index.Rmd

---
title: "Introduction to landscape ecology with R"
author: "Jakub Nowosad and Maximilian H.K. Hesselbarth"
institute: "..."
date: "2020-05-14"
output: 
  xaringan::moon_reader:
    seal: false
    css: [default, metropolis, metropolis-fonts, "style.css"]
    lib_dir: libs
    self_contained: false
    nature:
      highlightStyle: github
      highlightLines: true
      countIncrementalSlides: false
      ratio: 16:10
---
class: inverse, left, nonum, clear
background-image: url("figs/cover.jpg")
background-size: cover

.titlestyle[Introduction] 
<br>
.titlestyle[to]
<br>
.titlestyle[landscape]
<br>
.titlestyle[ecology]
<br>
.titlestyle[with R]


<link rel="stylesheet" type="text/css" href="//fonts.googleapis.com/css?family=Lora" />

.captionstyle[Jakub Nowosad and Maximilian H.K. Hesselbarth]

.captionstyle[2020-05-14, the IALE-North America 2020 Annual Meeting]


<!-- https://www.si.edu/object/bright-scene-cattle-near-stream:saam_1983.104.2 -->

```{r setup, echo  = FALSE, results = "hide", message = FALSE, warning = FALSE, purl=FALSE}
library(dplyr)
library(extrafont)
library(ggplot2)
library(gganimate)
library(knitr)
library(landscapemetrics)
library(purrr)
library(raster)
library(sp)

knitr::opts_chunk$set(warning = FALSE, message = FALSE, fig.align = "center")
extrafont::fonts()
options(tibble.print_min = 5)
options(tibble.print_max = 5)
```

---
# Rules of the workshop

- Please set your microphones to **mute**
- We encourage participants to use their **real names and a profile picture**
- Participants will be split into **random groups during the practical parts**. 
Feel free to discuss your problems and solutions with your group
- **The workshop will be recorded** and the video will be publicly posted afterward

```{r zoom, purl=FALSE, echo=FALSE}
knitr::include_graphics("figs/zoom.png")
```

- There are **two ways to ask questions** during the workshop:
    1. **The zoom chat** - this is a place to ask questions related to the currently discussed content and questions related to the practicals. 
    We will try to answer these questions as soon as possible.
    2. **The slido portal** - https://www.sli.do/IALENA - this is a place to ask questions related to overall spatial data analysis in R and landscape ecology with R.
    We will answer these questions at the end of the workshop
    
---
# Prerequisites

**Packages:**
- You can install the packages used in the workshop as follows:

```{r, eval=FALSE, purl=FALSE}
install.packages(c("landscapemetrics", "sf", "dplyr", "spData", "usethis", "remotes")
install.packages("spDataLarge", repos = "https://nowosad.github.io/drat/", type = "source")
remotes::install_github("mtennekes/tmaptools", upgrade = TRUE)
remotes::install_github("mtennekes/tmap", upgrade = TRUE)
remotes::install_github("rspatial/raster", upgrade = TRUE)
```

**Slides:** 
- https://r-spatialecology.github.io/ialena-2020/

**Scripts and data:**
- The code below can be used to download all the scripts and data for the workshop:

```{r, eval=FALSE, purl=FALSE}
usethis::use_course("https://github.com/r-spatialecology/ialena-2020/raw/master/ialena-2020.zip")
```

---
# Workshop agenda

- **Part I** (90 minutes): Introduction to the spatial data analysis in R:
    - Spatial data representations and reading
    - *First practical part*
    - Making maps in R
    - R spatial processing power
    - *Second practical part*
- **Break** (30 minutes)
- **Part II** (90 minutes): *landscapemetrics*: An open-source R tool to calculate landscape metrics:
    - Introdution to landscape metrics and how to calculate landscape metrics in R
    - *First practical part*
    - Sample metrics in buffer areas, moving window, and building blocks
    - *Second practical part*
- **Q and A**:
    - You can ask us any questions related to spatial data analysis in R and the *landscapemetrics* package during the workshop using https://www.sli.do/IALENA
    - The event code is `IALENA`

<!-- MH: Do we need one more slide already here talking quickly about the structure of the exercises? Like...they should download the scripts without solutions and try to fill them out etc... ?-->

---
class: inverse, left, bottom, clear, nonum

layout: false

background-image: url("figs/geocompr-logo.png")
background-position: 85% 10%
background-size: 30%

# Part I

## Introduction to the spatial data analysis in R

Jakub Nowosad

Institute of Geoecology and Geoinformation, Adam Mickiewicz University, Poznan, Poland

---
# Spatial data representations: raster data

.lc[
- The **raster** package
- Classes for spatial raster data: `RasterLayer`, and multilayer `RasterStack` and `RasterBrick`
- Raster data input/output
- Raster algebra and raster processing
- ?`raster-package`, [link1](https://geocompr.robinlovelace.net/spatial-class.html#raster-data), [link2](https://rspatial.org/raster/index.html), and [link3](http://www.rpubs.com/etiennebr/visualraster)
]

.rc[
```{r, echo=FALSE, purl=FALSE, out.height=200}
knitr::include_graphics("figs/raster-intro-plot-1.png", dpi = NA)
```

