exercise_solutions.Rmd

---
title: "exercise solutions"
output: html_document
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, error = TRUE, message = F)
```

```{r, echo=FALSE, message=FALSE, warning=FALSE}
library(raster)  
library(rasterVis)
library(sf)
library(dplyr)
library(ggplot2)
library(spData)

points <- st_multipoint(x = matrix(c(0, 0, 2, 0), ncol = 2, byrow = T))

line <- st_linestring(x = matrix(c(-1, -2, -0.5, -3, 2.5, -3, 3, -2), 
                                 ncol = 2, byrow = T))
```



## Vector spatial data

#### 1) add a nose!

Create a nose geometry, combine all the shapes into a single `sf` and then plot the face.

```{r}
nose <- st_linestring(x = matrix(c(1, -1, 1, -2), ncol = 2, byrow = T))
sfc <- st_sfc(points, line, nose)
face <- st_sf(data.frame(shape = c("eyes", "mouth", "nose"), geom = sfc))
plot(face)
```



#### 2) What are the coordinates for the 10th point in the Mexico polygon?

```{r}
mx_coords <- world %>% filter(iso_a2 == "MX") %>% st_coordinates() 

mx_coords[10, c("X", "Y")]
```


#### 3) How about in CRS [Mexico ITRF92 / UTM zone 15N](https://epsg.io/4488)

(hint: use `st_transform` to change the projection first)

```{r}
mx_utm15 <- world %>% 
    st_transform(crs = 4488) 

mx_coords <- mx_utm15 %>%
    filter(iso_a2 == "MX") %>% 
    st_coordinates() 

mx_coords[10, c("X", "Y")]
```

#### 4) Are these coordinates projected or not? Can you tell by just looking at the CRS in the transformed `sf` object?

```{r}
mx_utm15
```

We've already talked about UTM being projected CRSs but there is also a hint in the `proj4string`, in particular `+units=m` indicating that the units are linear (m). 

We can also extract the units from an `sf`s crs using function `sf::st_crs()` and accessing the units 

```{r}
st_crs(mx_utm15)$units
```



## Raster spatial data

#### 1) Create and plot a new `rasterLayer` of rough mean temperature in degrees C 

(rough because it would be much better to use more data at higher temporal resolution, eg at least monthly, not extremes).

```{r, echo=FALSE, message=FALSE, warning=FALSE}
library(rasterVis)
rough_mean <- raster(here::here("data", "raster", "rough_mean.tif"))
```

We can do this using a simple mean calculation:

```{r, eval=FALSE}
rough_mean <- ((full_stack[["mx.bio_5"]] + full_stack[["mx.bio_6"]])/2)/10
```

But we can even use r functions, in this case `mean()`

```{r eval=FALSE}
rough_mean <- mean(full_stack[["mx.bio_5"]], full_stack[["mx.bio_6"]])/10
```

```{r}
rough_mean
```

```{r}
levelplot(rough_mean, margin = F)
```




#### 2) Calculate mean precipitation seasonality for the **extraction bounding box area**. What is the value?

```{r echo=FALSE, message=FALSE, warning=FALSE}
mol_env_sf <- read_sf(here::here("data", "sf", "env_salamander.geojson"))
env_stack <- stack(here::here("data", "raster", "env_stack.grd"))
study_bbox <- mol_env_sf %>% st_bbox() %>% st_as_sfc()
extract_bbox <- study_bbox %>% st_buffer(dist = 1)
```


```{r}
mean_prec_seasonality <- extract(env_stack[["prec_seasonality"]], 
                as_Spatial(extract_bbox),
                fun = mean,
                na.rm = T)

mean_prec_seasonality
```


#### 3) Add a column indicating whether data points are greater than (`TRUE`) or less than (`FALSE`) extraction area mean precipitation seasonality.

```{r}
mol_env_sf <- mol_env_sf %>% 
    mutate(greater_mean_ps = 
               prec_seasonality > as.vector(mean_prec_seasonality))

mol_env_sf %>% select(locality, greater_mean_ps)
```