Future temperature in southwest Asia projected to exceed a threshold for human adaptability

SUPPLEMENTARY INFORMATION DOI: 10.1038/NCLIMATE2833 Future Temperature in Southwest Asia Projected to Exceed a Threshold for Human Adaptability Future temperature in southwest Asia projected to exceed a threshold for human adaptability Jeremy S Pal 1, 2 and Elfatih A B Eltahir 2 1. Department of Civil Engineering and Environmental Science, Loyola Marymount University, Los Angeles, CA 90045 2. Ralph M. Parsons Laboratory, Massachusetts Institute of Technology, Cambridge, MA 02139. Methods In this study, the Massachusetts Institute of Technology Regional Climate Model (MRCM) is used 1, which is based on the ICTP Regional Climate Model version 3 (RegCM3) 2 with several significant enhancements 3-6. The MRCM grid, centered at 24 N and 47 E on a Lambert Conformal projection, consists of 144 points in the x- direction and 130 points in the y- direction (Figure SI1). The grid cells are separated by 25- km in the horizontal, and 18 sigma levels are prescribed in vertical. Output from three global climate models (GCMs) from the Coupled model Intercomparison Project Phase 5 (CMIP5) database is used as atmospheric boundary conditions for the MRCM integrations 7. Present- day conditions are represented with historical greenhouse gas (GHG) concentrations for the period 1975 through 2005 8. To consider the impacts of climate change, two future GHG scenarios are considered based on the IPCC Representative Concentration Pathway (RCP) trajectories for the period 2070 through 2100: RCP 8.5 and RCP 4.5. RCP 8.5, which represents 8.5 W/m 2 of radiative forcing values in the year 2100 relative to pre- industrial values, is considered a high (or business as usual) GHG concentration scenario that does not consider any mitigation target 9. RCP 4.5, which represents 4.5 W/m 2 of radiative forcing, is considered a mitigation scenario 10. Recently, work has been carried out focusing on improving understanding of the regional climate of Southwest Asia and on improving the skill of MRCM in simulating the key processes in this arid region 11-15. As much of the land surface in Southwest Asia is characterized as desert and semi- desert, it is essential that the soil albedo and NATURE CLIMATE CHANGE www.nature.com/natureclimatechange 1

emissivity be accurately characterized. To correct the high soil albedo bias (0.06 over land) present in the default version of MRCM, albedo is prescribed based on the NASA/GEWEX Surface Radiation Budget (SRB) Project (Figure SI2) 16. In addition, emissivities over desert and semi- desert are reduced from 0.95 to 0.91 and 0.93, respectively, according to the NASA MODIS surface emissivity data 17. These two improvements significantly reduce an overall T and TW cold biases of approximately 1.5 C to less than 0.5 C over Southwest Asia when compared to the European Centre for Medium- Range Weather Forecasts Interim Reanalysis 18 (ERA- Interim) data, except in areas of complex topography (Figure SI3). The model includes a representation for the emission, transport and deposition of mineral aerosols and their direct radiative effects 11,14. Lastly, irrigated crop and marshland land cover types are included to better represent the surface conditions in southern Iraq. In order to objectively compare and select GCMs for use as boundary conditions for MRCM, we apply the following criteria: 1. The GCM provide representations of the Red Sea and Persian Gulf by use of an ocean model of adequate resolution. Each of the GCMs selected represents ocean processes between 0.4 and 1.11 horizontal resolution (Table SI1 and Figure SI4). While these resolutions are less than optimal to simulate some of the key processes in the Red Sea and Persian Gulf, they represent the best available from the CMIP5 archive. 2. The GCM accurately simulate surface T, TW and relative humidity over the Red Sea and Persian Gulf surrounding coastal regions, as well as over all of Southwest Asia. Output from more than 30 GCMs used in CMIP5 are objectively analyzed and compared to both the ERA- Interim 18 and Climate Research Unit 19 datasets. To assess the performance of each GCM, the normalized root mean square error for each variable (T, TW, and relative humidity) is averaged separately over each region (Persian Gulf, Red Sea, and Arabian Peninsula). As a result of applying the above objective criteria, the three GCMs with the lowest total sum of root mean square errors for each variable and region are selected: Community Climate System Model version 4 (CCSM4) 20, Max- Planck- Institute Earth System Model (MPI- ESM) 21 and Norwegian community Earth System Model (NorESM) 22.

