High noise levels hamper teleseismic sheer wave splitting measurements, whi
ch bandpass filtering does not always help, To investigate how robust split
ting measurements are to noise, we analysed a set of synthetic records with
known splitting parameters and added fixed levels of noise. In the presenc
e of weak anisotropy, single-waveform splitting measurements are unreliable
when operating with noisy data sets. A practical rule in terms of S/N rati
o and splitting delay time parameters is that splitting is confidently dete
ctable at S/N > 8, regardless of the wave's original polarization orientati
on. However, for the evidence of weak anisotropy to be detectable and measu
rable at an S/N value of 4, the backazimuth separation of the phases from t
he fast polarization direction needs to be higher than 20 degrees. Stacks o
f individual measurements consistently yield reliable results down to S/N v
alues of 4. Applying stacking to data from DSB (Dublin, Ireland), the fast
polarization direction phi and lag time delta t are 58 degrees and 0.95 s.
This orientation reflects surface trends of deformation in the area, as fou
nd elsewhere in the UK. Our result thus reinforces the proposed model that
the detected anisotropy in the British Isles originates from lithospheric c
oherent deformation preserved from the last main tectonic episode.