smoothie contains code originally contained as part of the package, SpatialVx; a package for performing weather forecast verification spatially. However, the code is potentially useful for much wider purposes than spatial weather forecast verification. It contains functions to perform convolution smoothing using several different types of kernels.
The manual for this package is given by Gilleland (2013).
Primary functions include:
Fourier2d, kernel2dsmooth and kernel2dmeitsjer
As well as the following wrapper functions, which can be useful in the context of having functions that take the same set of arguments (e.g., with the smoothing parameter as the second argument with the name lambda) for ease of allowing a user to supply their own desired kernel to a function as is utilized, for example, in the SpatialVx package.
hoods2dsmooth (neighborhood or boxcar kernel)
gauss2dsmooth (Gaussian kernel)
disk2dsmooth (Disk kernel)
identity2dsmooth (No smoothing, just returns the field)
See their help files for more information.
The functions utilize the convolution theorem along with the Fast Fourier Transform (FFT) to smooth the field (Hastie and Tibshirani, 1990; Souza, 2010)
Gilleland, E. (2013) Two-dimensional kernel smoothing: Using the R package smoothie. NCAR Technical Note, TN-502+STR, 17pp., 10.5065/D61834G2.
Hastie, T. J. and Tibshirani, R. J. (1990) Generalized Additive Models. Chapman \& Hall/CRC Monographs on Statistics and Applied Probability 43, 335pp.
Souza, C. R. (2010) Kernel Functions for Machine Learning Applications. 17 Mar 2010. Web. http://crsouza.blogspot.com/2010/03/kernel-functions-for-machine-learning.html.
# NOT RUN {
## See help files for above named functions and datasets
## for specific examples.
# }
Run the code above in your browser using DataLab