<!-- MH: The source will always be the path of the last person that knitted the slides. Is that a problem? Not really, right? -->
```{r, message=FALSE}
library(raster)
elev_data <- raster("data/example_elevation.tif")
elev_data
```
]


---
# Spatial data representations: vector data

.pull-left[
```{r,echo=FALSE, purl=FALSE, message=FALSE}
library(spData)
library(sf)
library(tmap)
study_area <- read_sf("data/study_area.gpkg")
tmap_mode("view")
tm_shape(study_area) + 
    tm_borders()
tmap_mode("plot")
```

]

.pull-right[

```{r}
library(sf)
study_area <- read_sf("data/study_area.gpkg")
head(study_area)
```

- The **sf** package
- A class system for spatial vector data 
- Vector data input/output
- Transformations between different coordinate reference systems (CRS)
- Geometric operations
- [link1](https://geocompr.robinlovelace.net/spatial-class.html#vector-data), [link2](https://r-spatial.github.io/sf/), [link3](https://github.com/rstudio/cheatsheets/blob/master/sf.pdf)
]


---
# Spatial data reading

.pull-left[
<!-- https://www.mrlc.gov/data/legends/national-land-cover-database-2011-nlcd2011-legend -->

```{r, echo=FALSE, purl=FALSE, eval=FALSE}
library(dplyr)
library(tidyr)
lc_df = tibble::tribble(
                                 ~V1,
                    "11. Open Water",
            "12. Perennial Ice Snow",
         "21. Developed, Open Space",
      "22. Developed, Low Intensity",
   "23. Developed, Medium Intensity",
      "24. Developed High Intensity",
           "31. Bare Rock/Sand/Clay",
              "41. Deciduous Forest",
              "42. Evergreen Forest",
                  "43. Mixed Forest",
                   "51. Dwarf Shrub",
                   "52. Shrub/Scrub",
         "71. Grasslands/Herbaceous",
              "72. Sedge/Herbaceous",
                          "74. Moss",
                   "81. Pasture/Hay",
              "82. Cultivated Crops",
                "90. Woody Wetlands",
  "95. Emergent Herbaceous Wetlands"
  ) %>% separate(V1, c("ID", "lc_category"), sep = "\\.") %>% 
  mutate(ID = as.integer(ID), 
         lc_category = stringr::str_trim(lc_category))
```


```{r, echo=FALSE, purl=FALSE, eval=FALSE}
library(raster)
lc_data <- raster("data/example_landscape.tif") 
levels(lc_data)[[1]] <- levels(lc_data)[[1]] %>% 
  left_join(lc_df, by = "ID") %>% 
  dplyr::select(-category)
levels(lc_data)[[1]] <- na.omit(levels(lc_data)[[1]])

writeRaster(lc_data, "data/example_landscape33.tif", options = "COMPRESS=DEFLATE")
```

```{r}
library(sf)
library(raster)
study_area <- read_sf("data/study_area.gpkg")
lc_data <- raster("data/example_landscape.tif") 
```

```{r, echo=FALSE, purl=FALSE, eval=FALSE}
library(elevatr)
elev_data <- elevatr::get_elev_raster(extend(lc_data, y = c(100, 100)), z = 11)
elev_data <- resample(elev_data, lc_data, method = "bilinear")
elev_data2 <- projectRaster(elev_data, crs = "+proj=longlat +datum=WGS84 +no_defs")
elev_data3 <- projectRaster(elev_data2, crs = lc_data)
writeRaster(elev_data2, "data/example_elevation.tif", options = "COMPRESS=DEFLATE", overwrite = TRUE)
```
]

.pull-right[
```{r}
plot(lc_data, axes = TRUE)
plot(st_geometry(study_area), add = TRUE)
```

```{r, echo=FALSE, purl=FALSE, eval=FALSE}
aa = mapedit::drawFeatures(mapview::mapview(sf::st_bbox(lc_data)))
bb = sf::st_transform(aa, sf::st_crs(lc_data))
bb$X_leaflet_id = NULL
bb$feature_type = NULL
bb$id = 1
st_crs(bb) = st_crs(bb)
write_sf(bb, "data/study_area.gpkg")
```




]


---
# Making maps in R: tmap

```{r}
library(tmap)
```

- Thematic mapping package
- It allows you to create static maps, animated maps and interactive maps.
- It works by adding subsequent layers to visualize and then modifying them.
- https://github.com/mtennekes/tmap - list of additional materials about the **tmap** package 

```{r, echo=FALSE, purl=FALSE, out.width="75%"}
knitr::include_graphics("figs/tmap-examples.png")
```

---
class: left, middle, clear, nonum

##  First practical part

Try to work through *Exercise_1_1.R*. 
The exercises will throw you in at the deep end - **you need to modify the provided code to improve the resulting map**.

During the exercise there are part where you are supposed to type your own solution indicated by the following structure:

```{r exercise_code_1_1, eval = FALSE, purl=FALSE}
# /Start Code/ #

print("Hello World") # This would be your code contribution

# /End Code/ # 
```

In case you run into problems or have any questions, please contact us at any time via *Zoom* private chat. 

