REviewing Climate change simulations for enhanced Adaptation in Sectors and Technical infrastructure

Transcription

1 RECAST REviewing Climate change simulations for enhanced Adaptation in Sectors and Technical infrastructure Adriaan Perrels (FMI) FICCA Final Seminar Helsinki 2&

2 Core Team Helsinki University: Jouni Räisänen, Olle Räty, Jussi Ylhäisi Finnish Meteorological Institute FMI Adriaan Perrels, Väinö Nurmi, Hanna Virta, Athanasios Votsis Technical Research Centre of Finland VTT Lasse Makkonen, Maria Tikanmäki, Irmeli Wahlgren, Erkka Rinne Links to / cooperation with: HENVI ENSURE project EU COST VALUE FP7 TOPDAD Adriaan Perrels/IL 2

3 RECAST Objectives & Structure Main: assess the impacts of climate change induced uncertainty and growth of weather variability on the social and economic coping range of selected infrastructure and real estate, with the aim to generate insights, methods and procedures that enable better adaptation Adriaan Perrels/FMI 3

4 Output 9 published peer reviewed articles and 2 submitted Several under preparation 1 PhD granted; 3 foreseen (2016) Conclusion of debate on plotting position New method for extreme value estimation Elaboration of cross-validation framework for climate model output corrections Application of cross-validation framework on hydrological modeling covering all Nordic countries Creation of large real estate urban features ecosystem services GIS database (Greater Helsinki, Pori, Rovaniemi) Modelling real estate value formation based on diverse characteristics Testing the effectiveness of flood risk mapping and the responsiveness of housing markets to natural hazards Adriaan Perrels/FMI 4

5 Two main streams 1 Methods Cross-validation for bias correction Applications Variability in precipitation & hydro-reservoir filling rates 2 Extreme value estimation in a non-stationary climate Response to (disclosed) flood risks Adriaan Perrels/IL 5

6 1 Which projection method is best? Principle of cross validation Recent climate in model N Recent climate in N-1 models Actual future climate in model N Projection for future climate in model N Verification statistics Future climate in N-1 models Repeat for all choices of Model N, using different (delta change and bias correction) methods for projection Räisänen and Räty Adriaan Perrels/FMI 6

7 Conclusions on correction methods Quantile mapping type bias correction approaches have best average performance in cross-validation No bias correction / delta change method is best under all circumstances Projection method uncertainty is generally smaller than climate model uncertainty Projection method uncertainty tends to be largest for extremes of temperature and precipitation Adriaan Perrels/IL 7

8 Example: First percentile of daily mean T in January, Helsinki, (SRES A1B) Adriaan Perrels/FMI 8

9 Precipitation and discharge variability in Nordic area - use of HYPE model Issue: variability of precipitation and filling rates of hydro power reservoirs in relation to climate change Prospects for operation and capital productivity favorable or unfavorable? HYPE model (by SMHI) , new version in 2015 Semi-distributed, process-based model Primary input: daily precipitation and temperature Baseline: ERA interim reanalysis data from ECMWF 1 Climate scenarios: 6 regional climate models from the ENSEMBLES project Correction methods: 1. Mean correction by delta change method 2. Quantile by quantile correction by delta change method 3. Quantile by quantile correction by bias correction method Emission scenario: A1B Adriaan Perrels/FMI 9

10 Changes in discharge by sub-basin change in averages // Indicated ranges are approximate Accurate modelling of snow melt and evapotranspiration difficult Challenge of parameter correction in changing climate -4% ~ -25% Overall average change: -0.6% 0% ~ +25% Adriaan Perrels/IL 10

11 Changes in discharge variability by sub-basin change in st. deviations // Indicated ranges are approximate Accurate modelling of snow melt and evapotranspiration difficult Challenge of parameter correction in changing climate - ~ -33% Overall change in standard deviations: -18% - 33% ~ 0% ~ 0% ~ +25% Adriaan Perrels/IL 11

12 Sensitivity of discharge to changes in precipitation (P) and temperature (T) // (one of 6 models) Temperature is really important Mean correction by delta change method Quantile by quantile correction by delta change method Adriaan Perrels/IL 12

