The world’s leading publication for data science, AI, and ML professionals.

Are heavy rain events increasing in frequency and intensity in Germany?

An investigation using data science

Photo by Inge Maria on Unsplash
Photo by Inge Maria on Unsplash

On July 14th, 2021, the amount of precipitation exceeded 100 litres per m² in one day in parts of North Rhine-Westphalia and Rhineland-Palatinate. The consequences were more than 100 people losing their lives to the floods and billions of euros in material damage.

Talking about it with a friend, he shared his impression that there is an increase in heavy rain events in recent years. I didn’t know whether to agree or disagree. Is he right and they increase in frequency in Germany? What about trends in their intensity? Are some cities unfortunate in their location and suffer more frequent than average events?

Climate Change inducing rising temperatures is a generally recognized observation, but trends in heavy rain events have been a less present topic in German news media. Follow me on the journey to find answers to the above questions. We’ll gather data from 1,084 weather stations across Germany for the period 1900–2020. The result is a dataset containing more than 16.5 million weather observations. Let’s uncover its secrets.

Where’s our data coming from? The German Weather Service (Deutscher Wetterdienst) is responsible for collecting and archiving meteorological data in Germany. Most of the data is publicly available and can be accessed via the Climate Data Center portal. I downloaded historic precipitation data for all available weather stations in Germany, removed missing data and filtered the data set for the period 1900–2020.

Starting our analysis, we can plot the location of all weather stations to check if each region is covered.

Data basis: Deutscher Wetterdienst
Data basis: Deutscher Wetterdienst

The 1084 weather stations are distributed all over Germany. Each state is represented in the data set with at least several weather stations. Nevertheless, there is a slight gradient with more weather stations in the south-west of Germany than there are in the north-east.

Let’s take a look at when the data was recorded.

Data basis: Deutscher Wetterdienst, aggregated over individual values
Data basis: Deutscher Wetterdienst, aggregated over individual values

In the years following the end of WWII, there was a visible boom in meteorological observations in Germany. This discrepancy in observations is important to note and means that in future time-series analyses, we’ll have to adjust the count of heavy rain events by the number of observations each year.

When working with a dataset, it is always good practice to verify the credibility of your data. One way to do this is to externally verify the dataset with a known event. For this project, we can take a look at the Elbe/Donau flood in August 2002. If the dataset is accurate, it should show high amounts of precipitation on August 11, 2002, in the south-east of Germany that caused the Elbe and Donau to step over their shores.

Data basis: Deutscher Wetterdienst
Data basis: Deutscher Wetterdienst

Clearly visible are numerous Weather stations reporting high precipitation figures. The dataset dimensions time, location and amount of precipitation have therefore just proven their credibility.


Enough verification at once, let’s dive into the analytics part: Are heavy rain events becoming more frequent in Germany? To answer this question, we first need to understand what a heavy rain event is. The definition naturally varies by climate zone. Heavy rain in Germany might be considered a usual day if you live in the Brazilian rain forest.

The German Weather Service talks of an extreme event if:

  • The amount of precipitation exceeds 40 litres per m² in one hour OR
  • The amount of precipitation exceeds 60 litres per m² in six hours

Since the historic weather observations are only available in a daily aggregate, I cannot stick to the exact definition of the German Weather Service. With a slight adjustment to the definition and accepting minor inaccuracies, I will subsequently talk of a heavy rain event if the amount of precipitation exceeds 60 litres per m² per day.

That said, here’s the chart for the aggregated yearly amount of heavy rain events – adjusted by the total amount of observations for the respective year.

Data basis: Deutscher Wetterdienst, averaged over individual values
Data basis: Deutscher Wetterdienst, averaged over individual values

Is it what you expected? I was surprised when I first saw the graph, as I secretly expected to see a clear increase. But let’s look at it in detail.

Visible is a high variance across the years, highlighting the unpredictable nature of heavy rain events. Extreme spikes occur every couple of decades, marking decade- or century-level heavy rain. However, the 25-year moving average curve shows no significant change in the average number of heavy rain events over the last 100 years.

We have seen that the average number of heavy rain events has not increased in Germany so far, now let’s check if the same holds for their intensity.

Data basis: Deutscher Wetterdienst, averaged over individual values
Data basis: Deutscher Wetterdienst, averaged over individual values

Regarding the intensity of heavy rain events, the yearly variance is again high. The mean precipitation intensity of the average event remains constant at around 73 litres per m² throughout the last 100 years. Data recorded before 1950 should be taken with caution, as the number of observations was much smaller, which explains the extreme scatter of the values.

The period 2000–2020 shows higher and more frequent peaks in comparison to the 1950–1990 period, suggesting that heavy rainfall events have been more extreme in recent years when they have occurred. The future must be awaited to see if this observation continues and the mean corrects upwards.

Next time someone asks you if the number or intensity of heavy rain events is increasing, you can now say that while the number does not seem to have increased so far, their intensity has spiked upwards several times in the last 20 years.

However, please note, that all statistics have been calculated to best knowledge and belief, but no guarantee can be given as to their accuracy. You should also always check the work of climate researchers who have access to sophisticated simulation tools and possibly more data that may or may not lead them to different conclusions. But more on that in the discussion part.

Now, what about regional differences you may ask. Are events evenly distributed across Germany or are several states more prone to suffering from these phenomena? Let’s find out.

Data basis: Deutscher Wetterdienst, averaged over individual values
Data basis: Deutscher Wetterdienst, averaged over individual values

Heavy rain events are most likely to occur in Bavaria, Baden-Wuerttemberg, Saxony, and Saxony-Anhalt. Note that I calculated a weighted count that is independent of the number of weather stations each state operates. Let’s zoom in a bit more. This time to the level of weather stations.

Data basis: Deutscher Wetterdienst, averaged over individual values
Data basis: Deutscher Wetterdienst, averaged over individual values

Visible is an above-average count of heavy rain events at the southern border of Bavaria and in Baden-Wuerttemberg. One likely explanation for this is the proximity of these weather stations to mountain ranges – the Alps and the Black Forest. Otherwise, there appear to be only minor differences across Germany.

Concluding words and discussion of results

This blog post investigated trends in the occurrence and intensity of heavy rain events in Germany over the last 120 years. The Data comes from the German Weather Service (Deutscher Wetterdienst). Despite the high variance of annual events, there has been no significant increase in heavy rain events in Germany so far. While the average intensity has also remained constant over the last 100 years, there have been peaks in several years this century, and it remains to be seen whether this trend will continue.

From the perspective of physics, an air mass that increases in temperature by 1 degree can hold up to 7 per cent more humidity. This is one reason why scientists have long argued that global warming will likely also affect extreme weather events.

Studies from the United States, Great Britain and Uruguay support this thesis and stress a likely effect of anthropogenic global warming on extreme rain events in the respective analyzed region. However, two studies from Germany found no clear effects of human-caused climate change on heavy rain events, indicating that the natural variability in these events probably still plays a bigger role in Germany today.

Nevertheless, have the consequences of heavy rain events increased in Germany in the last 20 years. Progressive soil sealing means that less water can be absorbed by the ground, which exacerbates the flooding in the affected areas.

Should you wish to dive deeper into the topic, here is another cool map that allows you to explore the outcomes of 405 different studies on the effects of climate change on extreme weather.

If you’re interested in the code for this project, you can check it out on my Github.

Acknowledgements

This project would not have been possible without the freely accessible data of the German Weather Service (Deutscher Wetterdienst).


Related Articles