16 September 2024
This question was part of the GJ Open Forecasting website.
***What will be the monthly global land and sea temperature anomaly for July 2024, according to the National Oceanic and Atmospheric Administration (NOAA)? Github***
Over the past several decades, recorded global surface temperatures have shown a consistent upward trend, largely driven by human-induced climate change. Each year, the National Oceanic and Atmospheric Administration (NOAA) closely monitors and reports on these shifts, providing crucial insights into the state of the planet's climate. Today we will estimate the land and sea temperatute anomaly for July 2024.
The data is actually available from 1850 onwards and for each month as well. Link
Also, I would like to point out that the anomalies are with respect to the 20th century average (1901-2000).
We can actually visualize the trend from the given plot. I took the mean anomaly for each year and plotted it on the graph below. This plot illustrates the average global land and sea temperature anomaly from 1850 to 2024. Each point represents the average temperature anomaly for a given year, calculated by taking the monthly anomalies and averaging them across the year. The upward trend visible in the plot highlights the steady rise in global temperatures over the past century, particularly accelerating in recent decades.
To predict the temperature anomaly for July 2024, I analyzed data from the past 48 months, focusing particularly on the last two years. This period was chosen to ensure the prediction reflects the most recent trends and changes. Below, you can see visualizations of the trend in temperature anomalies over this timeframe, as well as the fluctuations in anomalies from month to month. The volatility was really low, around 0.11 .
Given that the most recent data points were consistently between 1.2 and 1.4 °C with relatively low volatility, I was confident that the estimate for July 2024 would remain within this range. To forecast this, I employed a linear regression model. While linear regression might not be the ideal tool for capturing such trends, it performed well (for my purpose… in the sense I was lucky to have taken the past 48 months for prediction initially. I realized after the resolution, that any other number was giving way different results). The model predicted a July 2024 anomaly of 1.27 °C, which closely matched the actual value of 1.22 °C. Additionally, the model’s forecast for August was 1.28 °C, compared to the actual 1.27 °C. Given these results and considering the forecast options: