I'm not sure if you've ever grown a plant outdoors in your life. But do you know what such 'anomalies' can do to real-world agriculture? It might seem abstract in data, until a season's crop fails, because of just two days of excessive rain, or a few days that are too hot and dry at the wrong time. These so-called anomalies can mean the difference between a harvest and a disaster. You should seriously ask farmers. They don't care about your AI-based average. They care about deviation from what nature is 'supposed' to do this time of year. That's why anomalies matter, because biology, ecosystems and food security are built around expectation and adaptation, not raw global averages.

Here in Germany even one single week without rain and temperatures between 25 and 30 degree Celcius was enough: The grass literally died. Brown, brittle, gone. That’s what an 'anomaly' does. This is really essential for understanding: Either spend a year of trying to grow plants or observe what's happening in your environment. Vegetables do not grow in the supermarket.

Shuyun Zhao here is what I think as per my understanding of anomalies. Why Is the "Reference Line" (Baseline) So Elusive in Climate Reporting?

· Baseline Choice: Climate datasets (like ERA5) use baselines—such as 1850–1900, 1981–2010, or 1991–2020—as reference periods. “Anomaly” just means the difference between today and that baseline average.

Why Not Give Absolute Temperatures?

- Technical Variability: Absolute global mean temperatures differ slightly between datasets (for ERA5: your calculation of 287.53K and the AI recommendation of 287.96K for 1991–2020 both fall within commonly reported ranges for global surface temperatures). These differences are caused by methods of interpolation, corrections, and data gaps, and by how land and ocean temperatures are integrated.

- Instrument Changes: Over decades, thermometers, satellites, and algorithms have changed—so while anomalies track relative change very accurately, absolute values can have persistent small biases.

Why The Focus on Anomalies?

- Detecting Change: The main aim of climate science is to measure change over time. Anomalies remove fixed biases and errors, spotlighting the warming trend with much higher precision than absolute numbers could.

- Comparing Different Datasets: Different datasets (e.g., ERA5, GISTEMP, HadCRUT) may report slightly different baseline values, but their anomalies track very closely, letting researchers compare “rate of change” robustly.

- Communicating Impact: Statements like “+1.1°C since 1850-1900” are easier for scientists, policymakers, and the public to relate to the Paris Agreement goals.

Why Not Also List The Absolute Average?

- It’s Actually Available: Most research datasets (including ERA5) do publish absolute monthly/annual mean temperatures, but these rarely make headlines. Official reference values for 1991–2020 in ERA5 are not always highlighted because the anomaly framework is more robust for global change.

- Small Discrepancies: Even 0.1K difference can spark confusion (like in your case), but it doesn’t change the trend of warming.

Should We Provide Absolute Temperatures With Every Report?

Many scientists agree—it’s valuable to show both. Listing the global mean (e.g., “The 1991–2020 global ERA5 2m temperature is around 287.6K, but see the dataset documentation for specifics”) and then giving the anomaly can make science more transparent for both new researchers and the public.

Your Result vs. AI Value

Your ERA5 T2m calculation (287.53K) and the AI suggestion (287.96K) are both reasonable and within reported ranges. Such differences are expected due to data processing choices[1]. What matters most is being clear and consistent about methodology and communicating both the value used and the reference frame.

Final Perspective

- Anomalies are essential and robust for studying trends.

- Absolute values are less robust due to methodological differences, but still scientifically meaningful.

- Best practice: Publish both, but communicate trends (anomalies) most loudly.

- If in doubt for a study, clearly state your data source, processing method, and chosen baseline.

This approach makes climate science both accurate and accessible while embracing honesty about uncertainty and methodological differences.

Hope this help you Shuyun Zhao

From a climate science and data analysis perspective, the widespread use of temperature anomalies rather than absolute global mean surface temperatures (GMST) stems from fundamental methodological considerations. Different observational datasets (e.g., HadCRUT, GISTEMP, NOAA) and reanalyses (e.g., ERA5) employ distinct methodologies for spatial interpolation, bias correction, and land-ocean weighting, leading to inherent variations in absolute temperature estimates - as evidenced by the 0.43K difference between your ERA5-derived value (287.53K) and other references.

Anomalies provide a more robust metric for climate monitoring because they are less sensitive to these systematic offsets between datasets while preserving the critical temperature trends. This approach is particularly valuable when incorporating historical data from the 19th century, where sparse coverage makes absolute temperature calculations highly uncertain. The scientific community's emphasis on anomalies (typically relative to 1850-1900 or more recent 30-year baselines) reflects this practical need for consistent trend analysis across different observational systems. That said, your proposal to present both absolute temperatures and their anomalies would enhance public understanding by grounding abstract anomaly values in more intuitive absolute terms. Implementing such an approach would require dataset-specific baseline references (e.g., publishing the 1991-2020 mean absolute temperature for each major dataset), followed by current absolute values and their derived anomalies. This hybrid presentation could improve transparency while maintaining the scientific rigor of anomaly-based trend analysis, potentially bridging the gap between technical climate monitoring and public communication needs.

Thank you @Carmen @Rahul @Ahmed, your answers are very helpful.