Case studies

Eight case studies have been prepared in the project S2S4E to demonstrate the potential of the forecasts used in the Decision Support Tool (DST). The case studies analyze extreme events that occured in the past and present S2S4E forecasts that could have been available, to see if the forecasts could have added value to decision-making for the energy sector in preparation for these extreme events. The eight historical case studies have been selected as most relevant by industrial partners, due to unusual climate behaviour affecting the energy market.

The following sections provide short summaries of these case studies. For more detailed information on the climatic conditions and the forecasts analyzed in these case studies, please refer to:

Case Study #5 Factsheet: Icing Event in Romania

Case Study #6 Factsheet: Wind drought in USA

Case Study #7 Factsheet: The “Beast of the East“ Cold Spell in Europe

Deliverable 4.1: Additional factsheets for all the case studies listed below are coming soon. Meanwhile, detailed information is available in Chapter 6 of D4.1

Factsheets Guide: A guide to help understand the factsheets presented

 

Basic facts

Focus of the case study: electricity demand and wind power

Forecast range: sub-seasonal

Forecast variables: wind speed, temperature and electricity demand

What happened

A cold wave over Europe led to extremely low temperatures, which increased electricity demand for heating. Lower than usual wind speeds also resulted in a decrease in wind power generation and caused a high risk of energy imbalance in the energy grid.

Where it affected

The cold spell mostly affected areas in Europe. France in particular faced a shortage in energy supply due to planned maintenance outages in several nuclear power plants that coincided at the same time as  the cold spell.

When it occurred

The anomaly was observed during winter 2016-17, and was particularly significant from January 17th through 23rd, 2017.

Conclusions and potential of S2S4E forecasts

The fRPSS values computed for the whole hindcast demonstrate that S2S4E sub-seasonal forecasts have positive skill for this time period and region in anticipating episodes of low temperatures, especially two to one week before the event. According to these skills scores, S2S4E temperature forecasts demonstrate that they could have added insight to the decision-making process for industries and sectors who were affected by this cold spell, especially  one week before the event. On the other hand, wind forecasts also had increasingly high positive skill two and one week before the event, however the forecasts were not able to capture this specific cold spell. This means that the wind forecasts are still valuable, but were not able to predict the correct below normal tercile for the region and period in study.

In this cold spell case study, the forecasts were able to predict the below normal tercile for temperature and the above normal tercile for electricity demand two to one week before the event. Wind forecasts did not predict the below normal tercile, although the probability for this tercile was higher one week before the event.

For more detailed information on this case study, please refer to:

Deliverable 4.1

 

Basic facts

Focus of this case study: electricity demand, solar, hydro and wind

Forecast range: seasonal

Forecast variables: precipitation, inflows, solar radiation, photovoltaic capacity factor, wind speed, temperature, electricity demand

What happened

A sudden heat wave occurred. High solar radiation led to a moderate increase in solar power production. However, low wind speeds reduced wind power substantially and caused an imbalance in the energy system. In addition, backup coal power plants, relying on river transport, were compromised by low navigability in the Rhine and Neckar rivers due to low precipitation.

Where it affected

The anomaly occurred in Germany, affecting the assets (conventional power plants) in the Rhine and Neckar river banks.

When it occurred

Increased solar radiation and low precipitation occurred in the middle weeks of July 2013. High temperatures and low wind speeds were registered during the second half of July.

Conclusions and potential of S2S4E forecasts

The fRPSS values computed for the whole hindcast demonstrate that S2S4E seasonal forecasts have positive skill for this time period and region in anticipating river inflows at all lead times, especially two and one month before the event. Precipitation forecasts also demonstrated high positive skill at the one month lead time. According to these skills scores, S2S4E inflows and precipitation forecasts potentially could add insight to the decision-making for this context. On the other hand, solar radiation, photovoltaic capacity factor, wind, temperature and electricity demand forecasts and negative skill (fRPSS) for the region and time period studied, and therefore provide limited value to decision-making.

For this heat wave and low precipitation event, the forecasts predicted the correct tercile that the event occurred for inflows, solar radiation and photovoltaic capacity factor, at all lead times. Of these forecasts, the probabilities predicted were highest for inflows.

For more detailed information on this case study, please refer to:

Deliverable 4.1

 

Basic facts

Focus of this case study: electricity demand and wind

Forecast range: sub-seasonal

Forecast variables: wind speed, temperature, electricity demand and demand net wind

What happened

Above normal temperatures and low wind speeds led to increased electricity demand and low wind power generation.

Where it affected

The temperature and wind speed anomalies impacted central and south Europe. Spain and Portugal were the most affected countries.

When it occurred

The anomalies were observed during the end of summer 2016, notably from August 30th to September 5th.

Conclusions and potential of S2S4E forecasts

The fRPSS values computed for the whole hindcast demonstrate that S2S4E sub-seasonal forecasts have high positive skill for this time period and region in anticipating episodes of high temperatures, at all lead times. According to these skills scores, S2S4E temperature forecasts demonstrate that they could have added insight to the decision-making process for this heat wave. For the other forecasts, fRPSS scores were slightly positive for wind speed at all lead times, and fairly positive for electricity demand and demand for net wind only one week before the event.

For this heat wave, the forecasts were able to predict the correct terciles for temperature, electricity demand and demand of net wind, at all lead times. For these forecasts, the probabilities for the correct tercile was especially high for the two and one week lead times. The forecast for wind speed also predicted the correct below normal tericle two and one week before the event.

For more detailed information on this case study, please refer to:

Deliverable 4.1

 

Basic facts

Focus of this case study: hydro power

Forecast range: seasonal

Forecast variables: precipitation, snow water equivalent, and inflows

What happened

Substantial snowmelt and above normal precipitation filled up the reservoirs too early and led to unproductive water release.

Where it affected

The anomaly was registered North of Europe, mainly: Sweden, Norway and Finland. The study area is  the Umeälven river basin.

When it occurred

The abnormal events taking part in this case study occurred from spring to summer 2015, with its peak in July.

Conclusions and potential of S2S4E forecasts

The fRPSS values computed for the whole hindcast demonstrate that S2S4E seasonal forecasts have very high positive skill for this time period and region in anticipating snow water equivalent river inflows at all lead times. These forecasts are therefore considered to be very robust for decision making for this context. Precipitation forecasts also demonstrated fairly positive skill at the two and one month lead time.

For this high snowmelt and rain event in Sweden, the forecasts were able to predict (with high probability) the correct above normal tercile for precipitation and river inflows and the below normal tercile for snow water equivalent.

For more detailed information on this case study, please refer to:

Deliverable 4.1

 

Basic facts

Focus of this case study: wind energy

Forecast range: sub-seasonal

Forecast variables: average temperature and minimum temperature

What happened

An unexpected cold spell resulted in extremely low temperatures, freezing the rotors of wind turbines and stopping power production in several wind farms.

Where it affected

The anomaly was observed over eastern Europe, by the shores of the Black Sea. Romania was one of the countries highly affected.

When it occurred

The cold spell was registered during mid winter in 2014, from January 28th to February 3rd, 2014.

Conclusions and potential of S2S4E forecasts

Ice formation can heavily damage wind turbines and is therefore a critical issue to study for the wind industry. For the time period and region analyzed in this case study, skill scores were positive for both minimum and average temperature S2S4E forecasts. The fRPSS values were significantly higher for minimum temperature forecasts, demonstrating that these forecasts have greater potential value for decision-making than average temperature forecasts.

Although fRPSS values were positive, average temperature and minimum temperature forecasts had difficulty predicting the correct below normal tercile for this icing event in Romania. The forecasts, did, however, predict greater probabilities for the extreme below normal conditions that occurred, especially at shorter lead times.

