Scientific summary

Introduction

Climate change (CC) is expected to have a profound impact on the hydrological cycle. Increased temperatures (T) and potential evaporation as well as changes in precipitation (P) types and space-time distributions will modify the formation of runoff, with clear implications for society in terms of e.g. water resources, hydropower and flood risk.

Since ~15 years, “CC hydrology” has emerged as a distinct field of research with the overall aim of understanding and quantifying the future impacts. While this research has undoubtedly provided many important findings, at the time of designing HYDROIMPACTS 2.0 it was obvious that further progress required substantial development of various components involved.

  • Climate model inputs: Projections were generally not evaluated against a larger ensemble, to assess their magnitude of future change. Methods for bias correction and downscaling were developed only for P and T data with a daily time step or longer.
  • Hydrological impact assessment: This was generally performed for water quantity in single, medium-sized catchments. Very few examples of impacts (i) at small (e.g. urban) and large (e.g. continental) scales, (ii) also taking water quality into account, existed.
  • Support and outreach: There was a need for (i) open data and results from future hydrological climate change projections in Europe, (ii) tailored software for downscaling and visualization and (iii) pedagogical material for practitioners and the interested public.

In light of this, HYDROIMPACTS 2.0 was designed with the aims of developing the tools, performing the analyses and communicating the results required to make significant advancement in all the above aspects. The underlying hypothesis was that these aims were most effectively reached in a team combining expertise from hydrology, climate modelling and IT.

Material and methods

Climate model data: A large ensemble (~20) of RCM projections over Europe was produced including different GCMs, RCMs and IPCC emission scenarios and data with resolutions up to 30 min and 6 km9. P and T as well as several other variables have been used.   

Reference data: Concerning meteorological variables, high-resolution station data have been used as well as gridded daily products including PTHBV, ERA-40/Interim and GPCC. Further, large databases of discharge and water quality variables have been used.

Bias correction and downscaling: A range of methods has been developed and/or used, including the following.

  • Delta Change (DC): In the project, DC has been developed for high-resolution P by allowing for intensity-dependent changes as well as frequency adjustment15.
  • Weather-situation stochastic scheme: In a new approach, the weather situation (represented by cloud cover) is used to stochastically downscale extreme P from RCM grid scale to point scale11.
  • Seasonal atmospheric downscaling: Large-scale circulation patterns and teleconnection climate indices are linked to seasonal discharge volumes through novel combinations of advanced statistical tools including PCA, SVD and cross-wavelets8.   
  • Distribution-Based Scaling (DBS): In the project, the DBS method has been developed for (i) optimal application to P and T, (ii) supporting also wind speed and relative humidity20 and (iii) incorporating circulation patterns.

Hydrological and hydraulic models: The following (existing) models have been set up for various study areas in the project.

  • HYPE, HBV: Catchment-based hydrological models for integrated simulation of fluxes and turnover of water and nutrients, set up for Swedish catchments, entire Sweden, the Baltic Sea basin and Europe.
  • HYDRUS-1D: A numerical model for water and pollutant  transport in the unsaturated zone under different boundary conditions, set up for two representative soil types in southern Sweden.
  • SWMM, MIKE Urban, MOUSE, TSR: Dynamic models that describes catchment hydrology, hydraulic pipe flows and storm water quality in both single-event and continuous simulation. The models were set up for urban catchments in different Swedish cities.

Analyses: A very wide range of analyses have been performed in the project, mainly associated with the above methods and models.

  • Bias correction and downscaling: For example, evaluation of RCM bias for different variables and scales, distribution-fitting to observations and RCM data, analyses of links between large-scale circulation patterns and catchment discharge.
  • Evaluation of hydrological model results: Numerous hydrological variables have been analysed, characterising the runoff generation process (e.g. evapotranspiration, snow cover and soil moisture), the resulting water quantity (e.g. discharge, water levels and high/low flows) and quality (e.g. concentrations of nutrients, heavy metals and pollutants).

Besides these activities, other key analyses include evaluation of large RCM ensembles, studies of RCM historical simulations and reproduction of recent changes and extreme value analyses of high-resolution P.

Results

The project has generated a large amount of results and here only a very brief overview is given, see further the references given.

Short-term precipitation in RCM simulations: Through extreme value analysis, the expected future changes of short-term P extremes in Sweden have been quantified10,32. The impact of RCM spatial resolution on the reproduction of short-term P has been investigated12.

Bias correction and downscaling: The performance of the DBS method for correcting different variables in the RCM data has been demonstrated in various climate regions17,19,20. The performance of high-resolution DC and weather-type P downscaling has been demonstrated11,15. Links between large-scale circulation patterns and seasonal discharge have been identified8,33.

Small-scale hydrological impacts: The impact of P intensities and their future changes on storm water quality (suspended solids and heavy metals)5,6,7,21 as well as solute transport in soil16,30,35 has been assessed. Impacts of increased urbanisation are taken into account as well as different soil types. The impacts of CC on e.g. flood risk and overflows in different Swedish cities have been estimated13,31. 

Medium-scale hydrological impacts: CC impacts on inflows of water and nutrients to two major Swedish lakes (Vänern, Mälaren) have been assessed14,26.

Large-scale hydrological impacts: Through the large-scale set-ups of the HYPE model, CC impacts on hydrology in Sweden1,24,25, the Baltic basin2,4,29 and Europe25,28 have been assessed. A large number of hydrological variables have been analysed. Changes in nutrient transport and the effect of remedies and management plans have been assessed.

Outreach and support: Open access data on the estimated future hydrological changes have been made available on hypeweb.smhi.se. Various end-user interactions have taken place23. Recommendations for good use of RCM-P in hydrological CC impact assessment have been published22,34.

Publications: The project has so far generated 15 peer-reviewed published papers (+4 submitted), 2 books/theses, 13 reports/non-reviewed papers and 34 conference contributions. The reference numbers given above refer to the full publication list: Hydroimpacts2.0 Publication List.

A special issue of J. Water Manag. Res. was devoted to the project23,25,26,34,35.

Discussion and conclusions

We believe HYDROIMPACTS 2.0 has contributed to significantly advance the field of hydrological climate change impact assessment. The methods developed and the results obtained will be important in the process of designing policies and strategies for adapting hydrological systems to future changes. The emphasis on outreach and support ensured a widespread dissemination of project results as well as development of useful tools and pedagogical material for various end-users