0. Set up training dataset.Rmd

---
title: "0. Set up training dataset"
author: "Xiao Yang"
date: "8/10/2019"
output: html_document
---

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

merge manually classified areas and assign image id to each polygon

```{r}
dat1 = st_read("data/lake_ice_training_data1.shp")
dat2 = st_read("data/lake_ice_training_data2.shp")
dat = rbind(dat1, dat2)

save(dat, file = "outputs/training_polygons_20190808.RData")
summary(dat)

class_count_raw = dat %>% 
  ggplot() +
  geom_bar(aes(x = class)) +
  theme(axis.text.x.bottom = element_text(angle = 45, hjust = 1)) +
  labs(
    x = "Class",
    y = "Count"
  )

class_count_raw

class_count_raw %>% 
  ggsave(filename = "figs/class_count_raw.png",
         width = 5,
         height = 4)


hl = st_read("data/HydroLAKES_polys_v10_shp/HydroLAKES_polys_v10.shp")
imgIds = read_csv("outputs/subsamples_training_lakes_20190802.csv")

dat = dat %>% 
  mutate(polygonId = 1:nrow(dat))

dat_ids = dat %>% 
  st_join(hl %>% dplyr::select(Hylak_id, Elevation), left = T) %>% 
  left_join(imgIds %>% dplyr::select(LANDSAT_SCENE_ID, Hylak_id, lif, temp), by = "Hylak_id")

dat_ids = dat_ids %>% 
  filter(!is.na(LANDSAT_SCENE_ID)) %>% 
  mutate(area = as.numeric(st_area(geometry)) / 1000000)

st_write(dat_ids, dsn = "outputs/training_polygons_ids_20190810.shp")
```