13 Lessons on variability and climate change Most of Norway and to some extent Lapland will generally experience an increase of annual river discharges Most of Finland (outside Lapland) and of Central Sweden may generally expect very moderate change (+ or -) in annual discharge Southern Sweden and Denmark have generally a prospect of decreasing annual discharges The above conclusions are expected to get somewhat corrected (upwards) when a new version of HYPE is applied Intra-annual variability in discharges is expected to decrease in most parts of the study area, except in southern Sweden and to some extent in some areas of southern Finland (depends on regulation) The above (preliminary) results suggest that climate change is easing operational conditions for most hydro power capacity and raising its capital productivity Other changes such as growth in wind power capacity can affect these easing effects (+ & -) this is outside the realm of this study Adriaan Perrels/FMI 13

14 2 New objective method of fitting a cumulative distribution function (CDF) to extreme value data Selection of number of extremes is based on the fitted CDF No pre-conditioning because a curve is fitted Plotting positions are based on the true probability The weights of the data points are based on the inverse of the variance of the measured variable The parameters of the General Extreme Value distribution (GEV) are found by iteration Minimizing the error of the fit is done by the least squares with respect to the measured variable For the fitted GEV the weights correspond to the theoretical variance and the GEV parameters correspond to the weights Adriaan Perrels/FMI 14 Flickr Jacob Whiteboy - Storm cell #9,

15 Example: Wet snow load in Münsterland, Germany Adriaan Perrels/IL 15

16 Ecosystem benefits & hazards and real estate prices Urban ecosystem services provide both benefits and risks for real estate owners Real estate prices incorporate the benefits much easier than the risks Price premiums for ecosystems services are relatively sensitive to economic cycles Waterfront locations exhibit very high premiums Spatial price premium spillovers Regardless of (non disclosed) flood risks House buyers (and even sellers) are poorly aware of flood risks for aspired houses flood maps seem to correct this quite effectively Analysis based on large real estate urban features ecosystem services GIS database Adriaan Perrels/FMI 16

17 Price effects of various housing features based on spatial hedonic regression (Votsis 2014) Adriaan Perrels/FMI 17

18 Effect of flood risk disclosure housing price Identified price response based on HPE for Greater Helsinki Votsis and Perrels (submitted) Calculated expected value of flooding cost based on Finnish unit-cost estimates and flood map information Adriaan Perrels/FMI 18

19 Lessons on extreme events w.r.t. markets Rare events with economic consequences are usually poorly incorporated into market mechanisms Public disclosure of (flood)risks for real estate properties seems to be taken up fairly accurately by housing markets hence flood risk maps are an effective instrument to correct real estate markets and endogenize adaptation, but. for often more subtle climate change induced changes in risks flood risk maps may be less effective (though still a necessary ingredient) It seems that disclosed flood risks are fairly accurately incorporated into prices when occurrence probabilities are not very low (i.e. over ~ 2%/year according to this study) It is unclear whether this is attributable to lack of willingness or capability to assess costs with very low occurrence probability or that additional information would extend the accuracy ranges The information effect of the flood maps seems to last, but long term information decay effects could not be tested Complementarity from sociologic & social-psychologic research Adriaan Perrels/FMI 19

20 Further explorations Updated HYPE model jointly with Nordic electricity model BALMOREL (VTT) Application of new EVA method on localized flood risks and relating it to real estate market responsiveness Reviewing chains of uncertainty and relative significance of uncertainty contributions Urban simulation model incl. ecosystem services links with urban climate model Adriaan Perrels/FMI 20

21 References Makkonen, L., Pajari, M. & Tikanmäki, M., (2012), Closure to Problems in the extreme value analysis Structural Safety 40, * Räisänen, J. and Räty, O. (2012), Projections of daily mean temperature variability in 2 the future: cross-validation tests with ENSEMBLES 3 regional climate simulations, Climate Dynamics, DOI /s * Votsis, A. (2013), Modelling the effects of ecosystem services on the formation of urban real estate value: a spatial econometrics and geoinformatics approach, to be submitted April 2013 Votsis A., Perrels, A. 2014, Housing prices and the public disclosure of flood risk: a difference-in-differences analysis, submitted to the Journal of Real Estate Finance and Economics RECAST Posters in the seminar O. Räty, Th. Bossard, H. Virta, J. Räisänen, E. Rinne: Large-scale hydrological simulations with the HYPE model: uncertainties in future simulations and separate temperature and precipitation effects to simulated changes J. Räisänen and J.S. Ylhäisi: CO 2 induced climate change in Northern Europe in three generations of global climate models A. Votsis, A. Perrels: Climate Risks and Amenities in the Finnish Housing Stock V. Nurmi: Green Roofs in Helsinki Cost Benefit Analysis More information about RECAST can be found at the project website: Adriaan Perrels/FMI 21

22 Thank you Adriaan Perrels/IL 22