The KNMI'23 climate scenarios show that Dutch summers will become significantly drier in the future. Water managers have a need for information on the associated effects on hydrological variables such as groundwater levels and river discharges. This is important to anticipate on future freshwater availability. The Royal Netherlands Meteorological Institute (KNMI) has already made so-called transformed climate data available. In these datasets observed weather is converted into expectations for the future climate. However, these transformed weather data do not take into account the changing number of dry days in the future. Therefore, these data are not applicable to drought analyses.
In this study, HKV and KNMI, together with STOWA, applied an alternative methodology that describes the climate using statistics (see [1] for the full report). This correctly accounts for the expected increase in dry days. With this article, we want to encourage water managers to analyse the effects of meteorological drought on hydrological drought effects. This is vital in order to anticipate an increasingly drier future.
Deriving drought statistics
The aim of this study was to derive drought statistics from the KNMI'23 RACMO climate series. RACMO, the Regional Atmospheric Climate Model, is the regional climate model used by KNMI to calculate the future climate in the Netherlands. In practice, the RACMO series (240 years per scenario) are difficult to process within the widely used, computationally intensive hydrological models, e.g. groundwater models. For this reason, we have developed two practically applicable model products based on derived statistics:
Two research questions were central to the development of these products: (1) can different drought indicators be grouped in the same characteristic years? After all, meteorological drought, or dry weather, cannot be captured unambiguously and has multiple facets; and (2) how do drought statistics vary spatially across the Netherlands? With this approach, we aim to determine whether a single characteristic year can be derived for the entire Netherlands.
Principles
The focus of this study is on meteorological drought in the growing season: the period between April 1st and October 1st. For this growing season, recurrence times were derived for several drought indicators: the maximum cumulative precipitation deficit per year, the minimum precipitation sum over 45 days per year, the number of dry days per year and the minimum seven-day average river discharges of the Meuse and the Rhine per year. All these indicators describe only the summer half of the year. We express the recurrence times in, for example, once every 25 years. We derived such recurrence times for several KNMI'23 climate scenarios, considering both moderate and strong climate change, for the 'target years' 2050, 2100 and 2150.
The results show that the drought indicators do correlate with each other, but not strongly enough to select characteristic years for which the recurrence times are similar for all drought indicators. For example, a RACMO year may have a maximum cumulative precipitation deficit (MCPD) occurring once every 10 years, but a minimum 45-day precipitation sum with a recurrence time of 3 years and a minimum Rhine discharge with a recurrence time of 25 years. Therefore, in our products, we present all characteristic years based on the annual MCPD. This drought indicator is the one most commonly used by KNMI and water managers. Additionally, the derived statistics for the other indicators are available for reference through the mentioned model products.
Results
Figure 1 shows the annual maximum cumulative rainfall deficit for several climate scenarios. For reference, the chart also shows the MCPD in 2018. A dry summer like the one in 2018 has a recurrence time of over 20 years in today's climate. By 2050, that will shift to five to eight years, depending on the degree of climate change. In fact, for the year 2100, a dry year like 2018 may well become 'the new normal' for us. Statistics for the other included drought indicators also show that Dutch summers will become significantly drier in all climate scenarios. This change will have severe impacts on freshwater availability in the Netherlands.
Figure 1. Statistics for the MCPD, averaged over the Netherlands
The probability of dry weather occurrence varies spatially in the Netherlands. Figure 2 shows the spatial distribution in the annual maximum cumulative precipitation deficit for the current climate, at a recurrence time of 25 years. This is a situation with very dry weather conditions. The left panel shows the calculated results in the current climate based on the KNMI'23 scenarios. On the right, the same analysis is presented, but based on actual measurements. First of all, there are clear similarities between the two panels, supporting the validity of the KNMI'23 climate scenarios. In addition, clear spatial patterns are visible. The south of the Netherlands is clearly drier than the north. In addition, orographic effects (altitude effects) are visible, for example, on the Veluwe. According to the KNMI’23 scenarios, these spatial patterns will barely change in the future.
Figure 2. Spatial variation in MCPD occurring on average once every 25 years in the current climate. The variation based on this study is presented on the left, with the variation based on observations on the right
For regional water managers, it is important to take the spatial variability into account. An example: in a scenario with strong climate change in 2050 (scenario Hd2050), a maximum annual cumulative precipitation deficit of 350 mm in northern Groningen occurs once every 35 years on average. In Limburg, this occurs once every five years. That is a big difference. Therefore, the choice was made to provide both the representative 30-year RACMO series and the characteristic years per water board and not the average situation across the Netherlands.
Conclusions and recommendations
For many people, the fact that Dutch summers are getting drier will not come as a surprise. The results of this study provide insight into the effects of climate change on meteorological summer droughts. Both the moderate and strong climate change scenarios show a clear shift towards drier summer weather. Dry summers, such as the ones the Netherlands had in 2018 and 2022, will become more frequent and more intense in the future. These changes strongly affect groundwater levels, and river discharges. This will reduce, for example, the water availability for agriculture, nature, and shipping. All this calls for well-defined choices by water managers.
To facilitate the calculation of meteorological drought in hydrological models (including groundwater models), 30-year representative RACMO series and characteristic years were derived for each water board. These data and associated statistics are available on the STOWA website.
Guide
The study is summarised in a guide for the practical use of the drought statistics and their underlying data (see Figure 3). This guide aims to help water managers by applying the data in modelling studies for their own working area. In this manner, drought effects are accounted for in a clear and reproducible manner. The guide describes how water managers can select and apply characteristic years and representative series and how to deal with uncertainties in the scenarios. In this way, the guide forms a bridge between complex climate data and everyday water management practices.
Figure 3. Guide to using the KNMI'23 climate scenarios in drought studies
This study used the KNMI'23 climate scenarios to derive the recurrence times of several drought indicators for each water board in the Netherlands. This includes the analysis of the maximum cumulative precipitation deficit during the growing season. The results show that drought extremes are expected to become much more frequent in the future. A dry summer such as in 2018 occurs about once every 20 years in today's climate, but by 2050, it will occur every 5 to 8 years and possibly almost annually by 2100. These changes have a strong impact on groundwater levels and discharges. This requires our regional water managers to make well-defined choices.