New normal period in EMPS is 1986-2015

Dec. 10, 2020Christopher Rindahl

Wattsight is happy to announce that we are updating all our hydrological normals and our EMPS model with new climatic years. Previously our hydrological normals were using the 30 year period 1980-2010, while the EMPS model was using 45 climatic years covering the weather scenarios from 1967-2011.

During the next week we will change all these models to use the same 30 year reference period, using the climatic data from 1986-2015. As a consequence we will update all of the normals and forecast models for the following fundamentals:

  • Precipitation
  • Inflow
  • Reservoir
  • Snow and groundwater
  • Hydro production
  • HBV models
  • EMPS price models
  • EMPS details

A separate blog will be published covering the hydrological data, here we will focus on the methods used in the EMPS model as a basis for most of these models.


Instead of using statistical data for the last 30 years we use EMPS to simulate the new inflow statistics, reservoir statistics and hydro production. The reason for doing so, is because of the massive changes in the power system over the last 10 years. A wet year now, may also be followed by high wind, which makes it far more difficult to achieve the same hydro production as statistics would show for i.e 2006. On the other hand, far more interconnectors are now available, increasing the flexibility for hydro producers.

To properly take this into account, we have simultated in the EMPS how hydro producers should ideally operate on the 2020 power system, given that the different weather scenarios from 1986-2015 should occur again.

We use coherent data for consumption, temperature, wind and inflow, all adapted to 2020 capacities. So what would the inflow, the reservoir fillings and the production be, if the same weather that occured in 1986 should happen again, with current wind capacities, consumption levels and hydro production units?

Inflow statistics for Norway

The basis for the new normal period is based on updated data from NVE (The Norwegian Water Resources and Energy Directorate). The largest increase of new hydro production units has happened in Norway. Updated station data is obtained from NVE, where we get a dataset from 2019 showing all hydro production units and dams, with their capacities and efficiencies. In combination with this we receive about 80 waterflow series, which all the hydro production units are linked to. These data range all the way back to 1950, but we only use the data from 1986-2015 now to calibrate our inflow.

The reason for the new and shorter time period is the climate changes that are taking place. With the major changes seen in the last decades, it seems clear that recent data are more representative than the old ones. That is also the recommendation from NVE to use a shorter time period to set the normal levels now. Since the normal period used by NVE currently is 1980-2010, we have ran our models with 35 years to calibrate our models on the common period 1980-2010. Hence our levels are fully calibrated towards the normal levels reported by NVE for the period 1980-2010. But we use the newer normal period 1986-2015 which is approximately 1 TWh wetter for Norway due to some wetter years in the latest statistics.

Inflow statistics for Sweden

For Sweden we don’t have access to the same quality of metadata for all the Swedish stations, but we still have a very good Swedish model with a large amount of the older stations. And the amount of new builds and improvements have been modest in Sweden. These are solved with a couple of large and fairly unregulated dummy stations in Sweden, which is more than sufficient to represent the Swedish system and their new small scale hydro power which is mainly run of river production. In this update we have introduced several new water marks in Sweden to better represent the inflow profiles for all the stations. We have spent a long time finding the stations that best represent the inflow for each area, and changed quite a few of the old outdated watermarks. So our inflow statistics for Sweden should be better than ever.

New normal levels for EMPS

We have updated the reservoir capacities to more or less match reservoir capacities from nordpoolspot. For Norway we now have reservoir capacity of 86.890 TWh while for Sweden the reservoir capacity is 33.827 TWh.

Our inflow normals and corresponding production levels are now 136.2 TWh for Norway and 66.5 TWh in Sweden

Note that the numbers we mention here are pure EMPS results, that could have changed slightly when we make hourly profiles and adapt the data to a proper calendar function. So the other blog about hydrological numbers could have slightly different numbers. The EMPS model we have only operate with 52 weeks each year, hence it will not handle i.e week 53, which we will do with all the data you may download from wapi.


In this graph we show the comparison for Norway Total between NVE statistical average for the periode 1999-2018 (20 years) and our simulated normal for the period 1986-2015. Quite close to eachother and gives us certainty that our modelling work has been of high quality.

Change in consumption

As we have changed the normal period and see large changes to several of our inputs, we have also used this transition to change two other important inputs in the EMPS model that needed some brushing up.

The first one is the level of the residential temperature corrected consumption. The temperature corrected consumption provided by official sources is clearely off, most likely using quite old temperature statistics that are not representative for the climate we are in now. We have seen this giving to high temp adjusted residental consumption over time, and have decided to increase our climate adjustments quite a bit, lowering our general consumption levels.

Our previous climate adjustments for NP total was 3.3 TWh. This has now been increased by 2.9 TWh and we now have 6.2 TWh in consumption reduction due to warmer climate.

Sweden has the major part of the change with 2.4 TWh. Current climate adjustments are 3.7 TWh in Sweden and 2.5 TWh in Norway.

Change in CHP

In line with the assumptions and changes we have done to CHP in our LTP reports, we have also adjusted our CHP production forecasts in the regular medium term simulations. One effect is the warmer climate, with warmer weather, CHP as a pure by-product will be reduced as the heat demand is reduced.

The sceond effect is our findings that more units are changing from CHP factories to simple Heat Plants, they will just aim at delivering heat as a product. Hence they will have no generators, and to increase profits on the heat production, they are installing heatpumps with appr. 4X efficiency. As a consequece, this pulls consumption a bit upwards, and the CHP production is drastically reduced. The process is already ongoing and expected to increase further.

Our changes in CHP levels are:

Denmark -2TWh

Sweden -5 TWh

Finland -3 TWh

Price effect in simulation



The net effect on the prices are quite visible for the front year, dropping 1.7€. For the front year, the southern parts of Norway and Sweden suffer the largest price reduction.

Notice from the percentiles, screen shot 2, that the highest percentiles have the largest price reduction. This is probably because some cold and dry years have been removed from the climatic years, reducing the upside potential.

As we go forward in time, the effect on the years are more limited. Wetter and milder years push prices down, but the reduction in CHP production increase them. As a net effect for the last couple of years in the simulation, price volatility decreases, the lowest prices rise 3€ while the highest go down 3€. During this week our latest EMPS simulation based on the new climate statistics will be published.