Knowing climate for the next weeks and months: Which information can we trust?

Marta Terrado
IMPREX partners

Predictions on how the climate is going to be in the next weeks, months or seasons are available nowadays for Europe. They provide information that can be used to calculate the probability of experiencing water-related natural hazard events, such as floods or droughts.

However, any probability of occurrence of particular climate phenomena comes together with a probability of non-occurrence and decision-makers need to know if the available information is reliable enough to make a particular decision.


We continuously take decisions based on the weather. The weather forecast for tomorrow helps us decide if we should take the umbrella before going out, or whether we stick to our barbecue plan or book a table in a restaurant instead.

In the same way, forecasts for the next months can be a valuable piece of information on which our future plans can be based. Will this summer be rainier than usual, or will we be able to enjoy our vacation in the North Sea?

Climate information becomes even more valuable when it crosses the personal sphere and affects the public and private sectors. Water managers, for instance, need to know if it will be possible to satisfy water needs for next summer based on the water reserves that are expected to be available or if, conversely, restriction measures will need to be put in place.

Climate predictions are useful to answer these types of questions. However, they have an intrinsic associated uncertainty. Due to this uncertainty, average past data (climatology) has been traditionally used by decision-makers. For instance, to deduce the likelihood of having more, the same amount or less than average amount of rain next summer in Southern Spain, decision-makers have been looking at the average precipitation of the last 30 summers in the region of interest. Nevertheless, climatology- based approaches have some limitations since they assume that future conditions will be similar to past conditions, they cannot predict extreme events that have never happened before and they neglect atmospheric dynamics such as those caused by climate change.

To understand how reliable climate predictions are compared to current practice, predictions obtained by running models in the past are compared with observations; that is, with real measurements of what actually happened. Climatology is also compared with observations. These comparisons allow assessing whether predictions or climatology provide more accurate results. Different quality scores can be calculated to indicate how much better predictions are compared to using climatology.

Scores above zero indicate that using the prediction has an added value compared to using climatology, meaning that better decisions can be made than with the current approach. The minimum level of forecast quality required in each situation will depend on the decision to be made, the vulnerability of the sector, or the variability of a particular variable.


IMPREX research to improve the quality of climate predictions

Within IMPREX, the Barcelona Supercomputing Center (BSC) is quantifying the quality of sub-seasonal and seasonal climate predictions that are available for Europe and exploring ways to increase their quality through multi-model systems or increased model resolution. Both the sub-seasonal and seasonal time scales have been considered because they provide different types of actionable information that can enable decision-making according to the particular information requirements of each user.

Various prediction systems that can be used for hydrological purposes are assessed (systems constituting the EUROSIP
(1) multi-model system at seasonal scale and various systems belonging to the Subseasonal-to- Seasonal (S2S) project (2) at sub-seasonal scale).

The reason why different prediction systems are taken into account is because the quality of the predictions issued by each system differs according to the type of variable (e.g. temperature, wind or precipitation), the prediction period (e.g. summer or winter) and the location (e.g. Northern or Southern Europe). Whereas some prediction systems have higher quality scores for winter precipitation in Northern Europe, others can have lower scores in this area but alternatively provide higher quality scores for Eastern Europe instead. The key lies in being able to find windows of opportunity; that is, to identify which is the prediction system that decision-makers, who often are not climate experts, should use when making a particular decision. As said, this will depend on their variable, time period and location of interest.

Sometimes a single forecast system might not provide the best quality prediction. A potential way to obtain a forecast with increased quality is by combining the forecasts of several prediction systems. In this sense, BSC is also working to find the best way to combine the predictions issued by single prediction systems in order to obtain higher quality predictions that water managers are more willing to incorporate in their decision-making.


quality scores

Multi-model map of quality scores over various geographical regions in Europe between 1992-2012. Different colors indicate areas where a particular prediction system performs better. Dots mark areas where the skill is significant at 95% confidence level. Source: Mishra (2016).

To illustrate how higher quality predictions for Europe can benefit the water sector, IMPREX makes use of different case study narratives dealing with problems or activities that could be anticipated or better planned if proper and reliable climate information was available in advance. The narratives encompass flood inundation prediction (e.g. should population be evacuated?), planning of hydropower generation and fluvial transport (e.g. which is the best period for energy generation and transport?), and water allocation for urban and irrigation purposes (e.g. how much water should be prioritized for drinking? How many hectares of irrigated crops can be planted?), among others.


(1) Global seasonal forecasting system version 5 (Glosea5) from Met Office, System 4 from the European Center for Medium-Range Weather Forecasts (ECMWF), System 2 from National Centers for Environmental Prediction (NCEP) and System 5 from Météo France

(2) Monthly forecasting system from ECMWF, Global ensemble forecast system for monthly and seasonal predictions CFSv2 from NCEP, Climate Prediction System version1 from Beijing Climate Center (BCC) and Global ensemble prediction system  from Environment and Climate Change Canada (ECCC).


The first narratives can be read on the IMPREX website:

IMPREX website

Related documents:

  • N. Mishra. Skill assessment of seasonal temperature and precipitation forecast over Europe. Master in data science, Barcelona Graduate School of Economics and Barcelona Supercomputing Center, 2016.
  • C. Prodhomme, Batté, L., Massonnet, F., Davini, P., Bellprat, O., Guemas, V., Doblas-Reyes, F. J. (2016). Benefits of increasing the model resolution for the seasonal forecast quality in EC-Earth. Journal of Climate, JCLI-D-16-0117.1.
  • M. Turco, Ceglar, A., Prodhomme, C., Soret, A., Toreti, A, Doblas-Reyes, F.J. (2017) Summer drought predictability over Europe: empirical versus dynamical forecasts. Environmental Research Letters 12, 084006.

Leave a comment

Plain text

  • No HTML tags allowed.
  • Lines and paragraphs break automatically.
  • Web page addresses and email addresses turn into links automatically.
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.