spatialAccurayAssessment: Performances spatial accuracy assessment for change detection...

In hamun001/STEF: Space-time change analysis of earth observation data for forest cover change monitoring and detection

Description

This function performances spatial accuracy assessment of change detection results by comparing the date of change for validation data and the date of change from change detection algorithm. See the usage example

Usage

spatialAccurayAssessment(x, NoDataValue = NULL,
  EarlyDateIsCommission = NULL, TotalSamplesize = NULL,
  snumberOfsamplefromChange = NULL,
  columWithChangeDateFromValidationData = NULL,
  columWithChangeDateFromAlgorithm = NULL)

Arguments

x	A data frame containing the date for change as per validation data and the date of change from Algorithm.The date for validation data and that of change results from algorithm must be in different columns, and the date must be specified as decimal date e.g. 2015.467. At the location where no change occured and detected, the sample point must be allocated one common unique value.
NoDataValue	Numeric or character. A unique value allocated to sample points where no change occured or detected. It can be NA, for example
EarlyDateIsCommission	Logical. If TRUE, change detected earlier than the date from the validation data is regarded as error of commission
TotalSamplesize	Numeric. The total number of sample points in the validation data
snumberOfsamplefromChange	Numeric. Number of samples from change stratum
columWithChangeDateFromValidationData	Numeric. The column number in "x (the input data frame)" that contains the date of change as per validation data. The dates must be decimal year (2015.645).
columWithChangeDateFromAlgorithm	Numeric. The column number in "x (the input data frame)" that contains the date of change as per change detection algorithm outpu. The date stamps must be decimal year (2015.645)

Details

To be completed.

Value

Returns a overall, producer's and user's accuracy

Author(s)

Eliakim Hamunyela

Examples

## Not run: 

#create a raster stack
ra <- raster(ncols=360, nrows=180)
ra[] <- rnorm(ncell(ra))
for (i in 1: 86){
ro <- raster(ncols=360, nrows=180)
ro[] <- rnorm(ncell(ro))
ra <- stack(ra, ro)
}
 #generate fake  date stamps for layers
imagedate <- decimal_date(seq(as.Date("2010-1-1"), as.Date("2016-12-30"), by = "month"))

#single pixel processing example:
rad <- rasterEngine(inraster=rasterBrick, fun=spatioTemporalMetricsExtractortryCatch,window_dims=c(windowwidth=15,windowwidth =15),
                   args=list(mYear = 2014,density = F,my_dates =imagedate,threshold = 0.01,spatiaNormPercentile =95, windowwidth =15,tryCatchError=T))

#paralell processing example:
## register the cores
sfQuickInit(cpus=5)
rad <- rasterEngine(inraster=ra, fun=spatioTemporalMetricsExtractortryCatch,window_dims=c(windowwidth=15,windowwidth =15),
                   args=list(mYear = 2014,density = F,my_dates =imagedate,threshold = 0.01,spatiaNormPercentile =95, windowwidth =15,tryCatchError=T))
writeRaster(rad, filename="test.tif",datatype ="FLT4S", overwrite=TRUE)
# Unregister the cores
sfQuickStop()

## End(Not run)

hamun001/STEF documentation built on May 23, 2019, 8:37 a.m.