netcdfsubset.Rmd

---
title: "NetCDF subsetting"
author: "Quentin Read"
date: "9/28/2021"
output: html_document
---

First, do a query to get the latitude and longitude values for each cell. We get a matrix for each one in the `stars` object. The first one is latitude and the second longitude.

```{r get latlon}
library(stars)
library(glue)

baseurl <- 'http://thredds.aoos.org/thredds/dodsC/GOA_RUNOFF_DISCHARGE.ncml'

query_latlon <- glue('{baseurl}?lat,lon')

# QUery a matrix of lat and long values for all cells in the netcdf
query_latlon <- 'http://thredds.aoos.org/thredds/dodsC/GOA_RUNOFF_DISCHARGE.ncml?lat[0:1:899][0:1:1809],lon[0:1:899][0:1:1809]'
latlon <- read_ncdf(query_latlon)
```

Next, we want to translate Rachael's desired extent into cell indexes so we can do a reduced query for just the cells we need. The first step is to get vectors of lats and longs.

```{r lat lon vectors}
limits <- c(-151.40, 59.44, -151.63, 59.54)
lat_vec <- as.numeric(latlon[[1]][1,])
lon_vec <- as.numeric(latlon[[2]][,1])
```

Now find the range of indexes for the desired extent. We have to subtract one because the netCDF is indexed 0-based, and R is 1-based.

```{r index ranges}
lat_limits <- range(which(lat_vec >= limits[2] & lat_vec <= limits[4])) - 1
lon_limits <- range(which(lon_vec >= limits[3] & lon_vec <= limits[1])) - 1

lat_limits

lon_limits
```

For the final query, the dimensions of `q` are ordered `(time, y, x)`. If `y` is latitude and `x` is longitude this should be correct but please check this!

```{r final query}
query_final <- glue('{baseurl}?q[0:1:12784][{lat_limits[1]}:1:{lat_limits[2]}][{lon_limits[1]}:1:{lon_limits[2]}]')
dat <- read_ncdf(query_final)

dim(dat[['q']]) 
```

Dimensions check out!