RStoolbox (version 0.1.1)

radCor: Radiometric Calibration and Correction

Description

Implements several different methods for radiometric calibraion and correction of Landsat data. You can either specify a metadata file, or supply all neccesary values manually. With proper parametrization apref and sdos should work for other sensors as well.

Usage

radCor(img, metaData, method = "apref", bandSet = "full", hazeValues,
  hazeBands, atmosphere, darkProp = 0.02, verbose)

Arguments

img

raster object

metaData

object of class ImageMetaData or a path to the meta data (MTL) file.

method

Radiometric conversion/correction method to be used. There are currently four methods available (see Details): "rad", "apref", "sdos", "dos", "costz".

bandSet

Numeric or character. original Landsat band numbers or names in the form of ("B1", "B2" etc). If set to 'full' all bands in the solar (optical) region will be processed.

hazeValues

Starting haze value, can be estimated using estimateHaze. if not provided and method is "dos" or "costz" hazeValues will be estimated in an automated fashion. Not needed for apparent reflectance.

hazeBands

Bands corresponding to hazeValues.

atmosphere

Character. Atmospheric characteristics. Will be estimated if not expicilty provided. Must be one of "veryClear", "clear", "moderate", "hazy" or "veryHazy".

darkProp

Numeric. Estimated proportion of dark pixels in the scene. Used only for automatic guessing of hazeValues (typically one would choose 1 or 2%).

verbose

Logical. Print status information.

Value

• RasterStack with top-of-atmosphere radiance ($W/(m^2 * srad * \mu m)$), at-satellite brightness temperature (K), top-of-atmosphere reflectance (unitless) corrected for the sun angle or at-surface reflectance (unitless).

Details

The atmospheric correction methods (sdos, dos and costz) apply to the optical (solar) region of the spectrum and do not affect the thermal band.

Dark object subtraction approaches rely on the estimation of atmospheric haze based on *dark* pixels. Dark pixels are assumed to have zero reflectance, hence the name. It is then assumed further that any radiation originating from such *dark* pixels is due to atmospheric haze and not the reflectance of the surface itself.

The folloiwing methods are available: ll{ rad Radiance apref Apparent reflectance (top-of-atmosphere reflectance) dos Dark object subtratction following Chavez (1989) costz Dark object subtraction following Chavez (1996) sdos Simple dark object subtraction. Classical DOS, Lhaze must be estimated for each band separately. }

If either "dos" or "costz" are selected, radCor will use the atmospheric haze decay model described by Chavez (1989). Depending on the atmosphere the following coefficients are used: ll{ veryClear $\lambda^{-4.0}$ clear $\lambda^{-2.0}$ moderate $\lambda^{-1.0}$ hazy $\lambda^{-0.7}$ veryHazy $\lambda^{-0.5}$ }

The implemented sun-earth distances neglect the earth's eccentricity. Instead we use a 100 year daily average (1979-2070).

References

S. Goslee (2011): Analyzing Remote Sensing Data in R: The landsat Package. Journal of Statistical Software 43(4).

Examples

library(raster)
## Import meta-data and bands based on MTL file
mtlFile  <- system.file("external/landsat/LT52240631988227CUB02_MTL.txt", package="RStoolbox")
metaData <- readMeta(mtlFile)
lsat     <- stackMeta(mtlFile)

lsat <- readAll(lsat)

## Convert DN to top of atmosphere reflectance and brightness temperature
lsat_ref <- radCor(lsat, metaData = metaData, method = "apref")

## Correct DN to at-surface-reflecatance with DOS (Chavez decay model)
lsat_sref <- radCor(lsat, metaData = metaData, method = "dos")

## Correct DN to at-surface-reflecatance with simple DOS
## Automatic haze estimation
hazeDN    <- estimateHaze(lsat, hazeBands = 1:4, darkProp = 0.01, plot = TRUE)
lsat_sref <- radCor(lsat, metaData = metaData, method = "sdos",
                    hazeValues = hazeDN, hazeBands = 1:4)