 +170,7 @@ fig = Mke.Figure()
 Mke.Axis(fig[1,1])
 for r in [10,20,30,40,50]
   γ = GaussianVariogram(range=r)
-  interp = data |> Interpolate(grid, Kriging(γ))
+  interp = data |> Interpolate(grid, model=Kriging(γ))
   Mke.lines!(interp[seg, "z"], label = "range=$r")
 end
 Mke.axislegend(position = :lb)
@@ -252,7 +258,7 @@ fit the `Kriging` model with a maximum number of neighbors with the
 `InterpolateNeighbors` transform:
 
 ```{julia}
-interp = samples |> InterpolateNeighbors(img.geometry, Kriging(γ))
+interp = samples |> InterpolateNeighbors(img.geometry, model=Kriging(γ))
 ```
 
 ```{julia}

12-mining.qmd

@@ -230,7 +230,7 @@ We perform the interpolation of the `z` coordinate on the projected centroids of
 ```{julia}
 centroids = unique(shadow.(centroid.(blocks)))
 
-ztable = ztable |> Select("z") |> Interpolate(centroids, IDW())
+ztable = ztable |> Select("z") |> Interpolate(centroids, model=IDW())
 ```
 
 ```{julia}
@@ -307,9 +307,9 @@ up to a given maximum lag:
 ```{julia}
 maxlag = 300.0u"m"
 
-ns = setdiff(names(samples), ["geometry"])
+vs = setdiff(names(samples), ["geometry"])
 
-gs = [EmpiricalVariogram(samples, n, maxlag = maxlag) for n in ns]
+gs = [EmpiricalVariogram(samples, v, maxlag = maxlag) for v in vs]
 
 γs = [GeoStatsFunctions.fit(Variogram, g, h -> 1 / h^2) for g in gs]
 ```
@@ -330,19 +330,19 @@ function gammaplot(n, g, γ)
   Mke.current_figure()
 end
 
-gammaplot(ns[1], gs[1], γs[1])
+gammaplot(vs[1], gs[1], γs[1])
 ```
 
 ```{julia}
-gammaplot(ns[2], gs[2], γs[2])
+gammaplot(vs[2], gs[2], γs[2])
 ```
 
 ```{julia}
-gammaplot(ns[3], gs[3], γs[3])
+gammaplot(vs[3], gs[3], γs[3])
 ```
 
 ```{julia}
-gammaplot(ns[4], gs[4], γs[4])
+gammaplot(vs[4], gs[4], γs[4])
 ```
 
 Assuming that the variogram models are adequate, we can proceed to interpolation.
@@ -353,9 +353,11 @@ Given the domain of interpolation, the samples and the variogram models, we
 can perform interpolation with `InterpolateNeighbors`:
 
 ```{julia}
-models = [n => Kriging(γ) for (n, γ) in zip(ns, γs)]
+interps = map(vs, γs) do v, γ
+  samples |> Select(v) |> InterpolateNeighbors(blocks, model=Kriging(γ))
+end
 
-interp = samples |> InterpolateNeighbors(blocks, models...)
+interp = reduce(hcat, interps)
 ```
 
 Let's confirm that the interpolated values follow the same standard normal distribution: