Database of CMIP5 climate change projections across south Asia

Transcription

1 LAND AND WATER Database of CMIP5 climate change projections across south Asia Providing a consistent baseline projection of future changes in precipitation, temperature and potential evaporation across the south Asia region Hongxing Zheng, Francis Chiew, Steve Charles July 2015

2 Citation Zheng H, Chiew F, Charles SP (2015) Database of CMIP5 climate change projections across south Asia. CSIRO Sustainable Development Investment Portfolio project. CSIRO Land and Water, Australia. This report designed and implemented by CSIRO contributes to the South Asia Sustainable Development Investment Portfolio and is supported by the Australian Government. Copyright With the exception of the Australian government crest, Australian Aid and CSIRO logos, and where otherwise noted, all material in this publication is provided under a Creative Commons Attribution 4.0 International License Under this licence you are free to share (copy and redistribute in any medium or format) and adapt (remix, transform) for any purpose, even commercially, under the following terms: The authors request attribution as Australian Government Department of Foreign Affairs and Trade (DFAT) (Sustainable Development Investment Portfolio). Important disclaimer CSIRO advises that the information contained in this publication comprises general statements based on scientific research. The reader is advised and needs to be aware that such information may be incomplete or unable to be used in any specific situation. No reliance or actions must therefore be made on that information without seeking prior expert professional, scientific and technical advice. To the extent permitted by law, CSIRO (including its employees and consultants) excludes all liability to any person for any consequences, including but not limited to all losses, damages, costs, expenses and any other compensation, arising directly or indirectly from using this publication (in part or in whole) and any information or material contained in it. [EP156969]

3 Contents Summary v 1 Introduction 1 2 Data source CMIP5 global climate models (GCMs) Representative Concentration Pathways (RCPs) Gridded climate surfaces Empirical scaling factor of climate variables 5 4 Projected climate changes in south Asia Range of climate projections Overview of climate projections Exploring range and uncertainty in precipitation projections Description of the gridded database Empirical scaling factors for each ensemble run from each GCM Median and range of empirical scaling factors Application of the scaling factors Database of CMIP5 climate change projections across south Asia i

4 Figures Figure 1 Summary of GCM experiment runs in CMIP5. The climate variable names in the horizontal axis are standard abbreviations in CMIP5 (Taylor et al., 2012, and online references therein). The climate model names are shown in the vertical axis. The shading indicates the number of ensemble runs available for historical, RCP2.6, RCP4.5, RCP6.0, RCP8.5 and preindustrial control experiments. (Source: IPCC AR5)... 2 Figure 2 Global mean radiative forcing (RF, Wm 2) for the different RCPs. (Source: IPCC AR5).. 3 Figure 3 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average daily mean temperature for relative to current for the 12 months, 4 seasons and the annual scale for RCP Figure 4 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average daily mean temperature for relative to current for the 12 months, 4 seasons and the annual scale for RCP Figure 5 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average daily maximum temperature for relative to current for the 12 months, 4 seasons and the annual scale for RCP Figure 6 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average daily maximum temperature for relative to current for the 12 months, 4 seasons and the annual scale for RCP Figure 7 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average daily minimum temperature for relative to current for the 12 months, 4 seasons and the annual scale for RCP Figure 8 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average daily minimum temperature for relative to current for the 12 months, 4 seasons and the annual scale for RCP Figure 9 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average total potential evapotranspiration for relative to current for the 12 months, 4 seasons and the annual scale for RCP Figure 10 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average total potential evapotranspiration for relative to current for the 12 months, 4 seasons and the annual scale for RCP Figure 11 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average total precipitation for relative to current for the 12 months, 4 seasons and the annual scale for RCP Figure 12 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average total precipitation for relative to current for the 12 months, 4 seasons and the annual scale for RCP Figure 13 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average extreme high daily precipitation (left: RCP4.5; right: RCP8.5) ii Database of CMIP5 climate change projections across south Asia

5 for relative to current for the 12 months, 4 seasons and the annual scale. R001, R005 and R10 represents 1th, 5th and 10th percentile of daily precipitation series respectively Figure 14 Locations where range and uncertainty in the precipitation projections are explored in more detail Figure 15 Projected changes in future mean annual precipitation ( relative to current) from ensemble runs from the 42 GCMs (G01 to G42, see Table 1 to identify the GCMs) and range of results from first run of the 42 GCMs (G00 in first column) for RCP Figure 16 Projected changes in future mean annual precipitation ( relative to current) from ensemble runs from the 42 GCMs (G01 to G42, see Table 1 to identify the GCMs) and range of results from first run of the 42 GCMs (G00 in first column) for RCP Figure 17 Percentage change in future annual precipitation simulated by the 40 GCMs versus their RMSE weighting ( relative to current) for (top) RCP4.5 and (bottom) RCP Figure 18 Range of projected change in mean annual precipitation ( relative to current) for the eight locations using all 42 GCMs and the best 10 and best 25 GCMs measured by the RMSE indicating their ability to simulate historical precipitation for (top) RCP4.5 and (bottom) RCP Figure 19 Range of projected change in mean annual precipitation ( relative to current) for the eight locations using all 42 GCMs and the best GCMs up to 25%, 50% and 75% accumulated RMSE weightings for (top) RCP4.5 and (bottom) RCP Figure 20 Example format of data file in the ScalingFactor_GCM dataset with scaling factors for each of the 12 months, each of the 4 seasons and annual amounts across the south Asia region Figure 21 Example format of data file in the ScalingFactor_Summary dataset with scaling factors for each of the 12 months, each of the 4 seasons and annual amounts across the South Asia region Tables Table 1 The 42 CMIP5 models used to derived the range of future climate projections... 3 Table 2 Description of the eight selected locations where precipitation projections are explored in more detail Database of CMIP5 climate change projections across south Asia iii

6 iv Database of CMIP5 climate change projections across south Asia

7 Summary Climate change will impact water and related sectors. Temperature and potential evaporation will be higher. Changes in future precipitation will be amplified in the river flows. Security of water supply will be compromised due to longer and more severe droughts, more precipitation falling as rain rather than snow, increased seasonality of river flow and retreat of glaciers. Flood risk will increase due to more intense heavy precipitation events. This report describes the SDIP climate change database, which Specifically, the database presents empirical scaling factors for 0.5 o grids (~ 50 km) that reflect changes in six climate variables (precipitation, heave precipitation, potential evaporation, average daily temperature, maximum daily temperature and minimum daily temperature) for a future ( ) period relative to current. Changes are presented for each of the 12 months, 4 seasons and annual values. These are presented for each ensemble modelling run from each of the 42 CMIP5 GCMs, as well as the median and range (uncertainty) of plausible projections. This database therefore provides consistently derived projections for climate change impact modelling in the SDIP projects in the various basins and for hydrological modelling across South Asia to inform water management, planning and development, and their interactions with the energy and food sectors, to improve livelihoods in the region. A complementary report (Charles et al., 2015) provides an overview of the various climate change projections and basin and regional modelling studies across South Asia. [The climate change project data can be downloaded from Database of CMIP5 climate change projections across south Asia v

8

9 1 Introduction Climate change will impact water and related sectors. The warmer future climate will increase evapotranspiration and hence increase demand for water in irrigated agriculture, urban centres and water dependent ecosystems. The projected changes in precipitation will be amplified in the changes in river flows. The climate change impact will be reflected not only in the averages but also in other river flow characteristics. Hydrologic variability will be enhanced, with longer droughts and longer wet periods. In high altitude and cold regions, more precipitation will fall as rain rather than snow, and the snow will melt earlier, resulting in earlier peak flow and more winter and less spring and summer flows. The seasonality in river flow is also likely to increase with wet seasons becoming wetter and dry seasons becoming drier. The retreat of glaciers will increase river flows in the short term, but the contribution of glacier melt will gradually fall as glaciers shrink. This enhanced hydrologic variability and decrease in snow and ice storage will reduce the reliability of water supply, thus compounding water management and climate adaptation challenges. Heavy precipitation events will become more intense in the future. This will increase flood damage to settlements, infrastructures like roads and bridges, livestock and crops. Coastal areas, where many people live, will have increased risk from river and coastal flooding. This report describes the compilation of a database of future changes in precipitation, temperature and potential evaporation across the South Asia region (5.25 o o S, o o E), presented as empirical scaling factors reflecting the change in the climate variable in the future ( ) relative to current climatology for 0.5 o (~50 km) grids across the region. This consistent baseline climate change projection across the South Asia region is derived by analysing the latest CMIP5 global climate model results (as used and discussed in the IPCC AR5, Fifth Assessment Report of the Intergovernmental Panel on Climate Change). Section 2 describes the global climate models and gridded climate datasets used for the analyses. Section 3 describes the method used to estimate the empirical scaling factors for monthly, seasonal and annual values. Section 4 summarises the future climate change projections (as the median and the range or uncertainty in the climate projections), and explores the influence of GCM choice on the future precipitation projections. Section 5 describes the climate change projections database which presents the empirical scaling factors for each ensemble run from each of the 42 GCMs and the median and range of future climate projections. Database of CMIP5 climate change projections across south Asia 1

10 2 Data source 2.1 CMIP5 global climate models (GCMs) The range of future climate scenarios in this report is derived from the CMIP5 database ( pcmdi.llnl.gov/cmip5/). CMIP5 (the fifth phase of the Coupled Model Intercomparison Project) is sponsored by WCRP's Working Group on Coupled Modelling (WGCM) with input from the IGBP AIMES project. It involves global climate model experiments from more than 20 climate modelling groups around the world (Table 1). CMIP5 provides a standard set of model simulations to evaluate the models ability in simulating the recent past, to provide projections of future climate change, and to understand factors responsible for differences in model projections. The CMIP5 modelling results are used for reporting in the Intergovernmental Panel on Climate Change Fifth Assessment Report (IPCC AR5). Table 1 and Figure 1 summarise the 42 CMIP5 GCMs and the number of ensemble runs in each GCM. The range of future climate scenarios presented in this report are derived from the CMIP5 database on 15 March 2013 (the same as that adopted by IPCC AR5). Figure 1 Summary of GCM experiment runs in CMIP5. The climate variable names in the horizontal axis are standard abbreviations in CMIP5 (Taylor et al., 2012, and online references therein). The climate model names are shown in the vertical axis. The shading indicates the number of ensemble runs available for historical, RCP2.6, RCP4.5, RCP6.0, RCP8.5 and pre industrial control experiments. (Source: IPCC AR5) 2 Database of CMIP5 climate change projections across south Asia

11 Figure 2 Global mean radiative forcing (RF, Wm 2) for the different RCPs. (Source: IPCC AR5) Table 1 The 42 CMIP5 models used to derived the range of future climate projections ID Model Modeling Center (or Group) G01 ACCESS1 0 Commonwealth Scientific and Industrial Research Organization (CSIRO) and Bureau of G02 ACCESS1 3 Meteorology (BOM), Australia G03 BCC CSM1 1 M Beijing Climate Center, China Meteorological Administration G04 BCC CSM1 1 G05 BNU ESM College of Global Change and Earth System Science, Beijing Normal University G06 CANESM2 Canadian Centre for Climate Modelling and Analysis G07 CCSM4 National Center for Atmospheric Research G08 CESM1 BGC Community Earth System Model Contributors G09 G10 CESM1 CAM5 CESM1 WACCM G11* CMCC CESM Centro Euro Mediterraneo per I Cambiamenti Climatici G12 G13 CMCC CMS CMCC CM G14 CNRM CM5 Centre National de Recherches Météorologiques / Centre Européen de Recherche et Formation Avancée en Calcul Scientifique G15 CSIRO MK3 6 0 Commonwealth Scientific and Industrial Research Organization in collaboration with Queensland Climate Change Centre of Excellence G16 EC EARTH EC EARTH consortium G17 FGOALS G2 LASG, Institute of Atmospheric Physics, Chinese Academy of Sciences and CESS,Tsinghua University G18 FGOALS S2 LASG, Institute of Atmospheric Physics, Chinese Academy of Sciences G19 FIO ESM The First Institute of Oceanography, SOA, China G20 GFDL CM3 NOAA Geophysical Fluid Dynamics Laboratory G21 GFDL ESM2G Database of CMIP5 climate change projections across south Asia 3

12 ID Model Modeling Center (or Group) G22 GFDL ESM2M G23 GISS E2 H CC NASA Goddard Institute for Space Studies G24 GISS E2 H G25 GISS E2 R CC G26 GISS E2 R G27 HADGEM2 AO National Institute of Meteorological Research/Korea Meteorological Administration G28 HADGEM2 CC Met Office Hadley Centre (additional HadGEM2 ES realizations contributed by Instituto G29 HADGEM2 ES Nacional de Pesquisas Espaciais) G30 INMCM4 Institute for Numerical Mathematics G31 IPSL CM5A LR Institut Pierre Simon Laplace G32 IPSL CM5A MR G33 IPSL CM5B LR G34 MIROC ESM CHEM Japan Agency for Marine Earth Science and Technology, Atmosphere and Ocean Research G35 MIROC ESM Institute (The University of Tokyo), and National Institute for Environmental Studies G36 MIROC5 G37 MPI ESM LR Max Planck Institut für Meteorologie (Max Planck Institute for Meteorology) G38 MPI ESM MR G39* MRI ESM1 Meteorological Research Institute G40 MRI CGCM3 G41 NORESM1 ME Norwegian Climate Centre G42 NORESM1 M *Note: model runs are available only for RCP Representative Concentration Pathways (RCPs) The transient climate experiments in CMIP5 are conducted in three phases. The first phase covers the start of the modern industrial period through to the present day, years The second phase covers the future, , and is described by a collection of RCPs adopted in IPCC AR5 (Moss et al., 2010; Taylor et al., 2010). The third phase is described by a corresponding collection of Extension Concentration Pathways (Meinshausen et al., 2011). For the projections here, RCP4.5 and RCP8.5 are used, representing radiative forcing of +4.5 and +8.5 W/m2 respectively in the year 2100 relative to pre industrial values. Emissions in RCP 4.5 peak around 2040 and then decline, while emissions in RCP8.5 continue to rise throughout the 21st century (Figure 2). The median global mean temperature (median of simulations from the different GCMs) in relative to is 1.4 o C and 2.0 o C higher for RCP4.5 and RCP8.5 respectively. 2.3 Gridded climate surfaces The gridded climate surface datasets of South Asia (Chen et.al, 2014) are used to evaluate the performance of the CMIP5 models. The gridded climate surface datasets includes precipitation dataset from Aphrodite, Princeton and IMD. The Aphrodite (Asian Precipitation Highly Resolved Observational Data Integration Towards Evaluation of the Water Resources) is a database of 0.25 o 4 Database of CMIP5 climate change projections across south Asia

13 (~25 km) gridded daily precipitation for continental Asia from (Yatagai et al., 2012). The Princeton data product is a global 0.5 o (~50 km) gridded dataset of daily climate data from (Sheffield et al., 2006). The IMD (India Meteorological Department) dataset provides 0.25 (~25 km) gridded daily rainfall data across India from (Sridhar et al., 2014). Considering the relative reliability of the dataset, in the evaluation of the CMIP5 model, the first choice of the dataset is IMD followed by Aphrodite. The criteria of model evaluation are the root meansquared error (RMSE) of the ranked simulated annual rainfall against ranked observed annual rainfall from the datasets for the period Lower RMSE (or higher RMSE weight) means better simulation in the probability distribution of annual rainfall. The RMSE weight is the inverse of the proportion of single model s RMSE to total RMSE. 3 Empirical scaling factor of climate variables The range of potential changes to three key climate variables are evaluated, precipitation, temperature and potential evaporation. For precipitation, changes to long term average monthly, seasonal and annual amounts, as well as changes to extreme high daily precipitations (1st, 5th and 10th percentiles) are evaluated. For temperature, changes to long term average monthly, seasonal and annual mean daily temperature, maximum daily temperature and minimum daily temperature are evaluated. For potential evapotranspiration, changes to long term average monthly, seasonal and annual amounts are evaluated. The potential evaporation for each GCM grid is estimated from the solar radiation, maximum and minimum temperatures, and actual vapour pressure data using Morton s wet environment or equilibrium evaporation formulation (Morton 1983). The empirical scaling factor is defined as the change in future climate condition relative to current (or historical) climatology, where the historical and future here are represented by the GCM simulations for the period and respectively. The empirical scaling factor for the climate variables is expressed as the ratio of change, SF X /X (1) where X f and X h are the GCM simulation for the future and historical periods respective (e.g. X f may represent mean annual precipitation simulated by a GCM and X h may represent mean annual precipitation simulated by the same GCM). The scaling factor of the temperature is calculated as the difference between the two periods: SF X X (2) The scaling factors for each of the GCMs (the GCMs have different spatial resolutions) are resampled to present the results for common 0.5 o x 0.5 o grids (~50 km x 50 km) across the South Asia region (5.25 o o S, o o E). These scaling factors can be used to modify historical daily climate sequences for input into hydrological models, using the empirical scaling or delta change method, to assess potential climate change impact on water availability and hydrological and river flow characteristics (Chiew et al., 2009a). Using the large range of GCMs provides a useful indication of the plausible range (and uncertainty) of future climates and impact on water and related sectors. Database of CMIP5 climate change projections across south Asia 5

14 4 Projected climate changes in south Asia 4.1 Range of climate projections Figure 4 to Figure 13 show the projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) for RCP4.5 and RCP8.5 for relative to current for the 12 months, 4 seasons and the annual scale in the: long term average daily mean temperature (Figure 3 and Figure 4) long term average daily maximum temperature (Figure 5 and Figure 6) long term average daily minimum temperature (Figure 7 and Figure 8) long term average total potential evapotranspiration (Figure 9 and Figure 10) long term average total precipitation (Figure 11 and Figure 12) long term average extreme high daily precipitation (Figure 13). Figure 3 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average daily mean temperature for relative to current for the 12 months, 4 seasons and the annual scale for RCP4.5 Figure 4 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average daily mean temperature for relative to current for the 12 months, 4 seasons and the annual scale for RCP8.5 6 Database of CMIP5 climate change projections across south Asia

15 Figure 5 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average daily maximum temperature for relative to current for the 12 months, 4 seasons and the annual scale for RCP4.5 Figure 6 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average daily maximum temperature for relative to current for the 12 months, 4 seasons and the annual scale for RCP8.5 Database of CMIP5 climate change projections across south Asia 7

16 Figure 7 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average daily minimum temperature for relative to current for the 12 months, 4 seasons and the annual scale for RCP4.5 Figure 8 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average daily minimum temperature for relative to current for the 12 months, 4 seasons and the annual scale for RCP8.5 Figure 9 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the long term average total potential evapotranspiration for relative to current for the 12 months, 4 seasons and the annual scale for RCP4.5 8 Database of CMIP5 climate change projections across south Asia

17 Figure 10 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the longterm average total potential evapotranspiration for relative to current for the 12 months, 4 seasons and the annual scale for RCP8.5 Figure 11 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the longterm average total precipitation for relative to current for the 12 months, 4 seasons and the annual scale for RCP4.5 Figure 12 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the longterm average total precipitation for relative to current for the 12 months, 4 seasons and the annual scale for RCP 8.5 Database of CMIP5 climate change projections across south Asia 9

18 Figure 13 Projected range of change (median, 10th, 25th, 75th and 90th percentiles from the 40 GCMs) in the longterm average extreme high daily precipitation (left: RCP4.5; right: RCP8.5) for relative to current for the 12 months, 4 seasons and the annual scale. R001, R005 and R10 represents 1th, 5th and 10th percentile of daily precipitation series respectively 4.2 Overview of climate projections The ranges of climate projections presented here are relatively similar to those in the IPCC AR5 atlas of global and regional climate projections (IPCC, 2013). However, the projections here come from a much larger set of global climate models to more reliably inform hydrological modelling across South Asia to assess the potential range of climate change impact on water and related sectors. The text below provides a broad summary of the climate change projections presented in Section 4.1. A more detailed overview of climate change projections across south Asia and regional modelling studies are discussed in Charles et al. (2015). There is strong agreement between the GCMs in the temperature projections. Averaged across the south Asia region, the median projection for RCP4.5 is an increase in daily mean temperature of 2.1 o C by relative to current, with a 25 75th percentile range of 1.7 to 2.4 o C and 10th 90th percentile range of 1.4 to 2.8 o C. The median projection for RCP8.5 is an increase in daily mean temperature of 2.9 o C by relative to current, with a 25 75th percentile range of 2.6 to 3.5 o C and 10th 90th percentile range of 2.3 to 4.0 o C. The projected increase in daily minimum temperature is slightly higher and the projected increase in daily maximum temperature is slightly lower than the projected increase in daily mean temperature. Seasonally, the projected temperature increase is slightly higher in winter (DJF) than summer (JJA). Spatially, the projected temperature increase is noticeably higher in the north (high altitude regions) than in the south. There is also general agreement between the GCMs in the potential evaporation projections. The projected increase in potential evaporation is driven mainly by the increase in temperature. Most 10 Database of CMIP5 climate change projections across south Asia

19 models show an increase in relative humidity (which will reduce potential evaporation), but show little change and agreement between models in the direction of change of other variables (solar radiation and wind speed) that influence potential evaporation. Averaged across the South Asia region, the median projection for RCP4.5 is an increase in mean annual potential evaporation of 4.5 % by relative to current, with a 25 75th percentile range of % and 10th 90th %. The median projection for RCP8.5 is an increase in mean annual potential evaporation of 6.2 % by relative to current, with a 25 75th percentile range of % and 10th 90th %. Some models project a decrease in potential evaporation mainly due to higher relative humidity, and possibly decrease in wind speed and solar radiation and little increase in temperature. Seasonally, the projected increase in potential evaporation is considerably greater in winter (DJF) than summer (JJA). Spatially, the projected increase in potential evaporation tends to be greater in the north west. There is much greater uncertainty in the precipitation projections, with significant variation between models, and in the different seasons and regions. Nevertheless, a higher proportion of GCMs project an increase in precipitation, particularly in the north east and much more so in the monsoon summer (JJA) than winter (DJF). The projections also suggest possible intensification in the high extreme precipitation, but with the same disagreement between models and not much difference in the projected changes compared to the mean annual precipitation. 4.3 Exploring range and uncertainty in precipitation projections The range and uncertainty in the projected changes in future precipitation are explored for eight selected locations as listed in Table 2 and shown in Figure 14. Figure 14 Locations where range and uncertainty in the precipitation projections are explored in more detail Database of CMIP5 climate change projections across south Asia 11

20 Table 2 Description of the eight selected locations where precipitation projections are explored in more detail LOCATIONS LONGITUDE LATITUDE BASINS L Koshi L Bangladesh L Brahmani Baitarni L Indus L L L L Uncertainty within and across CMIP5 models Figure 15 and Figure 16 shows all the projected changes in future mean annual precipitation ( relative to current) for each of the ensemble GCM runs for each of the GCMs for RCP4.5 (40 GCMs) and RCP8.5 (42 GCMs) respectively for the eight locations. The first column in all the plots shows the results from the first run of each of the 40 or 42 GCMs, which is used to describe the range of projections in this study (in Section 4.1). It is interesting to note that the range of projections from ensembles of some GCMs can be as large as the range of projections between the different GCMs. Figure 15 Projected changes in future mean annual precipitation ( relative to current) from ensemble runs from the 42 GCMs (G01 to G42, see Table 1 to identify the GCMs) and range of results from first run of the 42 GCMs (G00 in first column) for RCP Database of CMIP5 climate change projections across south Asia

21 Figure 16 Projected changes in future mean annual precipitation ( relative to current) from ensemble runs from the 42 GCMs (G01 to G42, see Table 1 to identify the GCMs) and range of results from first run of the 42 GCMs (G00 in first column) for RCP Projections from all GCMs versus smaller set of better performing GCMs We explore here whether the use of the better GCMs can improve and/or reduce the uncertainty in the projected changes to mean annual precipitation. Figure 17 shows the projected change in mean annual precipitation in each of the 40 GCMs for RCP4.5 and 42 GCMs for RCP8.5 plotted against the weighted RMSE (root mean square error). The RMSE is calculated by ranking the annual precipitation simulated by the GCM and the observed precipitation and comparing the 30 values at the same ranks, therefore measuring the ability of the GCM in simulating the historical annual precipitation. The weighted RMSE is the inverse of the proportion of the particular GCM s RMSE to the total RMSE. A higher weighted RMSE (points towards the right hand side of the x axis in Figure 17) reflects that the GCM can reproduce the observed annual precipitation amounts and distribution relative to other GCMs. Figure 18 shows the range of projections for the eight locations for RCP4.5 and RCP8.5 respectively using all the GCMs and the best 10 and best 25 GCMs measured by the RMSE indicating the GCM ability in simulating the historical annual precipitation amounts and variability. Figure 19 shows the range of projections for the eight locations for RCP4.5 and RCP8.5 using all the GCMs up to 25%, 50% and 75% of accumulated relative RMSE weightings. There is no clear trend in Figure 17 showing different projections between the better and poorer GCMs. There is also no clear trend in Figure 18 and Figure 19, although using the better Database of CMIP5 climate change projections across south Asia 13

