Hmm.. I measured what mooshimeter outputs when it is on 10 kohm range.
1,79 volts when open, 99 microamps at “short circuit” (100 ohm shunt in another multimeter).
With 4k7 resistor and 100 ohms extra from a cheap multimeter at microamp range, mooshimeter was showing 4796 ohms, current was 99,56 micro amps and voltage at mooshimeter connectors 478 mV. Mooshimeter seems to use constant current and then check how much voltage drops with measured resistor.
Same setup with (yet another) cheap multimeter, it seems that there is nothing constant in resistance measurement. Voltage and currents are all over the place when trying different resistors.
So, what have we learnt about this? Probably nothing. I should try with my fluke 87v, but sleeping sounds much more fun right now, as i need to be an functional electrician in the morning.
For a conclusion, voltages and currents for resistance measurements are quite small. Voltages or currents in soil might be higher than the ones multimeters are using. That has to produce errors. Devices might handle those situations differently. I think that multimeter resistance measurements in soil cant produce any useful data. Maybe you could get better results with much bigger currents, like 200 or 250 mA, which are what most installation testers use when measuring continuity of protective earth conductors around a house (in Finland, how should i know what others do).