---
# Making maps in R

.left-code[
```{r tm1, fig.show="hide", error=TRUE}
tm_shape(study_area) #<<
```
]

---
# Making maps in R

.left-code[
```{r tm2, fig.show="hide", fig.width=8}
tm_shape(study_area) +
  tm_borders() #<<
```
]

.right-plot[
![](`r knitr::fig_chunk("tm2", "png")`)
]

---
# Making maps in R

.left-code[
```{r tm3, fig.show="hide", fig.width=8}
tm_shape(study_area) +
  tm_borders() +
  tm_scale_bar(position = c("right",      #<<
                            "bottom")) +  #<<
  tm_compass(position = c("left", "top")) #<<
```
]

.right-plot[
![](`r knitr::fig_chunk("tm3", "png")`)
]

---
# Making maps in R

.left-code[
```{r tm4, fig.show="hide", fig.width=8}
tm_shape(study_area) +
  tm_borders() +
  tm_scale_bar(position = c("right",      
                            "bottom")) +  
  tm_compass(position = c("left", "top")) +
  tm_layout(main.title = "Study area") #<<
```
]

.right-plot[
![](`r knitr::fig_chunk("tm4", "png")`)
]

---
# Making maps in R

```{r, eval=FALSE}
tmap_mode("view") #<<
tm_shape(study_area) +
  tm_borders() +
  tm_layout(main.title = "Study area")
```

```{r,eval=FALSE}
tmap_mode("plot") #<<
tm_shape(study_area) +
  tm_borders() +
  tm_layout(main.title = "Study area")
```


---
# Making maps in R

.left-code[
```{r tm5, fig.show="hide", fig.width=8}
tm_shape(lc_data) + #<<
  tm_raster()       #<<
```
]

.right-plot[
![](`r knitr::fig_chunk("tm5", "png")`)
]

---
# Making maps in R

.left-code[
```{r tm6, fig.show="hide", fig.width=8}
tm_shape(lc_data) + 
  tm_raster() +
  tm_layout(legend.outside = TRUE) #<<
```
]

.right-plot[
![](`r knitr::fig_chunk("tm6", "png")`)
]

---
# Making maps in R

.left-code[
```{r tm7, fig.show="hide", fig.width=8}
tm_shape(lc_data) + 
  tm_raster(drop.levels = TRUE,       #<<
            title = "Land cover:") +  #<<
  tm_layout(legend.outside = TRUE)   
```
]

.right-plot[
![](`r knitr::fig_chunk("tm7", "png")`)
]

---
# Making maps in R

.left-code[
```{r tm8, fig.show="hide", fig.width=8}
tm_shape(lc_data) + 
  tm_raster(drop.levels = TRUE,       
            title = "Land cover:") +  
  tm_shape(study_area) + #<<
  tm_borders() + #<<
  tm_scale_bar(position = c("right",      #<<
                            "bottom")) +  #<<
  tm_compass(position = c("left", "top")) + #<<
  tm_layout(main.title = "Study area", #<<
            legend.outside = TRUE)  
```
]

.right-plot[
![](`r knitr::fig_chunk("tm8", "png")`)
]

---
# Making maps in R

.left-code[
```{r tm9, fig.show="hide", fig.width=8}
tm_shape(lc_data) + 
  tm_raster(drop.levels = TRUE,       
            title = "Land cover:") +  
  tm_shape(study_area) +
  tm_borders(lwd = 3, col = "black") +  #<<
  tm_scale_bar(position = c("right",      
                            "bottom")) +  
  tm_compass(position = c("left", "top")) +
  tm_layout(main.title = "Study area",
            legend.outside = TRUE)  
```
]

.right-plot[
![](`r knitr::fig_chunk("tm9", "png")`)
]


---
# Making maps in R

```{r}
elev_data <- raster("data/example_elevation.tif")
```

.left-code[
```{r tm10, fig.show="hide", fig.width=8}
tm_shape(elev_data) +  #<<
  tm_raster()  #<<
```
]

.right-plot[
![](`r knitr::fig_chunk("tm10", "png")`)
]

---
# Making maps in R

.left-code[
```{r tm11, fig.show="hide", fig.width=8}
tm_shape(elev_data) + 
  tm_raster(style = "cont",  #<<
            title = "Elevation (m asl)") +  #<<
  tm_layout(legend.outside = TRUE)  #<<
```
]

.right-plot[
![](`r knitr::fig_chunk("tm11", "png")`)
]

---
# Making maps in R

.left-code[
```{r tm12, fig.show="hide", fig.width=8}
tm_shape(elev_data) + 
  tm_raster(style = "cont",
            title = "Elevation (m asl)",
            palette = "-RdYlGn") +  #<<
  tm_layout(legend.outside = TRUE)

# tmaptools::palette_explorer()
```
]

.right-plot[
![](`r knitr::fig_chunk("tm12", "png")`)
]

---
# Making maps in R

.left-code[
```{r tm13, fig.show="hide", fig.width=8, eval=FALSE}
map1 <- tm_shape(elev_data) + 
  tm_raster(style = "cont",
            title = "Elevation (m asl)",
            palette = "-RdYlGn") +
  tm_layout(legend.outside = TRUE)

tmap_save(map1, "my_first_map.png")  #<<

tmap_save(map1, "my_first_map.html")  #<<
```
]

.right-plot[
![](`r knitr::fig_chunk("tm12", "png")`)
]

---
# R spatial processing power

.pull-left[
Attribute data operations:
- Vector attribute subsetting, aggregation and joins
- Creating new vector attributes
- Raster subsetting
- Summarizing raster objects