22 GCMs in a couple of locations can give different results than using all GCMs, but with little difference in the range (e.g. Locations 1 and 7 in Figure 18). The results here suggest that it may be worth exploring assessing and weighting projections towards the better GCMs in detailed catchment or regional studies. However, assessing the GCMs against different criteria (e.g. against precipitation or other variables, against precipitation or correlation between large scale drivers and precipitation, against average precipitation or precipitation characteristics (e.g. monsoon) influencing runoff, etc) may give different results. As such, for broad scale hydrological modelling across South Asia, or probably also for detailed regional studies, with current GCM capabilities it is probably best to use the entire set of available GCMs to represent the entire range of plausible uncertainty (Chiew et al., 2009b). 14 Database of CMIP5 climate change projections across south Asia

23 Figure 17 Percentage change in future annual precipitation simulated by the 40 GCMs versus their RMSE weighting ( relative to current) for (top) RCP4.5 and (bottom) RCP8.5 Database of CMIP5 climate change projections across south Asia 15

24 Figure 18 Range of projected change in mean annual precipitation ( relative to current) for the eight locations using all 42 GCMs and the best 10 and best 25 GCMs measured by the RMSE indicating their ability to simulate historical precipitation for (top) RCP4.5 and (bottom) RCP Database of CMIP5 climate change projections across south Asia

25 Figure 19 Range of projected change in mean annual precipitation ( relative to current) for the eight locations using all 42 GCMs and the best GCMs up to 25%, 50% and 75% accumulated RMSE weightings for (top) RCP4.5 and (bottom) RCP8.5 Database of CMIP5 climate change projections across south Asia 17

26 5 Description of the gridded database 5.1 Empirical scaling factors for each ensemble run from each GCM The scaling factor database contains all the scaling factors at the spatial resolution of 0.5 o x 0.5 o across South Asia (5.25 o o S, o o E). The folder ScalingFactor_GCM contains the empirical scaling factors from all ensemble runs from all the 42 GCMs. There are six subfolders prcp 1, pexm, pet, tas, tasmax and tasmin containing the empirical scaling factors for precipitation, extreme high daily rainfall, potential evapotranspiration, mean temperature, maximum temperature and minimum temperature. In the subfolder, each text file represents the scaling factors of each GCM model and each run. The text file is named like: ModelName_r1i1p1_r45 ave.txt where r1i1p1 is the indication of the model run, r45 (or r85) represents RCP4.5 (or RCP8.5). For example GFDL CM3_run1_r45.txt refers to GFDL CM3 model, first ensemble run from the model, and for RCP4.5. Figure 20 shows an example of each file of empirical scaling factors. The first two columns show the longitude and latitude in decimal degree. The next 12 columns are scaling factors for each of the 12 months, followed by four columns with the seasonal scaling factors and the annual scaling factor in the last column. For extreme high daily rainfall, there are three times as many data columns showing the scaling factors for the 1%, 5% and 10% highest rainfall. Figure 20 Example format of data file in the ScalingFactor_GCM dataset with scaling factors for each of the 12 months, each of the 4 seasons and annual amounts across the south Asia region 5.2 Median and range of empirical scaling factors The folder ScalingFactor_Summary contains the summary empirical scaling factors (median and range of the 42 GCMs), as presented in Section 4.1. Like the above dataset for all GCMs, there are also six subfolders for the six climate variables. The data file name is relatively obvious, for 1 Precipitation scaling factors that are unrealistically low or high are set to 0.3 and 3.0.This can occur when the monthly (and sometimes seasonal) precipitation is very low. These low precipitation amounts have little impact on hydrological modelling and related outcomes. 18 Database of CMIP5 climate change projections across south Asia

