enstools.scores.das

enstools.scores.das(obs, fct, factor=4, threshold=None)

calculate DAS - Displacement and amplitude score for observation and forecast

based on IDL programm Version 2.0 20090918 C. Keil (DLR) , G.C. Craig (DLR), H. Mannstein (DLR) and python scripts from Bettina Richter (Meteorologisches Institut - LMU Muenchen)

DAS is a displacement-based error measure and includes two components: displacement and amplitude error, where

• displacement is scaled with maximum displacement c1

• amplitude error is scaled with ‘climatological’ observed mean c2

Philosophy: complete miss equals 100% amplitude error The smaller DAS the better the forecast quality.

\[das = dis_{nor} + lse_{nor}\]

Here both components have equal weight. This is a slight modification to the application of FQM on satellite imagery taking the maximum of both components published in the reference publication [1].

Parameters:

obsxarray.DataArray or np.ndarray

2d array with observations

fctxarray.DataArray or np.ndarray

2d array with forecasts

factor: int

sub-sampling factor or topmost pyramid level (default: 4)

thresholdfloat

threshold of data (default: None)

Returns:

dict

The score itself and all related values:

das:

the DAS score

dis:

the displacement component

amp:

the amplitude component

rms_obs:

root mean square of observation

morph_o:

morphed fct to obs

morph_f:

morphed obs to fct

xdis_o:

x-displacement of fct morphed to obs

ydis_o:

y-displacement of fct morphed to obs

xdis_f:

x-displacement of obs morphed to fct

ydis_f:

y-displacement of obs morphed to fct

lse_o:

least square error of fct morphed to obs

lse_f:

least square error of fct morphed to obs

References

[1]

Keil C, Craig G.C., 2009: A displacement and amplitude score employing an optical flow technique. WAF, :doi:`10.1175/2009WAF2222247.1`

[2]

Keil C, Craig G.C., 2007: A displacement-based error measure applied in a regional ensemble forecasting system. MWR 135, 3248-3259, :doi:`10.1175/MWR3457.1`