Spatial data operations:
- Spatial subsetting
- Topological relations
- Spatial joining
- Aggregation
- Map algebra
- Local, focal, and zonal raster operations
]

.pull-right[
Geometry operations:
- Geometric operations on vector data
- Geometric operations on raster data
- **Vector-raster interactions**

Coordinate reference systems:
- Understanding map projections
- **Reprojecting spatial data**
- Modifying map projections
]


---
# Vector-raster interactions: crop

.lc[
```{r, eval=FALSE}
lc_data
```
```{r, echo=FALSE, purl=FALSE}
tm_shape(lc_data) +
  tm_raster() +
  tm_shape(study_area) + 
  tm_borders(lwd = 3, col = "black") +
  tm_layout(legend.show = FALSE)
```

]

.rc[
```{r}
lc_data_cropped <- crop(lc_data, study_area) #<<
```
```{r, echo=FALSE, purl=FALSE, fig.width=8}
tm_shape(lc_data_cropped) +
  tm_raster() +
  tm_shape(study_area) + 
  tm_borders(lwd = 3, col = "black") +
  tm_layout(legend.show = FALSE)
```
]

---
# Vector-raster interactions: mask

.lc[
```{r, eval=FALSE}
lc_data
```
```{r, echo=FALSE, purl=FALSE}
tm_shape(lc_data) +
  tm_raster() +
  tm_shape(study_area) + 
  tm_borders(lwd = 3, col = "black") +
  tm_layout(legend.show = FALSE)
```

]

.rc[
```{r}
# lc_data_masked <- mask(crop(lc_data, study_area), study_area)
lc_data_masked <- mask(lc_data_cropped, study_area) #<<
```
```{r, echo=FALSE, purl=FALSE, fig.width=8}
tm_shape(lc_data_masked) +
  tm_raster() +
  tm_shape(study_area) + 
  tm_borders(lwd = 3, col = "black") +
  tm_layout(legend.show = FALSE)
```
]


---
# Spatial data coordinates

```{r, echo=FALSE, purl=FALSE}
polygon_sfc = st_sfc(st_polygon(list(rbind(
  c(1.2, 0.6), c(1.4, 0.6), c(1.4, 0.8), c(1.2, 0.8), c(1.2, 0.6)
))), crs = 4326)
polygon_sf = st_sf(geometry = polygon_sfc)
```

- **Geographic coordinates**: angles (degrees; longitude and latitude), pointing out locations on a spherical or ellipsoidal surface
- **Projected coordinates**: measured on a two-dimensional flat space (e.g. in metres; x and y), related to an ellipsoid by projection

.pull-left[
```{r, warning=FALSE, purl=FALSE, echo=FALSE, purl=FALSE, fig.height=6}
library(tmap)
library(sf)
polygon_sf$id = as.factor(c(1))
polygon_sf_af = polygon_sf
polygon_sf_af$geometry = polygon_sf_af$geometry + c(20, 40)
polygon_sf_afp = st_cast(polygon_sf_af, "MULTIPOINT")
tm_shape(polygon_sf_af) +
  tm_graticules() + 
  tm_polygons(col = "id", lwd = 7) +
  tm_shape(polygon_sf_afp) +
  tm_dots(size = 1) +
  tm_layout(main.title = "POLYGON", legend.show = FALSE, inner.margins = 0.15) +
  tm_xlab("Longitude", space = 0.5) + 
  tm_ylab("Latitude", space = 0.5)
```
]

.pull-right[
```{r, purl=FALSE, echo=FALSE, purl=FALSE, warning=FALSE, fig.height=6}
library(tmap)
polygon_sf_p = st_cast(polygon_sf, "MULTIPOINT")
tm_shape(polygon_sf) +
  tm_grid() + 
  tm_polygons(col = "id", lwd = 7) +
  tm_shape(polygon_sf_p) +
  tm_dots(size = 1) +
  tm_layout(main.title = "POLYGON", legend.show = FALSE, inner.margins = 0.15) +
  tm_xlab("X", space = 0.5) + 
  tm_ylab("Y", space = 0.5)
```
]


---
# Coordinate reference systems

.pull-left[
```{r}
st_crs(study_area) #<<
```
]


--


.pull-right[
```{r}
crs(lc_data) #<<
```
]

---
# Reprojecting spatial data

.pull-left[
- [A vast topic](https://geocompr.robinlovelace.net/reproj-geo-data.html#when-to-reproject)
- **Geographic Coordinate Reference Systems:** data integration, web mapping. Usually WGS84 (EPSG: 4326)
- **Projected Coordinate Reference Systems:** data analysis, local maps. Often some CRS is already used in the project/by the institution
- You can find different CRSs and their representations at https://epsg.io/

Vector data:
1. `study_area2 <- st_transform(study_area, st_crs(another_sf_obj))`
1. `study_area2 <- st_transform(study_area, crs(lc_data))`
1. `study_area2 <- st_transform(study_area,4326)`
1. `study_area2 <- st_transform(study_area, "ESRI:54030")`
]