Another factor to consider in this case study, is that the ECMWF monthly system model used for the forecasts was a previous version (CY4OR1). This choice was made because the case study occurs in 2014 and the newer model system was not available then. However, if the newer ECMWF monthly versions had been used, better forecasts could have potentially resulted.

For more detailed information on this case study, please refer to:

Deliverable 4.1

Factsheet #5

 

Basic facts

Focus of this case study: wind energy

Forecast range: seasonal

Forecast variables: wind speed and wind power capacity factor

What happened

Wind speeds were substantially below normal, reducing wind power generation. This reduction caused negative financial implications for wind farm owners in the western part of the country.

Where it affected

The wind drought especially affected states in the central U.S. and the west coast, such as Texas, Oklahoma, and Kansas, where the biggest wind farms are concentrated.

When it occurred

The most significant part of the wind drought occurred January through March 2015.

What caused it

The high pressure and low winds conditions over North America were caused by a High North Pacific Mode status with a positive SST anomaly in the Western Tropical Pacific.

Conclusions and potential of S2S4E forecasts

Episodes of prolonged low wind speed (also known as wind droughts) can negatively affect the wind power industry. The fRPSS values of the S2S4E forecasts demonstrate high positive skill for the region and time period analysed in this case study. This means that S2S4E forecasts are better than climatology and have potential to anticipate wind droughts a few months in advance.

For the 2015 wind drought studied, the S2S4E forecasts moderately predicted the correct below normal tercile for wind speeds, while the capacity factor forecasts predicted the correct below normal tercile only at the three month lead time.

According to the climate drivers assessment, this region is sensitive to the impact of teleconnections arising in the tropical Pacific, such as ENSO or NPM.

For more detailed information on this case study, please refer to:

Deliverable 4.1

Factsheet #6

 

Basic facts

Focus of this case study: electricity demand

Forecast range: Sub-seasonal

Forecast variables: temperature and electricity demand

What happened

An unanticipated cold spell resulted in below average weekly temperatures and triggered an increase in power demand for heating.

Where it affected

The cold spell affected mostly central Europe.

When it occurred

This case study analyzes the extreme cold temperatures that occurred  from February 27th to March 5th 2018.

Conclusions and potential of S2S4E forecasts

The fairRPSS values computed for the whole hindcast demonstrate that S2S4E sub-seasonal forecasts have skill for this time period and region to anticipate episodes of low temperature, especially two to one weeks in advance. This high skill highlights how sub-seasonal forecasts could be beneficial over the current practice of using climatological forecasts for decision-making with energy trading companies and within Distribution System Operators.

For this 2018 cold spell case study, the forecasts did correctly predict temperature in the below normal tercile and energy demand in the above normal tercile. For both forecasts, the probability of the correct tercile was very close to 100% one week before the event.

For more detailed information on this case study, please refer to:

Deliverable 4.1

Factsheet #7

 

Basic facts

Focus of this case study: wind energy

Forecast range: seasonal

Forecast variables: wind speed, capacity factor, temperature

What happened

Persistent conditions of high wind speed and precipitation led to record-breaking power production from renewable sources (wind and hydro). A subsequent cold spell increased power demand.

Where it affected

The cold spell affected most of Europe. Increased wind speed and precipitation was observed mostly in Spain, Portugal and Greece.

When it occurred

The anomaly was registered at the end of winter season, mostly during March 2018.

Conclusions and potential of S2S4E forecasts

The fRPSS values computed for the whole hindcast demonstrate that S2S4E seasonal forecasts have negative skill for this time period and region in anticipating wind speed, capacity factor and temperature at all lead times. Therefore, these forecasts have limited value for decision making for this context.

For this cold spell and high wind speeds event in Spain, the forecasts were able to predict the correct tercile for wind speed and temperature.

For more detailed information on this case study, please refer to:

Deliverable 4.1