This seems to cover two of your big concerns, atmospheric pressure and seismic zones.
All indications are that these devices certainly are measuring gravity directly, they time the speed an object falls, either a weight or using quantum measurements on atoms like this device. If an object falls faster, that is a very clear indication that gravity is pulling that object with more force, thus stronger gravity.
See also this project
https://www.bipm.org/utils/common/pdf/final_reports/M/G-K1/EURAMET.M.G-K1.pdf and this
https://www.bipm.org/utils/common/pdf/final_reports/M/G-K1/SIM.M.G-K1.pdf which are specific projects to determine the
uncorrected numbers from high precisision gravimeters. Table 3 in the first link gives the measurements at Walferdange. Table 5 in the second link gives measurements at Colorado.
I checked both sets of measurements against the IGS formula which is a theoretical value using inputs of just latitude and height. They agree closely, although the absolutely gravimeter values are clearly going to be ‘correct’.
The results conclusively indicate
(i) that observed acceleration changes significantly both with latitude and height. Thus the acceleration at Walferdange (lat 50.884635, elevation 405) is 980.96395 cm/s^2. At Colorado (lat 40.13080, elevation 1682) it is 979.62274 cm/s^2.
(ii) that observed acceleration agrees closely with predicted acceleration. My predictions were 981.02707 for Walferdange and 979.66495
for Colorado.
So to some extent Tom’s experiment is repeating existing scientific results, although I like the idea of an old school approach using spring based mechanisms that people can understand, and I fully support his idea for a project.