Climate Perturbation Tool

Transcription

1 HYDRAULICS SECTION Kasteelpark Arenberg Leuven, Belgium tel fax bwk.kuleuven.be/hydr Climate Perturbation Tool Manual ir. Els Van Uytven Prof. dr. ir. Patrick Willems October 2016

2 The climate change perturbation tool KU Leuven can be used freely provided that the following source(s) is(are) acknowledged: Quantile perturbation method precipitation (relative wet day intensity changes & change wet day frequency) tailored hydrological impact scenarios for Belgium: Ntegeka V., Willems P., Roulin E. and Baguis P. (2014). Developing tailored climate change scenarios for hydrological impact assessments. Journal of Hydrology, 508, The climate scenarios for Belgium: Tabari, H., Taye, M.T. and Willems P. (2015). Water availability change in central Belgium for the late 21th century. Global and Planetary Change, 131, Quantile perturbation method precipitation (absolute & relative wet day intensity changes) The weather typing method: Willems P. and Vrac M. (2011). Statistical precipitation downscaling for small-scale hydrological impact investigations of climate change. Journal of Hydrology, 402,

3 Table of contents Symbols and abbreviations Introduction What does the program do? Temporal and spatial scales Provided time horizons Hydrological impact scenarios Belgium Limitations Procedure for running the program Structure Climate Perturbation Tool Run the climate perturbation tool Installation of R and RStudio Start the tool graphical interface Graphical interface: general choices to be made by the user Graphical interface: variables general Graphical interface: variable precipitation Graphical interface: variable ETo Graphical interface: variable temperature Graphical interface: variable wind speed Graphical interface: directory of the different folders Graphical interface: start Observations Climate model data Output Information about the perturbation request Appendix A: Information on perturbation request References

4 Symbols and abbreviations ETo Evapotranspiration GCM Global climate model MSLP [Pa] Mean sea level pressure P Rainfall Tmean Daily mean temperature Tmin Daily minimum temperature Tmax Daily maximum temperature WS Wind speed WT Weather type SYMBOLS AND ABBREVIATIONS 4

5 1 Introduction Within the scope of the CCI-HYDR project on Climate change impact on hydrological extremes in Belgium for the Belgian Science Policy Office (Belspo), that ended in 2010, a first version of this Climate Perturbation Tool was developed (Ntegeka & Willems, 2009). It was based on tailored climate scenarios for climate change impact analysis on hydrological extremes in Belgium (Ntegeka et al., 2014). Historical time series of precipitation, (mean, minimum and maximum) temperature, potential evapotranspiration (ETo) and wind speed are by the tool modified based on the climate scenarios following an advanced version of the quantile perturbation approach (Ntegeka et al., 2014; Willems & Vrac, 2011; Willems, 2013). The latter approach is a technique for statistical downscaling. The tool hence produces perturbed historical series, which are representative for the future climate in Belgium. Time horizons till 2100 can be considered. The perturbed time series can be applied as inputs to models, e.g. hydrological models of river catchments, to simulate and assess future impacts of climate change. For more information or examples on how this can be done, the reader is referred to e.g. Vansteenkiste et al. (2014) for river catchment hydrological impact analysis, and Willems (2013) for urban drainage impact analysis. The current tool (version 2016) differs from earlier versions as it has been made available for hydrological impact modelers in other parts of the world. Given the availability of GCM output time series, the tool can automatically calculate the climate change signals and perturb historical time series following the delta change method, quantile perturbation approach and/or weather typing method. This can be done for each of the individual climate model runs. For Belgium, the tailored climate scenarios for impact analysis on hydrological extremes (wet winter, wet summer, mean, dry) can still be used as well. The scenarios for Belgium were updated based on the latest CMIP5 based climate model runs (Tabari et al., 2014; Tabari et al., 2015). In comparison with the previous versions of the tool that were workbooks in MS Excel, the current tool has been programmed in R. 2 What does the program do? The tool perturbs or changes the input series of precipitation, ETo, temperature and wind speed. For precipitation, both changes in the wet and dry day frequencies and changes in the precipitation intensities are considered. The changes in precipitation intensities are quantile based to account for the fact that the changes might depend on the magnitude or return period of the event. ETo, temperature and wind speed series are transformed by applying monthly average changes. For Belgian applications the user can perturb the observations with the perturbations of Uccle (Belgian climate scenarios - version 2015). Given the availability of GCM output time series, climate change signals here called perturbations (factors or absolute changes) are calculated for each climate model run. Subsequently these perturbations are used in the statistical downscaling methods. Both simple and advanced methods for precipitation are implemented in the tool and can be selected (see Paragraph 7.2.5). The mean delta change method is considered as a simple method and applies mean monthly average changes. The quantile perturbation method and the weather typing based method are at the other hand more advanced methods and INTRODUCTION 5

6 more appropriate when looking into extremes. For the other variables the choice is limited to the mean delta change method. 3 Temporal and spatial scales The tool can handle time series with a time step from daily to sub daily scales (12h, 6h, 3h, 1h, 30min, 20min, 15min, 10min). The scenarios were mainly developed for catchments up to 1000 km². This means that they can be applied to time series representing meteorological data at a point, or several locations in a catchment or limited region, or for catchment averaged data (e.g. catchment areal precipitation). Within the catchment a perfect spatial correlation between all the locations is assumed. Newly defined wet or dry days are hereby situated at the same time position within the time series of each station. 4 Provided time horizons The perturbation is done for a given time horizon in the future. Target years equal to 2020, 2025, 2030, 2035, 2040, 2045, 2050, 2055, 2060, 2065, 2070, 2075, 2080, 2085, 2090, 2095 and 2100 can be selected. 5 Hydrological impact scenarios Belgium For Belgium, next to the perturbations for each of the climate model runs individually, also the tailored climate scenarios for hydrological impact analysis including extremes, can be applied. These are based on the combined changes and effects of precipitation, temperature and ETo, and consist of the following four scenarios: high winter scenario, high summer scenario, mean scenario and low scenario (Ntegeka et al., 2014). The high winter scenario projects futures with wet winters (frontal rainfall) and dry summers, while the high summer scenario projects dry winters and wet summers (convective storm rainfall). The low scenario projects a future with dry winters and summers. (Ntegeka et al., 2014; Vansteenkiste et al., 2014). The current version of the tool makes use of the latest generation climate models (CMIP5 based; Tabari et al., 2014, 2015). Climate scenarios based on the entire ensemble of climate models as well for each representative concentration pathway (RCP) separately are available on request. The risk associated with flooding is most critical for the high impact scenarios, whereas the low scenario is most critical for droughts and related low flows. Users are encouraged to consider all the four impact scenarios to account for the overall uncertainty in the climate scenarios. TEMPORAL AND SPATIAL SCALES 6

7 6 Limitations The rainfall changes embedded in the program were based on climate model output time series with length of around 30 years. This implies that longer time periods required for high return period perturbations are not directly available. Time series longer than 30 years can be perturbed but these perturbations are not more accurate than for time series of 30 years or less. 7 Procedure for running the program 7.1 Structure Climate Perturbation Tool Figure 1 represents the structure of the perturbation tool after extraction of the Climate Perturbation Tool.zip file. Input Folder 1: Graphical user interface Folder 2: Observations Folder 3: Information Folder 4A: Climate model data Folder 4B: Belgian Climate scenarios Tool Folder 5: Source files (R scripts) Output Folder 6: Output - Analysis observations - Analysis climate change signals - Perturbed series Documentation Folder 7: Documentation Figure 1 Structure Climate Perturbation tool 7.2 Run the climate perturbation tool Installation of R and RStudio An installation guide for R and RStudio is provided by Torfs and Brauer (Torfs et al., see Folder 7 Documentation). The required R version is LIMITATIONS 7

8 7.2.2 Start the tool graphical interface Open the file Run GCM Extraction Tool in RStudio (Figure 2). Select all the lines in the R script and click on Run (Figure 3). A pop up window will appear (Figure 4) and ask you to provide the directory of the Climate Perturbation Tool. If required, the tool will automatically install packages and a graphical interface will appear. Figure 2 Open "Run Climate Perturbation Tool" in RStudio Figure 3 Start the interface (Click on run - right corner) PROCEDURE FOR RUNNING THE PROGRAM 8

