mosfet triode region

First comment: beware of SPICE2 modelling in the saturation region, as this depends only on the source-drain current, whihc means that they cannot possibly model this region. SPICE2 models will almost always overestimate the noise in the triode region of a MOSFET - and by a substantial factor if VGD < Veff.
BSIM noise models are better, albeit necessarily approximate.
Second comment: detailed behaviour will depend on the specific device.
BUT, if we take a rectangular-gate device and simplified behaviour (such that the gate-referred noise Voltage in saturated strong inversion is independent of bias), the gate-referred noise Voltage in the triode region will be somewhat smaller than that at the same gate Voltage in the saturation region, though this difference is only a couple of dB. This is mainly because the noise "averages" more uniformly. (This aspect will not necessarily apply to tapered devices)
Flicker noise current at constant device current will usually reduce significantly with increasing gate Voltage.