--


.pull-right[
Raster data:
1. `elev_data2 <- projectRaster(elev_data, crs = "+proj=longlat +datum=WGS84 +no_defs")`
1. `elev_data2 <- projectRaster(elev_data, crs = lc_data)`
1. `elev_data2 <- projectRaster(elev_data, crs = st_crs(study_area)$proj4string)`

Method is important here:
1. `elev_data2 <- projectRaster(elev_data, crs = "+proj=longlat +datum=WGS84 +no_defs", method = "bilinear")`
1. `lc_data2 <- projectRaster(lc_data, crs = "+proj=longlat +datum=WGS84 +no_defs", method = "ngb")`
]

---
# Reprojecting spatial data

<!-- ?resample -->
<!-- MH: Maybe that would be a bit too much? -->

.pull-left[
```{r, eval=FALSE}
elev_data
```
<br>
<br>
```{r, echo=FALSE, purl=FALSE, fig.width=8}
tm_shape(elev_data) + 
  tm_raster(style = "cont",
            title = "Elevation (m asl)",
            palette = "-RdYlGn") +
  tm_layout(legend.show = FALSE) + 
  tm_graticules()
```


]

.pull-right[
```{r}
elev_data2 <- projectRaster(elev_data, #<<
                            crs = lc_data, #<< 
                            method = "bilinear") #<<
```
```{r, echo=FALSE, purl=FALSE, fig.width=8}
tm_shape(elev_data2) + 
  tm_raster(style = "cont",
            title = "Elevation (m asl)",
            palette = "-RdYlGn") +
  tm_layout(legend.show = FALSE) + 
  tm_grid()
```



]

---
class: left, middle, clear, nonum

##  Second practical part

Try to work through *Exercise_1_2.R*. 
The exercises will focus on understanding spatial data, interactions between raster and vector representation, and reprojecting spatial objects. 

During the exercise there are part where you are supposed to type your own solution indicated by the following structure:

```{r exercise_code_1_2, eval = FALSE, purl=FALSE}
# /Start Code/ #

print("Hello World") # This would be your code contribution

# /End Code/ # 
```

In case you run into problems or have any questions, please contact us at any time via *Zoom* private chat. 

---
# Summary

<!-- add something about r spatial ecosystem?? -->

.lc[
<a href="https://geocompr.robinlovelace.net/">
```{r, echo=FALSE, purl=FALSE, out.width="100%"}
knitr::include_graphics("figs/book_cover.jpg")
```
</a>
]

.rc[
Recent talks:
- "The landscape of spatial data analysis in R": [video](https://www.youtube.com/watch?v=bLgoqnqJ4DY) and [slides](https://nowosad.github.io/whyr_19)
- "Recent changes in R spatial and how to be ready for them": [video](https://www.youtube.com/watch?v=Va0STgco7-4) and [slides](https://nowosad.github.io/whyr_webinar004)

Resources:

- [The Geocomputation with R website](https://geocompr.github.io/): open source book, blog posts, workshop materials, and more
- [The Spatial Data Science with R website (focus on raster)](https://www.rspatial.org/): materials focused on the **raster** package
- [The r-spatial website](https://www.r-spatial.org/): blog posts and a book-in-progress
- [The GIS Stackexchange website](https://gis.stackexchange.com/questions/tagged/r): place to ask questions related to spatial data analysis in R
- `#rspatial` and `#geocompr` on Twitter
]



---
class: inverse, left, bottom, clear, nonum

# Break (30 minutes)

---
# Break (30 minutes)

.large[
- You can ask us any questions related to spatial data analysis in R and the *landscapemetrics* package during the workshop using https://www.sli.do/IALENA

- The event code is `IALENA`
 ]

<!--We could ask for questions during the intro, and then during the break, we could look at them and prepare our answers.-->
<!--we should use some online timer (or rmarkdown timer), visible to the participants-->
<!-- MH: I assume there will be plenty of question coming up during the practical as well -->

---
class: inverse, left, bottom, clear, nonum

layout: false

background-image: url("figs/landscapemetrics.png")
background-position: 85% 10%
background-size: 25%

# Part II

## *landscapemetrics*: An open-source R tool to calculate landscape metrics

Maximilian H.K. Hesselbarth

Department of Ecosystem Modelling, University of Goettingen, Germany


---
# Introdution to landscape metrics

```{r load_data_part_2, echo = FALSE, purl = FALSE}
lc_data <- raster("data/toy_landscape.tif") 
```

--
- Tools to quantify composition and configuration of landscapes

  - Composition: number and abundance of land-cover classes
  - Configuration: spatial arrangement of land-cover classes

--
- Designed for discrete patches of different land-cover classes

<center> 

--
.pull-left[
```{r aerial_lsm, out.width = "75%", out.height = "75%", echo = FALSE, purl = FALSE}
knitr::include_graphics("figs/aerial_lsm.png")
```
]

.pull-right[
```{r example_lsm, out.width = "75%", out.height = "75%", echo = FALSE, purl = FALSE}
ras_df <- as.data.frame(lc_data, xy = TRUE)

ggplot(data = ras_df) + 
  geom_raster(aes(x = x, y = y, fill = as.factor(toy_landscape))) + 
  scale_fill_manual(values = c("#98D2E1", "#EAB8BD", "#9B9555")) +
  guides(fill = FALSE) +
  coord_equal() + 
  theme_void() + 
  theme(plot.margin = unit(c(0, 0, 0, 0), "mm"))
```
]