9 Figure 4 Provide the directory of the "Climate Perturbation Tool" The general size of the interface depends on the size configurations of RStudio (the zoom in or zoom out state) and the configurations of your computer. When no clear overview of the interface is present, the user is advised to maximize the size of the interface Graphical interface: general choices to be made by the user Two output options are included within this tool. The user can select Perturbed series for each climate model run or Perturbed series for each of the tailored climate scenarios for hydrological impact analysis in Belgium (Ntegeka et al., 2014). When the tailored scenarios for Belgium are selected, the Belgian region (coast or inland) needs to be specified. Figure 5 indicates the Belgian coastal area and the inland regions of Belgium. Figure 5 Belgian coastal area Graphical interface: variables general When observed series of a specific variable need to be perturbed, check the box left to the variable name. Provide in the box Number series the number of observed series. When Perturbed series for each climate model run is chosen as output, provide the number of climate model runs that have results for both the control and scenario period, in the box Number climate model runs. PROCEDURE FOR RUNNING THE PROGRAM 9

10 7.2.5 Graphical interface: variable precipitation Tailored climate scenarios for hydrological impact analysis in Belgium When the tailored hydrological impact scenarios for Belgium is chosen as output, the quantile perturbation method is applied (Ntegeka et al., 2014). This method takes into account the change in the wet day frequency and applies quantile relative changes when the total daily rainfall amount is at least 0.1mm. When multiple time series are available, the perturbation can be performed independently (point scale option) or dependently (catchment option). Within the catchment option, spatial correlation is considered when changing the wet day frequency. New wet or dry days are therefore defined at the same moment for all series. Perturbed series for each climate model run When the option perturbed series for each climate model run is selected, more statistical downscaling methods become available. The user can select between the simple mean delta change method (method 1), quantile perturbation methods (methods 2 and 3) and a weather typing based method (method 4). The mean delta change method applies mean changes for all rainfall events. Statistical downscaling methods 2, 3 and 4 are more appropriate when dealing with extremes. Quantile perturbation method 2 changes the wet day frequency and applies quantile relative changes when the total daily rainfall amount is at least 0.1mm. Quantile perturbation method 3 differs from quantile perturbation method 2 as additionally absolute changes are applied beneath a threshold. Absolute changes might first be required if high quantile perturbations are present for daily rainfall events with high exceedance probabilities (or low return periods). Applying absolute changes is moreover needed when the number of new to be defined wet days exceeds the number of present dry days and vice versa. Between 0.1mm/day and the indicated threshold, a linear, weighted combination of absolute and relative changes is applied. (Willems and Vrac, 2011 and Mora et al., 2014) Methods 1, 2 and 3 make direct use of observed precipitation and precipitation climate model data. Method 4, the weather typing based method, consists of two steps. In a first step the weather types for the observations (re-analysis data), control and scenario run are defined. Defining the weather types requires mean sea level pressure series in 16 points centered around the selected study area. The grid point and index scheme for these 16 points is presented in Figure 6 (Jones et al., 1993). Inter longitudinal and inter latitudinal distances measure 10 and 5. The mean sea level pressure in the 16 points and the latitude of the case study area is then used to calculate the pressure gradients and vorticity indices. Thresholds for these pressure gradients and vorticity indices lead finally to the weather types (modified Jenkinson-Collison weather types, Demuzere et al., 2009 and Brisson et al., 2011). The tool can define the weather types and exports them to Folder 2 Observations and Folder 4A Climate model data. The second step includes the resampling of the perturbed precipitation series (method SD-B-7, see Willems and Vrac, 2011). Compared to the mean delta change method and quantile perturbation methods, this method makes no direct use of the precipitation climate model data and requires additional data. Table 1 gives an overview of the required data. PROCEDURE FOR RUNNING THE PROGRAM 10

11 Figure 6 Grid point and index scheme (Jones et al. (1993)) Table 1 Data required for the weather typing method Latitude Observations Climate model data Precipitation Precipitation (control and scenario run) Latitude case study area Mean sea level pressure in 16 points (re-analysis data) or the weather types Mean sea level pressure in 16 points or the weather types (control and scenario run) Mean temperature (control and scenario run) Graphical interface: variable ETo For both output options ( perturbed series for each climate model run and the tailored hydrological impact scenarios for Belgium ), the observed ETo series are perturbed using the mean delta change method Graphical interface: variable temperature For both output options ( perturbed series for each climate model run and the tailored hydrological impact scenarios for Belgium ), the observed (minimum, mean and maximum) temperature series are perturbed using the mean delta change method. The tailored hydrological impact scenarios of minimum and maximum temperature series for (Belgian) coastal applications are not available. PROCEDURE FOR RUNNING THE PROGRAM 11

