-
-
Notifications
You must be signed in to change notification settings - Fork 10
/
modelling.qmd
153 lines (113 loc) · 5.97 KB
/
modelling.qmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
```{r,echo=FALSE,message=FALSE,warning=FALSE}
library(lidR)
```
# The Area-Based Approach (ABA) to forest modelling {#sec-modeling-aba}
This section presents a complete workflow about processing ALS point cloud data to create wall-to-wall predictions of selected forest stand attributes using an area-based approach (ABA) to forest inventories. The steps used in this vignette as well as further details about enhanced forest inventories (EFI) are described in depth in [White et al. 2013](https://cfs.nrcan.gc.ca/publications?id=34887) and [White et al. 2017](https://cfs.nrcan.gc.ca/publications?id=38945). This vignette assumes that the user has a directory of classified and normalized `.las/.laz` files. First we load the package `sf` to work with spatial data.
``` r
library(sf)
```
## Read in data {#sec-modeling-read-data}
The function `readLAScatalog()` seen in @sec-engine builds a `LAScatalog` object from a folder. Make sure the ALS files have associated index files - for details see: [Speed-up the computations on a LAScatalog](https://cran.r-project.org/web/packages/lidR/vignettes/lidR-computation-speed-LAScatalog.html).
``` r
ctg <- readLAScatalog("folder/")
print(ctg)
#> class : LAScatalog
#> extent : 297317.5 , 316001 , 5089000 , 5099000 (xmin, xmax, ymin, ymax)
#> coord. ref. : +proj=utm +zone=18 +ellps=GRS80 +towgs84=0,0,0,0,0,0,0 +units=m +no_defs
#> area : 125.19 km²
#> points : 1.58 billion points
#> density : 12.6 points/m²
#> num. files : 138
plot(ctg)
```
<center></center>
Now that our ALS data is prepared lets read in our inventory data using the `st_read()` function from the `sf` package. We read in a `.shp` file where each features is a point with a unique plot ID and corresponding plot level inventory summaries.
``` r
plots <- st_read("PRF_plots.shp")
```
Our `plots` object contains 203 plots with coordinates of plot centers, plot name, and stand attributes: Lorey's height (HL), Basal area (BA), and gross timber volume (GTV).
``` r
plots
#> Simple feature collection with 203 features and 4 fields
#> geometry type: POINT
#> dimension: XY
#> bbox: xmin: 299051.3 ymin: 5089952 xmax: 314849.4 ymax: 5098426
#> epsg (SRID): NA
#> proj4string: +proj=utm +zone=18 +ellps=GRS80 +units=m +no_defs
#> First 5 features:
#> PlotID HL BA GTV geometry
#> 1 PRF002 20.71166 38.21648 322.1351 POINT (313983.8 5094190)
#> 2 PRF003 19.46735 28.92692 251.1327 POINT (312218.6 5091995)
#> 3 PRF004 21.74877 45.16215 467.1033 POINT (311125.1 5092501)
#> 4 PRF005 27.76175 61.55561 783.9303 POINT (313425.2 5091836)
#> 5 PRF006 27.26387 39.78153 508.0337 POINT (313106.2 5091393)
```
To visualize where the plots are in our study area we can overlay them on the `LAScatalog`.
``` r
plot(ctg)
plot(plots, add = TRUE, col = "red")
```
<center></center>
We have now prepared both our ALS and plot inventory data and can begin ABA processing.
## ABA database building {#sec-modeling-build-data}
To build a database ready for modeling the steps are the following:
1. Clip the regions of interest (stands) using a shapefile of stand locations.
2. Calculate derived metrics for each stand.
3. Associate stand attributes (ID,HL,BA,GTV) to the metrics.
This can be done with a single function `plot_metrics()`. We also use the `opt_filter()` function on the `LAScatalog` to ignore noise below 0 m at read time so that they do not influence future processing. We clip discs with a `radius` of 14.1 meters because our plot radius is 14.1 m.
``` r
opt_filter(ctg) <- "-drop_z_below 0" # Ignore points with elevations below 0
D <- plot_metrics(ctg, .stdmetrics_z, plots, radius = 14.1)
```
We have now calculated a variety of ALS metrics for each plot.
## Statistical modeling {#sec-modeling-build-model}
Here we provide a simple example of how to create an OLS model (`lm`) for Lorey's Mean Height (HL). Using the `summary` and `plot` functions we found that the 85^th^ percentile of height (`zq85`) for ALS metrics explain a large amount of variation in `HL` values. We are more than happy with this simple model.
``` r
m <- lm(HL ~ zq85, data = D)
summary(m)
#> Call:
#> lm(formula = HL ~ zq85, data = D)
#>
#> Residuals:
#> Min 1Q Median 3Q Max
#> -4.640 -1.053 -0.031 1.030 4.258
#>
#> Coefficients:
#> Estimate Std. Error t value Pr(>|t|)
#> (Intercept) 2.29496 0.31116 7.376 4.2e-12 ***
#> zq85 0.93389 0.01525 61.237 < 2e-16 ***
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#>
#> Residual standard error: 1.5 on 201 degrees of freedom
#> Multiple R-squared: 0.9491, Adjusted R-squared: 0.9489
#> F-statistic: 3750 on 1 and 201 DF, p-value: < 2.2e-16
```
Here we visualize the relationship between the observed (measured) and predicted (ALS-based) HL values.
``` r
plot(D$HL, predict(m))
abline(0,1)
```
<center></center>
## Wall to wall modelling
Now that we have created a model using plot data, we need to apply that model across our entire study area.
To do so we first need to generate the same suite of ALS metrics (`.stdmetrics_z`) for our ALS data using the `pixel_metrics()` function on our original collection of files. Note that the `res` parameter is set to 25 because we want the resolution of our metrics to match the area of our sample plots (14.1 m radius = 625m^2^).
``` r
metrics_w2w <- pixel_metrics(ctg, .stdmetrics_z, res = 25, pkg = "terra")
```
To visualize any of the metrics you can use the `plot` function.
``` r
plot(metrics_w2w$zq85)
plot(metrics_w2w$zmean)
plot(metrics_w2w$pzabovezmean)
```
## Calculate wall-to-wall predictions
We can use the `predict()` function to produce `HL_pred` which is a wall-to-wall raster of predicted Lorey's mean height.
``` r
HL_pred <- predict(metrics_w2w, m)
```
To visualize wall-to-wall predictions use
``` r
plot(HL_pred)
```
<center></center>