</center>


---
layout: false

background-image: url("figs/landscapemetrics.png")
background-position: 90% 5%
background-size: 15%

# The R package *landscapemetrics*

--
- Based on [*raster*](https://CRAN.R-project.org/package=raster) package <font size="3" color="grey"> (Hijmans 2017) </font>

--
- Allows to calculate landscape metrics in transparent and reproducible workflow

--
- Tidy <font size="3" color="grey"> (sensu Wickham 2014) </font> and type-stable output of all metric functions

--
- Includes many utility functions for e.g. plotting or sampling

<br>
<center>

--
```{r levels_lsm, fig.hold = "hold", fig.align = "default", out.width = "30%", echo = FALSE, purl = FALSE}
lsm_patch <- get_patches(lc_data, class = 1)[[1]] %>% 
  as.data.frame(xy = TRUE) %>% 
  mutate(level = "patch") %>% 
  ggplot() + 
    geom_raster(aes(x = x, y = y, fill = as.factor(layer))) + 
    scale_fill_manual(values = c("#85D4E3", "#F4B5BD", "#9C964A", "#CDC08C", 
                                 "#FAD77B", "#B77F46", "#282220", "#E0D146", 
                                 "#D18F41"),
                      na.value = "grey75") +
    guides(fill = FALSE) +
    labs(title = "Patch level") +
    coord_equal() + 
    theme_void(base_size = 17.5) + 
    theme(plot.margin = unit(c(0, 0, 0, 0), "mm"), 
          text = element_text(family = "Metropolis"))

lsm_class <- ras_df %>% 
  mutate(toy_landscape = case_when(toy_landscape == 1 ~ 1), 
         level = "class") %>% 
  ggplot() + 
    geom_raster(aes(x = x, y = y, fill = as.factor(toy_landscape))) + 
    scale_fill_manual(values = "#98D2E1", na.value = "grey75") +
    guides(fill = FALSE) +
    labs(title = "Class level") +
    coord_equal() + 
    theme_void(base_size = 17.5) + 
    theme(plot.margin = unit(c(0, 0, 0, 0), "mm"), 
          text = element_text(family = "Metropolis"))

lsm_landscape <- ras_df %>% 
  mutate(level = "landscape") %>% 
  ggplot() + 
    geom_raster(aes(x = x, y = y, fill = as.factor(toy_landscape))) + 
    scale_fill_manual(values = c("#98D2E1", "#EAB8BD", "#9B9555")) +
    guides(fill = FALSE) +
    labs(title = "Landscape level") +
    coord_equal() + 
    theme_void(base_size = 17.5) + 
    theme(plot.margin = unit(c(0, 0, 0, 0), "mm"), 
          text = element_text(family = "Metropolis"))

lsm_patch 
lsm_class
lsm_landscape
```

</center>


---
# Getting ready to use *landscapemetrics*

--
Import and check data

```{r check_lsm}
# install.packages("landscapemetrics")
library(landscapemetrics)
library(dplyr)
lc_data <- raster("data/toy_landscape.tif") 
check_landscape(lc_data)
```

--
List available metrics

```{r list_metrics}
list_lsm()
```


---
# How to calculate landscape metrics

--
.pull-left[

Calculate single metric

```{r simple_calc_a}
lsm_p_area(lc_data)
```
]

--
.pull-right[

Combine several metrics to one *tibble*

```{r simple_calc_b}
class_mean <- lsm_c_area_mn(lc_data)
lsm_total <- lsm_l_ta(lc_data)

bind_rows(class_mean, lsm_total)
```
]


---
# Easy integration into larger workflows

--
Integrate easily into larger workflows

```{r larger_workflow}
set.seed(42)

sample(x = c(1, 2, 3), prob = c(0.5, 0.25, 0.25), size = 2500, replace = TRUE) %>%
  matrix(nrow = 50, ncol = 50) %>%
  raster() %>%
  lsm_p_enn() %>% #<<
  filter(value <= quantile(value, probs = 0.25) | value >= quantile(value, probs = 0.75)) %>%
  group_by(class) %>%
  summarise(n = n()) %>%
  arrange(-n)
```


---
# Calculation of multiple metrics

--
Easily calculate multiple metrics based on e.g. level and type

```{r multiple_metrics}
calculate_lsm(landscape = lc_data, 
              level = "landscape", type = "diversity metric", 
              classes_max = 5)
```


---
class: left, middle, clear, nonum

##  First practical part

Try to work through *Exercise_2_1.R*. 
The exercise will teach you how to calculate single and multiple landscape metrics and how to integrate *landscapemetrics* into larger workflows.

During the exercise there are part where you are supposed to type your own solution indicated by the following structure:

```{r exercise_code_1, eval = FALSE, purl = FALSE}
# /Start Code/ #

print("Hello World") # This would be your code contribution

# /End Code/ # 
```

In case you run into problems or have any questions, please contact us at any time via *Zoom* private chat.

For a general overview of *landscapemetrics*, please see <https://r-spatialecology.github.io/landscapemetrics>.


