unexpected "path gains" results from tdl channel

when i set the tdl profile to 'Custom', i have to set the AveragePathGains.

Uploaded is a a script for tdl channel with custom profile, when plotting the pathgains (using mesh function) , i got unexpected result.

As shown in the attached figure the path gain for the first tap is around -16 dB. while in the configuration i set the first tap original path gain in (AveragePathGains) to -32 dB. all the values that have been set in the AveragePathGains suffer from shifting up(or adding positive values to those gains).

according to my understanding, the added values to the original path gains comes from the rayleigh distribution, but the question here, why the random value we got from the rayliegh distribution is positive always and very high. Is the random number generated from the rayliegh distribution related to the power of the tap?