Let's correct two of these. Of course there are empirical observations related to climate. There is a definition of 54 "essential climate variables" (ECVs), see [1]. One of these is greenhouse gases [2]. At the bottom of the link, you can see observing programs for CO2 and CH4, both in situ and satellite.
These are well-measured variables, with well-characterized error bars, and models are tested against these observations. For instance, the ~400 ppm CO2 ECV is measured to within 1ppm quite regularly and has been for many years. This is how we improve the models. The same kind of monitoring/model-improvement feedback loop is in place for almost all of the ECVs, including clouds and sea surface properties like temperature and salinity.
Another correction, relating to "no one knew" about a future crisis in the 1980s. Here is a paper [3] from James Hansen in Science magazine, in 1981, that clearly warns of crisis-level consequences. From the abstract -- In the 21st century, we will see "creation of drought-prone regions in North America and central Asia, erosion of the West Antarctic ice sheet with a consequent worldwide rise in sea level, and opening of the fabled Northwest Passage." For whatever it's worth, this appeared on the NYTimes front page in 1981.
That's just one publication. Throughout the 1980s there were broad-based studies, mandated by Congress, that indicated the scope of the problem. See [4], from 1983.
[1] https://gcos.wmo.int/en/essential-climate-variables
[2] https://ane4bf-datap1.s3.eu-west-1.amazonaws.com/wmod8_gcos/...
[3] https://pubs.giss.nasa.gov/abs/ha04600x.html
[4] https://www.nap.edu/catalog/18714/changing-climate-report-of...