27 example prcp_rcp4.5_p10 refers to the 10th percentile of the scaling factor for mean annual precipitation for RCP4.5. Similar to the scaling factors for each GCM model, the first two columns of the files in summary dataset are the longitude and latitude in decimal degree, then followed by 12 columns of scaling factors for each of the 12 months and four columns with the seasonal scaling factors and the annual scaling factor in the last column. For extreme high daily rainfall, there are three times as many data columns showing the scaling factors for the 1%, 5% and 10% highest rainfall. Figure 24 shows an example of the summary file. Figure 21 Example format of data file in the ScalingFactor_Summary dataset with scaling factors for each of the 12 months, each of the 4 seasons and annual amounts across the South Asia region 5.3 Application of the scaling factors These empirical scaling factors can be used to guide the plausible range of changes in the climate variables in impact adaption vulnerability assessment under climate change. For example, in hydrological and related modelling, the observed historical data is perturbed or scaled by these empirical factors to reflect a future climate. For example, the daily precipitation time series can be scaled by either the monthly, seasonal or annual scaling factors. Where seasonal scaling factors are used, the generated future daily precipitation series should be rescaled to reflect/match the annual scaling factor. Likewise, when monthly scaling factors are used, the future precipitation time series should be rescaled to match the seasonal scaling factors, and then rescaled again to reflect the annual scaling factor. In most applications in integrated basin analysis, impacts and outcomes to different sectors in water uses are influenced by different climate variables. In these applications, it is best to run simulations guided by all the available GCMs to estimate the likely range (and uncertainty) in the future outcomes, using the internally consistent changes (empirical scaling factors) in the different variables of each GCM. Where resources allow only limited modelling runs, empirical scaling factors from the median projection and the 10 th and 90 th percentile can be used to reflect the extreme range of changes. Here, the median and 10 th and 90 th percentile GCM should be chosen based on the key climate variable impacting the objective of the modelling, for example, based on precipitation where water supply and security is the main consideration. Database of CMIP5 climate change projections across south Asia 19

28 References Chen Y, Singh R and Liu R, Gridded climate datasets for hydrological modelling across South Asia. Sustainable Development Investment Portfolio. CSIRO Land and Water, Canberra. Charles S., Climate change and water in south Asia overview and literature review. Sustainable Development Investment Portfolio. CSIRO Land and Water, Perth. Chiew FHS, Teng J, Vaze J, Post DA, Perraud J M, Kirono DGC and Viney NR, 2009a. Estimating climate change impact on runoff across south east Australia: method, results and implications of modelling method. Water Resources Research, 45, W10414, doi: /2008wr Chiew FHS, Teng J, Vaze J and Kirono DGC, 2009b. Influence of global climate model selection on runoff impact assessment. Journal of Hydrology, 379, , doi: /j.jhydrol IPCC, Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Stocker, T.F., D. Qin, G. K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex and P.M. Midgley (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 1535 pp Meinshausen M, Smith SJ, Calvin K, Daniel JS, Kainuma MLT, Lamarque JF, Matsumoto K, Montzka SA, Raper SCB, Riahi K, Thomson A, Velders GJM, van Vuuren DPP, The RCP greenhouse gas concentrations and their extensions from 1765 to Climatic Change, 109, Morton FI, Operational estimates of areal evapotranspiration and their significance to the science and practice of hydrology. Journal of Hydrology 66, Moss RH, Edmonds JA, Hibbard KA, Manning MR, Rose SK, van Vuuren DP, Carter TR, Emori S, Kainurma M, Kram T, Meehl GA, Mitchell JFB, Nakicenovic N, Riahi K, Smith SJ, Stouffer RJ, Thomson AM, Weyant JP, Wibanks TJ, The next generation of scenarios for climate change research and assessment. Nature, 463, Sheffield J, Goteti G, Wood EF, Development of a 50 yr high resolution global dataset of meteorological forcings for land surface modeling. Journal of Climate 19 (13), Sridhar DSPL, Rajeevan M, Sreejith OP, Satbhai NS, Mukhopadyay B, Development of a new high spatial resolution ( ) Long Period ( ) daily gridded precipitation data set over India and its comparison with existing data sets over the region. MAUSAM 67, Taylor KE, Balaji V, Hankin S, Juckes M and Lawrence B, CMIP5 and AR5 Data Reference Syntax (DRS) Version 0.25, < pcmdi.llnl.gov/cmip5/docs/cmip5_data_reference_syntax_v0 25_clean.pdf>. 20 Database of CMIP5 climate change projections across south Asia

29 Yatagai A, Kamiguchi K, Arakawa O, Hamada A, Yasutomi N, Kitoh A, APHRODITE: Constructing a Long Term Daily Gridded Precipitation Dataset for Asia Based on a Dense Network of Rain Gauges. Bulletin of the American Meteorological Society, 93, Database of CMIP5 climate change projections across south Asia 21

30 CONTACT US t e enquiries@csiro.au w FOR FURTHER INFORMATION Land and Water, Water Resources Management Hongxing Zheng t e hongxing.zheng@csiro.au w YOUR CSIRO Australia is founding its future on science and innovation. Its national science agency, CSIRO, is a powerhouse of ideas, technologies and skills for building prosperity, growth, health and sustainability. It serves governments, industries, business and communities across the nation. 22 Database of CMIP5 climate change projections across south Asia