SpatialVx (version 0.3)

upscale2d: Upscaling Neighborhood Verification on a 2-d Verification Set

Description

Perform upscaling neighborhood verification on a 2-d verification set.

Usage

upscale2d(object, thresholds = NULL, quantiles = NULL, q.gt.zero
                 = FALSE, time.point = 1, model = 1, levels = NULL,
                 max.n = NULL, smooth.fun = "hoods2dsmooth",
                 smooth.params = NULL, verbose = FALSE)
## S3 method for class 'upscale2d':
plot(x, ..., set.pw = FALSE)
## S3 method for class 'upscale2d':
print(x, ...)

Arguments

object

list object of class SpatialVx.

thresholds,quantiles

(optional) alternative thresholds (or quantiles of upscaled fields) to use on the upscaled fields. If both are NULL, the thresholds from the SpatialVx object are used.

q.gt.zero

logical, If quantiles are not NULL, then should the quantiles be calculated only for values greater than zero in the upscaled fields?

time.point

numeric or character indicating which time point from the SpatialVx verification set to select for analysis.

model

numeric indicating which forecast model to select for the analysis.

levels

numeric vector giving the successive values of the smoothing parameter. For example, for the default method, these are the neighborhood lengths over which the levels^2 nearest neighbors are averaged for each point. Values should make sense for the speci

max.n

(optional) single numeric giving the maximum neighborhood length to use. Only used if levels are NULL.

smooth.fun

character giving the name of a smoothing function to be applied. Default is an average over the n^2 nearest neighbors, where n is taken to be each value of the levels argument.

smooth.params

list object containing any optional arguments to smooth.fun. Use NULL if none.

verbose

logical, should progress information be printed to the screen?

x

list object of class upscale2d as returned by upscale2d.

set.pw

logical, should panels for the plot be set? Recommended here if you want to see all displays. Set to FALSE for printing to a pdf, for example, or for use with adding to other plots.

...

optional arguments to the image.plot function from package fields. Can also include the argument type, which must be one of all, gss, ts, bias or

Value

• upscale2d returns a list of class upscale2d with components:
• rmse
  numeric vector giving the root mean square error for each neighborhood size provided by object.
• bias,ts,ets
  numeric matrices giving the frequency bias, ts and ets for each neighborhood size (rows) and threshold (columns).

Details

Upscaling is performed via neighborhood smoothing. Here, a boxcar kernel is convolved (using the convolution theorem with FFT's) to obtain an average over the nearest n^2 grid squares at each grid point. This is performed on the raw forecast and verification fields. The root mean square error (RMSE) is taken for each threshold (Yates et al., 2006; Ebert, 2008). Further, binary fields are obtained for each smoothed field via thresholding, and frequency bias, threat score ts) and equitable threat score (ets) are calculated (Zepeda-Arce et al., 2000; Ebert, 2008).

References

Ebert, E. E. (2008) Fuzzy verification of high resolution gridded forecasts: A review and proposed framework. Meteorol. Appl., 15, 51--64. doi:10.1002/met.25

Yates, E., Anquetin, S., Ducrocq, V., Creutin, J.-D., Ricard, D. and Chancibault, K. (2006) Point and areal validation of forecast precipitation fields. Meteorol. Appl., 13, 1--20.

Zepeda-Arce, J., Foufoula-Georgiou, E., Droegemeier, K. K. (2000) Space-time rainfall organization and its role in validating quantitative precipitation forecasts. J. Geophys. Res., 105(D8), 10,129--10,146.

See Also

hoods2d, kernel2dsmooth, kernel2dmeitsjer, fft

Examples

x <- matrix( 0, 50, 50)
x[ sample(1:50,10), sample(1:50,10)] <- rexp( 100, 0.25)
y <- kernel2dsmooth( x, kernel.type="disk", r=6.5)
x <- kernel2dsmooth( x, kernel.type="gauss", nx=50, ny=50, sigma=3.5)
hold <- make.SpatialVx(x, y, thresholds=c(1e-8, 0.05, 0.1),
    field.type="random")
look <- upscale2d( hold, thresholds=seq(0.01,1,,5), levels=c(1, 3, 20))
look
plot( look)

plot( look)

data(geom001)
data(geom000)
data(ICPg240Locs)
hold <- make.SpatialVx(geom000, geom001, thresholds=c(0.01,50.01),
    loc=ICPg240Locs, projection=TRUE, map=TRUE, loc.byrow = TRUE,
    field.type="Precipitation", units="mm/h",
    data.name=c("Geometric", "geom000", "geom001"))

look <- upscale2d(hold, quantiles=c(0.25, 0.5, 0.75),
    levels=c(1, 3, 9, 17, 33, 65, 129, 257),
    verbose=TRUE)
plot(look, set.pw=TRUE)
look <- upscale2d(hold, quantiles=c(0.25, 0.5, 0.75),
    q.gt.zero=TRUE, verbose=TRUE)
plot(look)
look <- upscale2d(hold, thresholds=c(1e-8, 0.1, 0.5, 45, 50.1),
    verbose=TRUE)
plot(look)