SPOFF it's a value that comes from difference in offsets. If you want to apply selective shaped pulse at the spectra offset (o1p) just leave SPOFF 0. On the other hand if you want to excite another region of spectrum (not in frequency offset - o1p) you must calculate SPOFF for each shaped pulse (for example pulse1 has SPOFF1, etc.). How to calculate it ? If your region of interest lies on the left from the offset just substract it's frequency from offset (one must do the calculations in Hz; SPOFF should be also enter in Hz not ppms !), value of SPOFF will be positive. In case of right sided region of o1p, final SPOFFS value will be negative, calcluations look the same as in a case of left side of spectrum.

SPOAL is a mystery for me. I have never touched it during my work with seletive pulses, so I can't help you. All I can do is quote some piece from Bruker pulse programming manual:

Phase alignment
Shapes with frequency offsets do vary the phase in order to efficiate the
frequency shift. By this variation the total phase of the shape is affected as
well. The parameter SPOAL[0..31] determines whether the phase is
aligned relative to the start or the end of the pulse. SPOAL has a range of 0
to 1. If SPOAL = 0, the relative phase shift is 0 at the beginning of the pulse
whereas it is determined by the frequency offset SPOFFS and the pulse
length at the end of the pulse. If SPOAL = 1 the relative phase shift is 0 at
the end of the pulse.

If you did not calibrate SPOFF, all selective pulses were "hitting" precisely in frequency offset (o1p or o1).

WET is rather sensitive to radiation damping effects. What probe do you have? If it's a BBI or TXI, or even worse a TXI/TCI cryoprobe, then the experiment will require major adjustment of the power of the first shape in the WET sequence (you will need more power than calculated, i.e. pl in dB should become more negative). Even with a 5mm BBO/BBFO probe at low fields, on 90% H2O sample some adjustment of sp7 may be required to get the best result.

You can get an idea of whether this is the problem, by trying a shorter pulse in the wet sequence - use 5ms for p11 instead of the default 20ms, and change the power for all of the shapes by -12dB. You may find that this improves things somewhat, at the cost of suppressing rather more of the spectrum. If this does improve things, you can use the original values, and adjust sp7 in GS mode (going to less positive values) - start in steps of 1dB (on my 500 TXI room temperature probe I needed almost -12dB adjustment), and then when you find the minimum fidarea go to steps of 0.1dB. If you have an inverse probe or cryoprobe however I would recommend max selective pulse length of 10ms rather than the default 20ms.

Is there a reason you don't want to use presat? In general one can get better results with noesygppr1d than with WET, in terms of selectivity and overall suppression, but O1 needs to be optimized carefully for best results (usually acquiring a spectrum with a 360 degree pulse, and taking the frequency of the residual sharp component as o1, is sufficient). However automating this is easier than automating the adjustment of WET to compensate for radiation damping.

NB if radiation damping is significant, then picking the top of the H2O peak in a single pulse experiment is not accurate - the peak position is shifted by radiation damping. What's the linewidth of the water in a single pulse experiment?