time_raster: Time raster

In ropenscilabs/timeraster: Time Series Rasters

Time raster

Description

Time raster

Usage

time_raster(x, dates = NULL)

Arguments

x	one of RasterStack-class or RasterBrick-class
dates	An object with dates, one of object of class xts, or a character or Date class vector of dates. If character, coerced internally to Date

Details

This function creates TimeRaster-class, a subclass of stack in the raster package. It adds an xts time series (using only the index).

You can create a TimeRaster-class in a variety of ways:

• TimeRaster(stack, dates) - Raster/Brick stack and date vector

• TimeRaster(filelist, dates) - List of files and date vector

• TimeRaster(stack, xtsobj) - Raster/Brick stack and xts object

• TimeRaster(filelist, xtsobj) - List of files and xts object

Where stack is an already loaded Raster stack or filelist is a list of files. A simple example to get a list of files from all geotifs in a directory would be files <- list.files(path= "/data/subdir", pattern=".tif$", all.files=T, full.names=T)

dates is a list of dates of the same length as nlayers(stack). A simple creation of a datelist for the 365 days in 2014 is as.Date("2014-01-01")+0:364

If an xts object is already made you can pass that in, but only index(xtsobj) is used

The dates object need not be sequentially ordered although that would be an intuitive way to store the data - however, the dates object must be in the same order as the files or stack layers

Once created, you can use all raster methods including plot, show, arithemetic, etc

The layer subscripting as tr[[3]] works. However, the main feature is that the layer subscripting has been greatly extended. If the [[]] index is a string then time subscripting is invoked. The overall syntax for time subscripting is [daterange][aggregation][cycling], where:

• daterange - startdate [TO enddate] - as with xts, dates can be higher level. So even if the data is daily an index of "2010" can be used for all days in 2010 or "2010-10" for all days in October

• aggregation - UPTO WEEKS|MONTHS|QUARTERS|YEARS [BY SUM|MIN|MAX|MEAN|SD|COUNT][KEEPNA] - this will summarize a data set to a more aggregate time level default is to use sum with NA's omitted

• cycling - ACROSS MONTH|YEAR [BY SUM|MIN|MAX|MEAN|MEDIAN|VAR|STD|COUNT][KEEPNA] - this allows one to take say monthly averages across multiple years so if one starts with monthly date from Jan 2010 to Dec 2015 ACROSS YEAR would give stack with 12 layers (JAN-DEC) with each layer aggregated or averaged by month across the 6 years

A full complex time subscript for daily data from 1980-2015 would be: tr[["2000 TO 2009 UPTO MONTH ACROSS YEAR BY MEAN"]]. This would subscript the data to one decade, aggregate daily up to monthly data then average all 10 January data points together giving a stack with 12 layers for the 12 months each month being an average across the years 2000-2009

Two additional methods are supported to ensure that when a vector is extracted across the layers it returns an xts object:

• get_ts(tr, xcoord, ycoord) - returns an xts object at (xcoord, ycoord) in the raster

• cellStats(tr, func, ...) - returns an xts object that has been summed/averaged, etc (depending on func) across the layers

Value

An object of class TimeRaster-class

Examples

## Not run: 
zip <- system.file("examples", "prismrain.zip", package = "timeraster")
dir <- paste0(tempdir(), "/prismrain")
dir <- "prismrain"
dir.create(dir)
unzip(zip, exdir = dir)
files <- list.files(dir, full.names = TRUE, pattern = ".tif$", all.files = TRUE)

# Create raster stack from file paths
rf <- raster::stack(files)

# Create a time series object
library("xts")
ts <- xts(1:365, as.Date("2014-01-01") + 0:364)

# Create TimeRaster object
## Using new()
res <- new("TimeRaster", rf, ts = ts)

## Use convenience function time_raster()
### With list of files as input - & xts object
res <- time_raster(files, ts)
### With RasterStack class as input - & xts object
res <- time_raster(rf, ts)
### With RasterStack class as input - & Date class object
res <- time_raster(rf, as.Date("2014-01-01") + 0:364)

# plot data
plot(res[["2014-10-01 TO 2014-10-03"]])
plot(res[["UPTO MONTHS"]])

# Get data at a lat/long location
get_ts(res, -67, 45)

# Get cell stats
cellStats(res, "mean")

## End(Not run)

ropenscilabs/timeraster documentation built on May 18, 2022, 8:33 p.m.