Two-dimensional kernel smoothing: Using the R package "smoothie" / by Eric Gilleland
Series: | NCAR Technical NotesBoulder, CO : National Center for Atmospheric Research (NCAR), 2013Content type:- text
- unmediated
- volume
- 2153-2397
- 2153-2400
Item type | Current library | Call number | Copy number | Status | Date due | Barcode | Item holds |
---|---|---|---|---|---|---|---|
REPORT | NCAR Library Mesa Lab | 03704 | 1 | Available | 50583020000026 |
2013
Technical Report
Applying a kernel smoother to a two-dimensional field can be a laborious and computationally expensive process if carried out in the most obvious fashion (applying a double loop). One might consider the alternative of stacking a location matrix and applying the kernel smoother in a more efficient way (e.g., through the R function aggregate). However, for large grids this may not be practical or even possible. The methods have the advantage of being able to handle missing values and edges (with fewer neighbors) directly, but if these are not pivotal concerns, the convolution theorem along with the fast Fourier transform (FFT) provides a very speedy alternative. The R package smoothie provides functionality for smoothing a two-dimensional field (or image) using the convolution theorem and FFT, and is used extensively in the spatial weather forecast verification package, SpatialVx.