Comment by avmich

First, Fermi estimate. You can estimate how much something weights, costs, takes to construct etc.

Then you become particularly good in it when you read texts from this area, so you gradually replace your estimates with data. And correct your errors when they disrespect reality too much. It's harder to estimate CO2 emissions from cement production from first principles; but you may have actual numbers from typical plants, which include inefficiencies.

All of that favors data and usually requires calculations, and also often requires understanding of natural sciences - conservation laws, energy conversions, speed of processes - and some economical modeling too.

A collection of links with data and models is on Azimuth Project - https://www.azimuthproject.org/azimuth/show/HomePage .