12 7.2.8 Graphical interface: variable wind speed For both output options ( perturbed series for each climate model run and the tailored hydrological impact scenarios for Belgium ), the observed wind speed series are perturbed using the mean delta change method Graphical interface: directory of the different folders The directory of the different folders needs to be provided. Click on the buttons and select the appropriate folders. When the original folders are used, the user can activate the option Fast folder selection. In that case no further specification of the directories is required Graphical interface: start Analysis observations Before starting the perturbation, the user can verify the required format for observational data (see Paragraph 7.3) by clicking on Start analysis observations. When fulfilled requirements, a basic statistical summary is returned in the interface. Analysis bias The user can assess the skill of the climate model runs by a bias analysis. Click therefore on Start analysis bias. Figures illustrating the bias are returned after a verification of the observations and climate model data (see Paragraph 7.3 and Paragraph 7.4). Statistics are calculated on the entire range of historical climate model data and observations (not for common data range of climate model data and observations). By hand of these figures the user might indicate inappropriate climate model runs and exclude these from the ensemble. Each time an inappropriate run is identified, the run will be excluded from the figure and no perturbed series will be generated for that run. The figures adapt each time a change is made within the interface. Analysis climate change signals The user can directly perturb the observations without a prior analysis of the climate model data. Although, it is advised not to skip the analysis of the climate model data. The behavior of the climate model runs can be investigated through an analysis of the change signals or perturbations. Click therefore on Start analysis climate change signals. Figures illustrating the climate change signals are returned after a verification of the climate model data format (see Paragraph 7.4). By hand of these figures the user might indicate inappropriate climate model runs and exclude these from the ensemble. Each time an inappropriate run is identified, the run will be excluded from the figure and no perturbed series will be generated for that run. The figures adapt each time a change is made within the interface. Perturbation Click on the Start perturbation button in the tab folder Start. During the calculation a popup window will appear and show the progress of calculation. The calculation sequence is: precipitation ETo minimum temperature mean temperature maximum temperature wind speed. When the calculations are finished, a second popup window, similar to that one in Figure 7, will appear. Click on No or Cancel for closing the application. Results can be found in the assigned folders. PROCEDURE FOR RUNNING THE PROGRAM 12

13 Figure 7 End of the calculation 7.3 Observations The input series should be stored as text files in Folder 2. Keep in mind following format. Examples can be found in Folder 7 Documentation Examples Naming of text files: Precipitation series Variable: P1.txt, P2.txt, P3.txt Time: time_p1.txt, time_p2.txt, time_p3.txt (If only one series: P1.txt & time_p1.txt) ETo, temperature and wind speed series: The naming is similar to that one of the precipitation perturbation at point scale : P will be replaced by ETo, Tmin, Tmean, Tmax and WS Mean sea level pressure or weather types: Variable: MSLP.txt or WT.txt Time: time_mslp.txt or time_wt.txt Contents of the text files: No blank space is allowed at the end of the text files. Time format: yyyy-mm-dd hh:mm The tool can handle daily and sub daily time scales (12h, 6h, 3h, 1h, 30min, 20min, 15min, 10min). If sub daily scales are used, the time series should not end during the time span of the last day: Example 1: time step = 1h; the last time notification should be :00 Example 2: time step = 15 minutes; the last time notification should be :45 Exceptions on the rule are mean sea level pressure and weather types. These series should have a daily time step. PROCEDURE FOR RUNNING THE PROGRAM 13

14 Precipitation, ETo, temperature, wind speed and weather types: Data should be put in one column. Mean sea level pressure: Data should be put in 16 columns (spacing by tab), each column represents one of the 16 points defined around the studied station (see Figure 6). Feedback observations: Before perturbing the time series, the format of the observations is verified. Results for this verification are exported to the folder Output Analysis observations. A FALSE in the text files equal_length_obs_p/eto/ _.txt means that for that observational time series (line number in the text file), the time and values do not have the same length. A FALSE in the text files good_format_obs_p/eto/ _.txt indicates possible NA (not a number value), missing values and/or comma in digital numbers. 7.4 Climate model data The climate change scenarios for Belgium (Uccle) are available in Folder 4B. Next paragraphs are important if the option Perturbed series for each climate model run is selected. The climate change signals (perturbations) will be determined using the GCM results stored in Folder 4A. The control and scenario period are often defined as to and to The climate change signals represent then the changes between 1975 (center of the control period) and 2085 (center of the scenario period). Based on the chosen future target year and the time span of your observations, the tool rescales the climate change signals during downscaling. One is advised to define the control and scenario period such the duration is around 30years. GCM results should be stored in text files with following format. Examples can be found in Folder 7 Documentation Examples. Naming of text files: Precipitation: Variable: values_gcmrcm_1_p_contr, values_gcmrcm_1_p_scen.txt, values_gcmrcm_2_p_contr, values_gcmrcm_2_p_scen.txt Time: time_gcmrcm_1_p_contr, time_gcmrcm_1_p_scen.txt, time_gcmrcm_2_p_contr, time_gcmrcm_2_p_scen.txt ETo, Temperature, wind speed, mean sea level pressure and weather types: The naming is similar to that one of precipitation: P will be replaced by Tmin, Tmean, Tmax, WS, MSLP and WT. PROCEDURE FOR RUNNING THE PROGRAM 14

15 Example: - eu_pr_day_bnu-esm_historical_r1i1p1_ (time and precipitation values for the historical period) - eu_pr_day_bnu-esm_rcp26_r1i1p1_ (time and precipitation values for the future period, rcp 2.6) - eu_pr_day_bnu-esm_rcp45_r1i1p1_ (time and precipitation values for the future period, rcp 4.5) Separate the precipitation values and the time. Make text files with the time and precipitation values of the climate model run eu_pr_day_bnu- ESM_historical_r1i1p1_ : values_gcmrcm_1_p_contr.txt, time_gcmrcm_1_p_contr.txt, values_gcmrcm_2_p_contr.txt, time_gcmrcm_2_p_contr.txt. (Two versions are required as there are two future climate model runs.) Make text files with the time and the precipitation values of the climate model run eu_pr_day_bnu-esm_rcp26_r1i1p1_ : values_gcmrcm_1_p_scen.txt, time_gcmrcm_1_p_scen.txt. Make text files with the time and the precipitation values of the climate model run eu_pr_day_bnu-esm_rcp45_r1i1p1_ : values_gcmrcm_2_p_scen.txt, time_gcmrcm_2_p_scen.txt. Contents text files: The name of the model run should be positioned on the first line in all the text files and no blank space is allowed at the end of the text files. Time format: yyyy-mm-dd Precipitation, ETo, temperature, wind speed and weather types: Data should be put in one column. Mean sea level pressure: Data should be put in 16 columns (spacing by tab), each column represents one of the 16 points defined around the studied region (see Figure 6). Feedback climate change signals: Before the calculation of the change signals, the format of the climate model runs is verified. Results for this verification are exported to the folder Output Analysis climate change signals. A FALSE in the text files equal_length_cmr_p/eto/ _contr/scen.txt means that for that specific climate model run (line number in the text file), the text files for the time and the values have not the same length. A FALSE in the text files good_format_cmr_p/eto/ _contr/scen.txt indicates possible NA (not number values), missing values and/or comma in digital numbers. 7.5 Output The tool generates the perturbed time series in text files and stores them in folder Output - Perturbed series. Depending on choices made by the user, the text files will have different names. PROCEDURE FOR RUNNING THE PROGRAM 15

