gapfill_raster: Fill NA values in Raster* time series by interpolation (gap...
In antoinestevens/MODISExtra: Download and extract MODIS imagery

Description Usage Arguments Details Value Note Author(s) References See Also Examples

Fill NA values in Raster* time series

1
2
3

gapfill_raster(x, w=NULL, t=NULL, timeInfo = orgTime(x), method,
                      lambda = 5000, nIter= 3, outlierThreshold = NULL,
                      df = 6, ws = 2, cl = NULL, filename=rasterTmpFile(),...)

`x`	A `Raster-class` object to interpolate
`w`	Optional, a `Raster-class` object with weight information. See e.g. `makeWeights`.
`t`	Optional, a `Raster-class` object with MODIS 'composite day of the year' data.
`timeInfo`	object returned by `orgTime` defining the input and output time stamps to interpolate to.
`method`	character `vector` of length 1 with one of the following value: 'spline' for smoothing splines ,'linear' for linear approximation ,'constant' for constant approximation, 'nn' for nearest neighbour , 'whittaker' for whittaked smoothing.
`lambda`	Yearly lambda value. Default is 5000. See `whittaker.raster`.
`nIter`	numeric `vector` of length 1 giving the number of iterations for the upper envelope fitting ('whittaker' `method`). Default is 3. See `whittaker.raster`.
`outlierThreshold`	Numeric. If provided, `outlierTreshold` allows to remove outliers before smoothing by fitting a whittaker function and removing values with residuals higher than `outlierTreshold`.
`df`	Yearly degree of freedom for 'spline' `method`. Default is 6. See `smooth.spline.raster`
`ws`	window size for the 'fensholt' `method`. Default to 2. See details
`cl`	cluster object for parallel processing (do not use it if beginCluster() has been called. Default is `NULL`
`filename`	name of the file where the interpolated raster is saved. Default is created through `rasterTmpFile`
`...`	arguments passed to `writeRaster`

The 'fensholt' method fill gaps in a time serie according to the following rule: If t0 is flagged as NA (e.g. cloudy) an average between t-1, t+1 is performed; if only one of t-1 or t+1 is not missing t0 will be replaced by this value. If both t-1 and t+1 are missing then the filter is extended to cover also t-2 and t+2 following the same rules as for t-1 and t+1

A Raster-class object with NA values filled

Most of the code is borrowed from smooth.spline.raster and whittaker.raster

Antoine Stevens & Matteo Mattiuzzi

Fensholt, R., Rassmussen, K., Nielsen, T.T., and Mbow, C. (2009). Evaluation of earth observation based long term vegetation trends - Intercomparing NDVI time series trend analysis consistency of Sahel from AVHRR GIMMS, Terra MODIS and SPOT VGT data. Remote Sensing of Environment 113, 1886-1898.

interpolate_raster