Two Thermometers and a Warming Claim
A student downloads monthly mean temperatures from two weather stations: one at an airport 2 km from a city center, surrounded by asphalt and a parking lot, and one at a rural farm 40 km away. From 1970 to 2020, the airport station's annual mean rose 1.4 °C; the farm station's rose 0.5 °C. The student concludes that the city's global warming is 1.4 °C and that the farm data prove warming is not happening. The airport station also moved 300 m in 1998 and switched from liquid-in-glass thermometers to an automated sensor.
1. Evaluate the student's conclusion. What is the urban heat island effect, and how does station siting affect the trend?
Read the worked answer
Both conclusions are wrong.
The urban heat island is the local warming of urban and paved areas relative to surrounding rural land. Asphalt and parking lots absorb solar energy, release it slowly, reduce evapotranspiration, and add waste heat. An airport station surrounded by pavement can warm faster than the regional background, especially at night.
Therefore the airport's 1.4 °C trend is a local, contaminated trend. It is not 'the city's global warming,' and it is not a clean measure of greenhouse-driven change.
The farm's 0.5 °C rise is still warming, not proof that warming is absent. A single rural station is also local, not global. The two records differ in magnitude because of land use, elevation, latitude, instrumentation, and microsite-not because one proves the other false.
The 1998 relocation and instrument change create an inhomogeneity: a step change in the record that can masquerade as a trend. Absolute temperatures from different sites should not be compared directly; anomalies relative to a common baseline are more appropriate.
2. Why do the two records differ, and which is more useful for detecting global climate change? Explain.
Read the worked answer
They differ because each station samples a different local environment.
- The airport site has urban heat island influence and land-use change.
- The farm site is rural but still influenced by local soil moisture, elevation, and instrumentation.
- Instrument changes and relocations add non-climatic jumps.
For detecting global climate change, neither single station is sufficient. Global assessments use thousands of land stations, ocean ship and buoy data, and satellites, then homogenize records to remove station moves and instrument changes. The rural station is less contaminated by urban land use, so it is more representative of the background trend, but one farm cannot 'prove' a global trend either.
The scientifically sound comparison is of anomalies and of many stations, not of two absolute records. Both records show warming; the disagreement is about magnitude and local influence, not about the sign.
3. Describe two adjustments or data choices that would make the comparison more scientifically sound.
Read the worked answer
Two adjustments:
- Use homogenized station data. Choose a dataset that documents and corrects for relocations, instrument changes, and time-of-observation bias. The 1998 move and sensor swap must be adjusted, not ignored.
- Compare like with like and use many stations. Compare rural-only stations, or pair urban and rural stations, and compute anomalies against a common baseline. Use spatial averaging and report uncertainty.
Additional choices: satellite lower-troposphere records, borehole temperatures, or phenology as independent lines of evidence. The key judgment is that a trend from one station is a local estimate with known biases; global claims require a network and explicit uncertainty.