16 Tailored climate scenarios for hydrological impact analysis in Belgium: Precipitation: If the perturbation is done at point scale: P_series_1_high_winter_ impact_scenario.txt, P_series_1_high_summer_impact_scenario.txt P_series_1_mean_impact_scenario.txt P_series_1_low_impact_scenario.txt If the perturbation is done at catchment scale, series is replaced by station. ETo, temperature and wind speed: The naming is similar to that one of precipitation: P is replaced by ETo, Tmin, Tmean, Tmax or WS. The perturbed series will have the same length as the observations. Perturbed series for each climate model run: In this case, perturbed series for the selected range of climate model runs are generated. No perturbed series are generated for the deviating climate model runs, indicated in the tab folders Analysis climate change signals and/or Analysis bias. Precipitation: P_method_1_ series_1_climatemodelrun_1.txt, P_method_1_ series_1_climatemodelrun_2.txt, P_method_1_ series_2_climatemodelrun_1.txt, P_method_1_ series_2_climatemodelrun_2.txt Besides the number of the observed series and the climate model run, a number representing the chosen downscaling method is included in the name of the generated text file. Naming for the output files of ETo, temperature and wind speed is similar. The definition of the method numbers is shown in Table 2 and Table 3. In case of the mean delta change method and quantile perturbation methods the output series will have a length and time step equal to these of the observations. Output series of the weather typing based method cover the time span of the scenario run (30 years) with the time step of the observations. PROCEDURE FOR RUNNING THE PROGRAM 16

17 Table 2 Definition method number for precipitation downscaling methods Method 1 Mean delta change method Method 2 Quantile perturbation method (only relative changes - Ntegeka et al., 2014) Method 3 Quantile perturbation method (both absolute and relative changes - Ntegeka et al., 2014) Method 4 Weather typing based statistical downscaling method Table 3 Definition method number for ETo, minimum, mean and maximum temperature and wind speed downscaling methods Method 1 Mean delta change method 7.6 Information about the perturbation request Information about the perturbation request, such as the variables to be perturbed, the number of series, the target scenario year, the statistical downscaling method, is provided and stored in Folder 3. Table 4 on p.18 includes more information about the variables in the exported files. PROCEDURE FOR RUNNING THE PROGRAM 17

18 8 Appendix A: Information on perturbation request Table 4 Information about the perturbation request Variable Value Meaning Output 0 or 1 0 = Perturb the observations for each climate model run 1 = Tailored hydrological impact scenarios for Belgium Region 0 or 1 When the tailored hydrological impact scenarios for Belgium are selected: 0 = Belgium inland region 1 = Belgium coastal region Target year scenario period {2020, 2025, 2030,, 2095, 2100} Pmodule, ETomodule, Tminmodule, Tmeanmodule, Tmaxmodule, WSmodule Number P series Number ETo series, Number Tmin series, Number Tmean series, Number Tmax series, Number WS series Number climate model runs P (& MSLP/WT), Number climate model runs ETo, Number climate model runs Tmin, Number climate model runs Tmean, Number climate model runs Tmax, Number climate model runs WS Excluded climate model runs P (& MSLP/WT) Excluded climate model runs ETo Excluded climate model runs Tmin Excluded climate model runs Tmean Excluded climate model runs Tmax Excluded climate model runs WS Ensemble downscaling methods P TRUE or FALSE , 1, 2, 3 or 4 The target scenario year (the year for which future projections are required) FALSE = no observed series for the variables P, ETo, Tmin, Tmean, Tmax, WS TRUE = observed series for the variable P, ETo, Tmin, Tmean, Tmax, WS are present Number of observed series Number of climate model runs which have results for both the control and the scenario period. The climate model runs are used for the calculation of the climate change signals Similar text files are provided for the other variables (MSLP/WT, ETo, Tmin, Tmean, Tmax and WS) Number of the climate model run(s) that shows (show) deviating behavior and needs (need) to be excluded from the ensemble. 0 = No downscaling for precipitation 1 = Downscaling using all the perturbation methods and the weather typing method 2 = Downscaling using all the perturbation methods 3 = Downscaling using a selected perturbation method APPENDIX A: INFORMATION ON PERTURBATION REQUEST 18

19 4 = Downscaling using the weather typing method Downscaling method P 0, 1, 2, 3,4 0 = No additional specification for the precipitation downscaling methods is required If ensemble downscaling methods P = 3, then a perturbation method needs to be selected. 1 = mean delta change method 2 = quantile perturbation method (relative intensity changes + change in the wet day frequency) 3 = quantile perturbation method (absolute & relative intensity changes) 4 = weather typing based statistical downscaling method Catchment scale 0 or 1 1 = Perturbation of the precipitation series at catchment scale spatial correlation is included in the change in the wet day frequency 0 = Perturbation of the precipitation series at point scale Threshold absolute changes 1mm/day If perturbation of the observations occurs using absolute and relative intensity changes, a threshold for absolute change changes needs to be specified. Define WT Downscaling method ETo Downscaling method Tmin Downscaling method Tmean Downscaling method WS TRUE or FALSE TRUE = The weather types are defined by the tool. The user needs to provide the MSLP. FALSE = The weather types are defined by the user. 1 The mean delta change method is applied on the observed series. Latitude case study region [ ] or NA The weather typing method requires the specification of the latitude of the case study area. APPENDIX A: INFORMATION ON PERTURBATION REQUEST 19

