Validation of ERA5 and ERA5-Land Reanalysis Data for Daily Air Temperature and Extreme Temperature Indices over Bangladesh

The Earth's climate is shifting, and understanding how temperatures have changed over the last few decades is essential for predicting what lies ahead. To track these changes, scientists rely on two main types of information. The first comes from weather stations on the ground, where thermometers record the actual air temperature at specific locations. While these readings are direct and accurate, they are often scattered, with gaps in time and space, making it difficult to see the full picture of a region's climate. The second type of information comes from computer models known as reanalysis datasets. These are not simple forecasts but rather a reconstruction of the past weather. Scientists feed decades of observations from satellites, balloons, and weather stations into powerful computer models that simulate the atmosphere. The result is a complete, continuous map of the weather, filling in the blanks where no station exists. However, because these maps are built from models, they must be checked against the real ground data to ensure they are telling the truth. This is especially important in places like Bangladesh, a country where the climate is complex and the stakes for accurate data are high.

In a recent study, researchers set out to test two of the most popular global weather maps, known as ERA5 and ERA5-Land, to see how well they capture the daily heat and cold across Bangladesh. The team compared the computer-generated temperatures from these two systems against records from thirty-one actual weather stations scattered throughout the country, covering the years from 1989 to 2019. They looked at both the hottest part of the day and the coldest part of the night, as well as how well the systems tracked extreme weather events, such as the hottest days of the year or the frequency of unusually warm nights. The goal was to determine which of the two digital maps provides a more reliable picture of the real world for this specific region.

The investigation revealed that both digital maps are generally very good at matching the real weather. When the researchers looked at the daily maximum temperature—the highest heat reached during the day—both systems showed a strong connection to the ground observations. However, a closer look showed that the standard ERA5 map performed slightly better than its more detailed cousin, ERA5-Land. The standard map tracked the daily highs with greater accuracy, showing less error in its predictions. Both systems tended to underestimate the heat, meaning they predicted days to be slightly cooler than they actually were, but the standard map was closer to the truth. This difference was not just a small fluctuation; statistical tests confirmed that the standard map's superior performance was a consistent and significant finding across the country.

The story changed slightly when the researchers examined the daily minimum temperature, which represents the coldest point of the night. Here, the two maps were much more evenly matched. Both systems did a good job of tracking the overnight lows, though they both had a tendency to predict nights that were slightly warmer than reality. Interestingly, the more detailed ERA-Land map actually had a slight edge in reducing this warm bias, making it a bit more accurate for nighttime temperatures in some seasons. The study also found that both systems struggled the most during the monsoon season. When heavy rains and thick cloud cover dominate the sky, the ability of the computer models to predict day-to-day temperature changes dropped noticeably, particularly for the nighttime lows. This suggests that the complex interactions between the wet land and the atmosphere during the rainy season are still difficult for even the most advanced models to capture perfectly.

Beyond daily temperatures, the team also checked how well these maps recorded extreme weather events. They looked at indices that measure the intensity of the hottest and coldest days, as well as how often extreme heat or cold occurs. For the absolute hottest and coldest days, the standard ERA5 map again showed a slight advantage, capturing the severity of these extremes more accurately. However, when it came to tracking long-term trends in how often extreme heat or cold happens, the more detailed ERA-Land map performed better. It was more successful at matching the observed changes in the frequency of these events over the thirty-year period. This suggests that while the standard map is better at getting the specific temperature right on a given day, the higher-resolution map is better at showing how the pattern of extreme weather is shifting over time.

The researchers concluded that for most general purposes in Bangladesh, the standard ERA5 map is the most reliable choice for daily temperature data. It provides a solid, consistent record of both daily highs and the severity of extreme heat. However, if a researcher needs to study how the frequency of extreme weather is changing over time, or if they need the most accurate picture of nighttime temperatures, the ERA-Land map offers distinct advantages. The study highlights that while these computer models are powerful tools that can fill in the gaps where weather stations are missing, they are not perfect. They work best when their specific strengths are understood and matched to the right task. For scientists and planners in Bangladesh, knowing which map to use for which job is now a clearer path forward, ensuring that decisions based on climate data are as accurate as possible.