lab8_oilSpills.Rmd

---
title: 'Intro To Git: Oil Spills'
author: "Ian Ladner"
date: "March 8, 2018"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

```{r loading}

library(tidyverse)
library(sf)
library(spatstat)
library(tmap)
library(leaflet)
library(maptools)

oil <- read_csv("data/oil_spills.csv")


texas <- filter(oil, `Accident State` == "TX" & `Accident Year` < 2017) %>% 
  group_by(`Accident Year`) %>% 
  summarise(Loss = sum(`Net Loss (Barrels)`))

colnames(texas)<- c("Year","Loss")


ggplot(texas, aes(x = Year, y = Loss))+geom_col()+theme_classic()

```

```{r}
df_loc <- oil %>% 
  filter(`Accident State` == "TX" & `Accident Year` == 2016) %>% 
  select(Latitude, Longitude, `Net Loss (Barrels)`)

colnames(df_loc) <- c("Lat","Long","Loss")

oil_sf <- st_as_sf(df_loc, coords = c("Long","Lat"), crs = 4326)

leaflet(oil_sf) %>% 
  addTiles() %>% 
  addMarkers()
```

```{r tmap}

states <- st_read(dsn = "data", layer = "states")

tex_border <- states %>% 
  filter(STATE_NAME == "Texas") %>% 
  st_transform(4326)

plot(tex_border)

tm_shape(tex_border)+
  tm_polygons()+
  tm_shape(oil_sf)+
  tm_dots(size = 0.7)

```


Point pattern analysis. R is finicky.

```{r spatialAnalysis}

spill_sp <- as(oil_sf, "Spatial")

spill_ppp <- as(spill_sp, "ppp")

#but what's our bounding window

tx_sp <- as(tex_border, "Spatial")
tx_owin <- as(tx_sp, "owin")


#but outer window and points together
all_ppp <- ppp(spill_ppp$x, spill_ppp$y, window = tx_owin)
```


##A density plot

```{r}
plot(density(all_ppp, sigma = 0.4))
```



```{r quadrantTest}


#nx = number of horz subdivisions, ny = number of vert subdivisions
#null hyp: CSR is demonstrated by data
oil_qt <- quadrat.test(all_ppp, nx = 5, ny = 5)


#What do these numbers mean? Ask Allison.
plot(all_ppp)
plot(oil_qt, add = TRUE, cex = 0.5)

```

G function for nearest neighbor analysis

```{r Gfunction}

r <- seq(0,1,0.1)

oil_gFun <- envelope(all_ppp, fun = Gest, r = r, nsim = 100)


ggplot(oil_gFun, aes(x = r, y = obs))+geom_line(color = "black")+geom_line(aes(x = r, y = theo),color = "red")
```

Nearest neighbor using the L (or K) function which looks at concentration of points on the whole

```{r}

lag2 <- seq(0,3,0.5)

oil_lFun <- envelope(all_ppp, fun = Lest, r = lag2, nsim = 20, global = TRUE)


ggplot(oil_lFun, aes(x = r, y = obs))+geom_line(color = "coral3")+geom_line(aes(x = r, y = theo), color = "darkgreen")

#clustered
```