20 9 References Brisson E., Demuzere M., Kwakernaak B. and Van Lipzig N. (2011). Relations between atmospheric circulation and precipitation in Belgium. Meteorology and Atmospheric Physics, 111, 27-39, doi: /s y Bultot F., Coppens A. and Dupriez G. (1983). Estimation de l évapotranspiration potentielle en Belgique. Publications/publicaties série/serie A, No/Nr 112, Institut Royal Météorologique de Belgique - Koninklijk Meteorologisch Instituut van België, 28 p. Demuzere, M., Werner, M., van Lipzig, N.P.M., Roeckner, E., (2009). An analysis of present and future ECHAM5 pressure fields using a classification of circulation patterns. International Journal of Climatology, 29, , doi: /joc.1821 Jones, P., Hulme, M., and Briffa, K. (1993). A comparison of lamb circulation types with an objective classification scheme. International Journal of Climatology, 13, Mora D. E, Campozano F., Cisneros F., Wyseure G. and Willems P. (2014). Climate changes of hydrometeorological extremes in the Paute basin, Ecuadorean Andes, Hydrology and Earth System Sciences, 18, , doi: /hess Ntegeka, V. and Willems P. (2008). Trends and multidecadal oscillations in rainfall extremes, based on a more than 100-year time series of 10 min rainfall intensities at Uccle, Belgium. Water Resources Research, 44, W07402, doi: /2007wr Ntegeka V. and Willems P. (2009). CCI-HYDR Perturbation Tool: a climate change tool for generating perturbed time series for the Belgian climate. Manual version January 2009, KU Leuven Hydraulics section & Royal Meteorological Institute of Belgium, 7 p. Ntegeka V., Willems P., Roulin E. and Baguis P. (2014). Developing tailored climate change scenarios for hydrological impact assessments. Journal of Hydrology, 508C, , doi.org/ /j.jhydrol R Core Team, (2014). R: A language and environment for statistical computing. R Founda tion for Statistical Computing, Vienna, Austria. URL Tabari H., Taye M.T. and Willems P. (2014). Bijsturing van de Vlaamse klimaatscenario s voor hydrologische en hydrodynamische impactanalyse inclusief hydrologische extremen. Studie uitgevoerd in opdracht van de Afdeling Operationeel Waterbeheer van de Vlaamse Milieumaatschappij en MIRA 2014, door KU Leuven Afdeling Hydraulica, november 2014, 106 p. Tabari H., Taye M.T. and Willems P. (2015). Water availability change in central Belgium for the late 21th century. Global and Planetary Change, 131, , doi.org/ /j.gloplacha Torfs P. and Brauer C. (2014). A (very) short introduction to R, Wageningen University - Hydrology and Quantitative Water Management Group URL REFERENCES 20

21 Vansteenkiste T., Tavakoli M., Ntegeka V., De Smedt F., Batelaan O., Pereira F. and Willems P. (2014). Intercomparision of hydrological model structures and calibration approaches in climate scenario impact projections. Journal of Hydrology, 519, , doi.org/ /j.jhydrol Willems P. and Vrac M. (2011). Statistical precipitation downscaling for small-scale hydrological impact investigations of climate change. Journal of Hydrology, 402, , doi.org/ /j.jhydrol Willems, P. (2013). Revision of urban drainage design rules after assessment of climate change impacts on precipitation extremes at Uccle, Belgium. Journal of Hydrology, 496, , doi.org/ /j.jhydrol REFERENCES 21

22

23 AFDELING HYDRAULICA Kasteelpark Arenberg 40 bus HEVERLEE (LEUVEN), BELGIË tel tel. secr fax Patrick.Willems@bwk.kuleuven.be bwk.kuleuven.be/hydr