---
# Visualize patches

--
Easily visualize patches of several classes or the whole landscape

```{r, show_patches}
show_patches(augusta_nlcd, labels = FALSE, class = 42)
```


---
# Further visualization

--
.pull-left[

Visualize core areas of patches

```{r, show_cores_dummy, eval = FALSE}
show_cores(augusta_nlcd, 
           labels = FALSE, class = 42)
```

```{r, show_cores, echo = FALSE, cache = TRUE, purl = FALSE}
show_cores(augusta_nlcd, 
           labels = FALSE, class = 42, edge_depth = 5)
```
]

--
.pull-right[

Visualize landscape metrics on patch level

```{r, show_lsm_dummy, eval = FALSE}
show_lsm(augusta_nlcd, labels = FALSE, 
         class = 42, what = "lsm_p_area")
```

```{r, show_lsm, echo = FALSE, cache = TRUE, purl = FALSE}
show_lsm(augusta_nlcd, labels = FALSE, 
         class = 42, what = "lsm_p_area") + 
  theme(legend.position = "none")
```
]


---
# Sample metrics in buffer areas

--
- Samples metrics in a buffer area (sample plot) around sample points 

--
- Also able to use sample lines or polygons (as [*sp*](https://CRAN.R-project.org/package=sp)-objects; <font size="3" color="grey"> Pebesma & Bivand, 2005) </font>. 

--
- Metrics can be specified identical to `list_lsm()` and `calculate_lsm()`

--
```{r, plot_sample_points, echo = FALSE, cache = TRUE, purl = FALSE}
sample_points <- matrix(c(1253709, 1249455, 
                          1261857, 1256055,
                          1252009, 1257055), ncol = 2, byrow = TRUE)

buffer <- construct_buffer(coords = sample_points,
                           size = 1000, shape = "square", 
                           return_sp = FALSE)

ggplot(data = raster::as.data.frame(augusta_nlcd, xy = TRUE)) + 
  geom_raster(aes(x = x, y = y, fill = nlcd_2011_landcover_2011_edition_2014_10_10_NLCD.2011.Land.Cover.Class)) +
  geom_point(data = data.frame(sample_points), aes(x = X1, y = X2), 
             shape = 19, size = 2.5) +
  geom_polygon(data = data.frame(buffer), aes(x = X1, y = X2, group = X3), 
               fill = NA, col = "black", size = 1.5) +
  coord_equal() + 
  scale_fill_manual(name = "", values = c("Herbaceuous" = "#FDE9AA", 
                                          "Evergreen Forest" = "#38814E", 
                                          "Deciduous Forest" = "#85C77E", 
                                          "Hay/Pasture" = "#FBF65D", 
                                          "Developed, Open Space" = "#E8D1D1", 
                                          "Mixed Fores" = "#D4E7B0", 
                                          "Shrub/Scrub" = "#DCCA8F", 
                                          "Open Water" = "#5475A8", 
                                          "Woody Wetlands" = "#C8E6F8", 
                                          "Developed, Low Intensity" = "#E29E8C", 
                                          "Developed, Medium Intensity" = "#FF0000", 
                                          "Developed, High Intensity" = "#B50000", 
                                          "Barren Land" = "#D2CDC0", 
                                          "Emergent Herbaceuous Wetlands" = "#FDE9AA", 
                                          "Cultivated Crops" = "#CA9146")) +
  guides(fill = FALSE) + 
  theme_void()
```


---
# Sample metrics in buffer areas

--
Slightly different output *tibble*

```{r sample_lsm}
sample_points <- matrix(c(1253709, 1249455, 
                          1261857, 1256055,
                          1252009, 1257055), ncol = 2, byrow = TRUE)

sample_lsm(landscape = augusta_nlcd, y = sample_points, 
           size = 1000, what = "lsm_l_np")
```


---
# Moving window

--
- Show gradients and variability in the landscape

--
- Use matrix to specify moving window and focal cell

--
```{r moving_window_plot, echo = FALSE, cache = TRUE, purl = FALSE}
focal_i <- cbind(ras_df[ ,c(1, 2)], id = 1:nrow(ras_df))

focal_poly <- purrr::map_dfr(1:nrow(focal_i), function(i) {
  
  x_min <- focal_i[i, 1] - 4.5
  x_max <- focal_i[i, 1] + 4.5
  
  y_min <- focal_i[i, 2] - 4.5
  y_max <- focal_i[i, 2] + 4.5
  
  data.frame(x = c(x_min, x_max, x_max, x_min), 
             y = c(y_min, y_min, y_max, y_max))
}, .id = "id")

focal_poly$id <- as.integer(focal_poly$id)

focal_i <- focal_i[1:250, ]
focal_poly <- focal_poly[focal_poly$id %in% 1:250, ]

gg_window <- ggplot(data = ras_df) + 
  geom_raster(aes(x = x, y = y, fill = as.factor(toy_landscape))) + 
  geom_point(data = focal_i, aes(x = x, y = y), pch = 1) +
  geom_polygon(data = focal_poly, aes(x = x, y = y),
               fill = NA, col = "black") +
  scale_fill_manual(values = c("#98D2E1", "#EAB8BD", "#9B9555")) +
  guides(fill = FALSE) +
  coord_equal() + 
  theme_void() + 
  transition_states(id,
                    transition_length = 2,
                    state_length = 1)

gganimate::animate(gg_window, nframes = 250, fps = 5, duration = 50)
```


