Scientists have a lot of confidence in the historical surface temperature records, because they have been calculated in several different ways, and the results

Scientists have a lot of confidence in the historical surface temperature records, because they have been calculated in several different ways, and the results generally agree with each other. For instance, the ECMWF (the European weather prediction centre) has a very nice model that uses just barometer readings to calculate historical weather patterns, temperature and all. The results agree remarkably well with actual observed and reported weather. We have those records going back a long time, and because no thermometer readings are used, there is no potential for urban heat islands to bias the results. Because barometer readings generally are consistent over time (as in, sensor technology improvement doesn't really change the bias), there is no potential for a time-specific bias in surface temperature.