Extract_Data_NetCDFfiles.Rmd

---
title: "Extract_Data_NetCDFfiles"
author: "Gustavo Facincani Dourado"
date: "8/11/2020"
output: html_document
---

```{r}
#load packages that are going to/can be used
library(dplyr)
library(raster)
library(ncdf4)
library(maptools)
library(foreign)
library(RNetCDF)
library(rgdal)
library(lubridate)
library(tidyverse)
```


```{r}
#Trying it with one file - daily total runoff for the year 2007
file<- brick("C:/Users/gusta/Desktop/PhD/CERCWET/GCMs/ACCESS1-0/rcp45/tot_runoff.2007.v0.CA_NV.nc")
#extent(file)

# number of layers
names(file)
nl <- nlayers(file)

df = list() #make the dataframe we'll produce as a list
```

```{r}
# begin of loop
for (i in 1:nl){

# Extract the raster file of layers
r <- raster(file, layer = i)

# Extract relevant information from shapefile
#Using Merced as an example, this is a shapefile with multiple polygons
#the basin is divided into 6 subwatersheds
Mer_shp <- shapefile("C:/Users/gusta/Desktop/PhD/CERCWET/GCMs/Shapefiles/Mer.shp")
Mer <-Mer_shp["SUBWAT"] #attribute to divide the subwatersheds

# Ensure command extract is from raster package
extract <- raster::extract
# Extract the mean value of cells within AMC polygon
# Alternative: look to "mask" function ?mask
masked_file<-extract(r, 
                    Mer, #shapefile
                    fun = mean, #the mean observed values in the region
                    #if I don't use the mean, I'll get the data from each point and respective weights of each point
                    na.rm=TRUE, 
                    df=T, #as a dataframe
                    small=T, 
                    sp=T,  
                    weights=TRUE, 
                    normalizedweights=TRUE)

# Generate variable depicting the date
# Extract the information about the time
date <- r@z[[1]] #x and y are lat-long, z = time
date
date <- as.Date(date, origin = "1800-01-01") #dates in NetCDF files are numbers that represent daily count of days since 01-01-1800, so let's read it as date


# Compile the codes for variable and time in one dataframe
df <- data.frame(date, masked_file)

colnames(df)[3] <- "total_runoff" #rename the column with the variable


#set directory where it's going to be saved
setwd("C:/Users/gusta/Desktop/PhD/CERCWET/GCMs/ACCESS1-0/rcp45/")

# save data as .csv
#If I use write.csv, I get one csv per day, so I used write.table with append = TRUE to save it all as one file
#in this case, if we use the shape with multiple polygons, each subbasin will be repeated per row
#if we use individual polygons, then we have individual csvs per subbasin
write.table(df,file="df.csv", append=TRUE,sep=",",col.names=FALSE,row.names=FALSE) 
}

```