---
# Moving window

Results in list with *raster* objects

```{r moving_window, cache = TRUE}
window <- matrix(1, nrow = 5, ncol = 5)
window_result <- window_lsm(lc_data, window = window, what = "lsm_l_np")
```

```{r moving_window_result, echo = FALSE, out.width = "40%", cache = TRUE, purl=FALSE}
ggplot(data = raster::as.data.frame(window_result[[1]]$lsm_l_np, xy = TRUE)) + 
  geom_raster(aes(x = x, y = y, fill = factor(layer))) + 
  scale_fill_viridis_d(option = "A") +
  coord_equal() + 
  theme_void() + 
  theme(legend.position = "none")

```


---
# Building blocks

--
Fast building blocks to create your own metrics

.pull-left[

*raster* package <font size="3" color="grey"> (Hijmans 2017) </font>

```{r adj_raster}
adj_raster <- function(x) {
  adjacencies <- adjacent(x,cells = 1:ncell(x))
  table(x[adjacencies[, 1]],x[adjacencies[, 2]])
}
adj_raster(lc_data)
```
]

.pull-right[

*landscapemetrics* package

```{r adj_lsm}
get_adjacencies(lc_data)
```
]

--
```{r bench, eval = FALSE, purl = FALSE}
# A tibble: 2 x 3 
expression                      mean        mem_alloc
<chr>                           <bch:tm>    <bch:byt>
1 adj_raster(landscape)         8900 μs     1470 KB
2 get_adjacencies(landscape)    562 μs      6 KB
```


---
##  Second practical part

Try to work through *Exercise_2_2.R*.
The exercise will teach you how to use all utility functions of the *landscapemetrics* package.

During the exercise there are parts where you are supposed to code your own solution indicated by the following structure:

```{r exercise_code_2, eval = FALSE, purl = FALSE}
# /Start Code/ #

print("Hello World") # This would be your code contribution

# /End Code/ # 
```

In case you run into problems or have any questions, please contact us at any time via *Zoom* private chat.

For a general overview of *landscapemetrics*, please see <https://r-spatialecology.github.io/landscapemetrics>.


---
# Citation 

- Questions, feature requests or bugs at [https://github.com/r-spatialecology/landscapemetrics/issues](https://github.com/r-spatialecology/landscapemetrics/issues)


```{r citation}
citation("landscapemetrics")
```


---
class: left, top, clear

.pull-left[
```{r, out.width="25%", echo=FALSE, purl=FALSE, fig.align="left"}
knitr::include_graphics("figs/geocompr-logo.png")
```

## Summary:

- R gives a ["console-based ability to seamlessly switch between geographic and non-geographic data processing, modeling and visualization tasks"](https://geocompr.robinlovelace.net/intro.html)
- **landscapemetrics** allows for calculating landscape metrics for categorical landscape patterns in a tidy workflow
- We encourage bug reports, new metrics or functions suggestions, and code contributions.
For more information see [CONTRIBUTING](https://github.com/r-spatialecology/landscapemetrics/blob/master/CONTRIBUTING.md)
- This workshop is just a tip of an iceberg (of possibilities), *but you do not need to know it all*
- Questions?
]

.pull-right[
```{r, out.width="20%", echo=FALSE, purl=FALSE, fig.align="left"}
knitr::include_graphics("figs/landscapemetrics.png")
```

## Contact:

`r icon::ii_social_twitter()` [jakub_nowosad](https://twitter.com/jakub_nowosad)

`r icon::ii_social_twitter()` [MHKHesselbarth](https://twitter.com/mhkhesselbarth)

## Resources:

https://nowosad.github.io

https://mhesselbarth.rbind.io/
]

---
# References

Hijmans, R.J., 2019. raster: Geographic data analysis and modeling. R package version 2.9-5. <https://cran.r-project.org/package=raster>.

<!-- McGarigal, K., Cushman, S.A., Ene, E., 2012. FRAGSTATS v4: Spatial pattern analysis program for categorical and continuous maps. Computer software program produced by the authors at the University of Massachusetts, Amherst. <http://www.umass.edu/landeco/research/fragstats/fragstats.html>. University of Massachusetts, Amherst. -->

Pebesma, E.J., Bivand, R.S., 2005. Classes and methods for spatial data in R. R News 5(2).

<!-- Porta, C., Spano, L.D., Pontedera, F., 2017. r.li - Toolset for multiscale analysis of landscape structure. -->

<!-- VanDerWal, J., Falconi, L., Januchowski, S., Shoo, L., Storlie, C., 2019. SDMTools: Species distribution modelling tools: Tools for processing data associated with species distribution modelling exercises. R package version 1.1-221.1. <https://CRAN.R-project.org/package=SDMTools>. -->

Wickham, H., 2014. Tidy Data. J. Stat. Softw. 59, 1–23.

<!-- Wickham, H., Averick, M., Bryan, J., Chang, W., McGowan, L., François, R., Grolemund, G., Hayes, A., Henry, L., Hester, J., Kuhn, M., Pedersen, T., Miller, E., Bache, S., Müller, K., Ooms, J., Robinson, D., Seidel, D., Spinu, V., Takahashi, K., Vaughan, D., Wilke, C., Woo, K., Yutani, H., 2019. Welcome to the Tidyverse. J. Open Source Softw. 4, 1686. -->