Sound power varies as the inverse square of the distance. Half the distance, four times the power.
The proper sound level unit is the Bel, which is a tenfold increase of sound power. For some reason this got mangled into decibels, which are a tenth of the proper unit. Bels work to a base 10 log law, thus 2 Bels is 100 times the power of 1 Bel, 3 Bels 1000 times, and so on.
So, halving the distance form the source (in free space) will quadruple the power, but this is less than 10x, so it will be less than 1Bel, or 10dB. In fact it will be the log of 4, which is just over 0.6 of a Bel, x10 if in decibels. Because negative logs indicate smaller values, doubling the distance instead will cause the same change in a negative direction, or -6dB change.
In circuits, power is proportional to voltage squared, so a 1Bel or 10dB increase is given by a voltage increase of the square root of 10, or just over 3x. A 10x increase in voltage gives a 100x increase in power, and that one is easy - 2 Bels or 20dB.
100x (10^2) voltage gain is likewise 10^4 power increase, so 4 Bels or 40dB in sound terms.
As you mention, this only applies when the distances are significantly larger than the size of the radiating element. In the 'near field' you have one of those 'chaotic systems' where conventional maths gets into difficulties.
HTH. The whole thing is made needlessly complex by using an oddball multiplier of the unit. Bit like those people who insist on using hectopascals, dekalitres or whatever yuckspeak. Why the hell can't they use the STANDARD values?