The accuracy of the simulations of future T and TW conditions by MRCM in Southwest Asia reflects to a significant degree the accuracy of the SSTs projected by the coupled GCMs used as boundary conditions. This is particularly true for SSTs in the Persian Gulf and Red Sea. Although the CMIP5 GCMs used in this study are screened to resolve the Persian Gulf and Red Sea, their representations are nevertheless spatially limited, especially CCSM4 and NorESM1 (Table SI1 and Figure SI4). The associated limited representation of the SSTs in the Persian Gulf and Red Sea may pose shortcomings that should be evaluated and addressed in future studies. TW is computed by the formulation developed by Davies- Jones 23. The ERA- Interim reanalysis data are considered the best available combined spatial and temporal representation of observations for the region, and are therefore used for the following bias correction procedure: 1. The maximum 6-hour average TW and T for each day are computed for both the MRCM hourly output and the ERA-Interim Reanalysis 3-hourly 0.75 x 0.75 data, denoted by TW max and T max, respectively. 2. The ERA-Interim T max and TW max data are interpolated from the 0.75 x 0.75 horizontal grid to the 25-km MRCM grid. 3. Consistent MRCM and ERA-Interim climatologies of T max and TW max are computed for each day of the year on the MRCM 25-km grid. 4. The magnitude of the bias for each day of the year is estimated by the difference between 30-day running means of the two climatologies. 5. The daily bias is finally applied to the MRCM daily values of T max and TW max for the present-day and future climates. The corrections are on the order of 1-2 C for TWmax (Figure SI3 and Figure SI5). It assumed that the bias in the present day is the same as the bias in the future, which is commonly done in impacts studies 24,25. The adjusted values are in turn used to compute annual maxima and histograms at each grid point. Since maximum values of TW in the region occur in July, August, and September (JAS), histograms of TWmax and Tmax are computed for this period. The JAS TWmax and Tmax values are additionally sorted to determine the 50 th (median) and 95 th percentile values. The 50 th percentile value is on average exceeded half of the days and

provides a measure of the mean TWmax, while the 95 th percentile is exceeded on average every 4.6 days of the three- month period suggestive of a typical hot TWmax summer event. To verify the quality of the ERA- Interim Reanalysis data, the observed annual maximum TWmax is estimated for six stations from the region: two stations adjacent to the Persian Gulf; two adjacent to the Red Sea; and two in the desert interior (Figure SI6). Present- day TWmax values exceeding 31 C are observed adjacent to the Persian Gulf and the Red Sea, and lower values are observed in the desert interior, both of which are consistent with our simulation results (Figures 1 and 2). Furthermore, an increasing trend significant at the 95% level is observed for each of the stations, also consistent with the simulations. It is important to note that the station data represent point values, while the ERA- Interim reanalysis data represent spatial averages over a large area (1,000s km 2 ). As a result, there may be some inconsistencies between the station data and ERA- Interim data. The latter is the dataset used in the bias correction procedure. Extreme Temperature Annual Tmax increases monotonically in the different locations in the Southwest Asia region (Figure SI8). The severest values of annual Tmax occur in Kuwait and Al Ain, where the 60 C threshold is frequently exceeded several times by the end of the century. In these locations, 50 C events become normal days during JAS as indicated by the 50th percentile Tmax (Figure SI9). While Kuwait is protected from extreme TW conditions due to is geographic position, the annual Tmax approaches 60 C in several years, with 50th and 95th percentile summer events of 50 and 56 C, respectively (Figure SI9). In Doha, where the trends in the annual TWmax values are also substantial, the annual Tmax exceeds 60 C in many years, with respective 50 th and 95 th percentile summer events of 45 and 50 C. On the coast of the Red Sea and in the interior regions of Southwest Asia, Tmax conditions are projected to be less severe but still extreme. For example, in Jeddah and nearby Mecca, annual Tmax in many years exceeds 50 and 55 C, respectively, while maintaining annual TWmax values approaching 33 and 32 C, respectively.

References 1 Im, E.- S., Gianotti, R. L. & Eltahir, E. A. B. Improving the Simulation of the West African Monsoon Using the MIT Regional Climate Model. Journal of Climate 27, 2209-2229, doi:10.1175/jcli- d- 13-00188.1 (2014). 2 Pal, J. S. et al. Regional Climate Modeling for the Developing World: The ICTP RegCM3 and RegCNET. Bulletin of the American Meteorological Society 88, 1395-1409, doi:10.1175/bams- 88-9- 1395 (2007). 3 Winter, J. M., Pal, J. S. & Eltahir, E. A. B. Coupling of Integrated Biosphere Simulator to Regional Climate Model Version 3. Journal of Climate 22, 2743-2757, doi:10.1175/2008jcli2541.1 (2009). 4 Gianotti, R. L. & Eltahir, E. A. B. Regional Climate Modeling over the Maritime Continent. Part I: New Parameterization for Convective Cloud Fraction. Journal of Climate 27, 1488-1503, doi:10.1175/jcli- d- 13-00127.1 (2014). 5 Gianotti, R. L. & Eltahir, E. A. B. Regional Climate Modeling over the Maritime Continent. Part II: New Parameterization for Autoconversion of Convective Rainfall. Journal of Climate 27, 1504-1523, doi:10.1175/jcli- d- 13-00171.1 (2014). 6 Gianotti, R. L., Zhang, D. & Eltahir, E. A. B. Assessment of the Regional Climate Model Version 3 over the Maritime Continent Using Different Cumulus Parameterization and Land Surface Schemes. Journal of Climate 25, 638-656, doi:10.1175/jcli- d- 11-00025.1 (2012). 7 Taylor, K. E., Stouffer, R. J. & Meehl, G. A. An Overview of CMIP5 and the Experiment Design. Bulletin of the American Meteorological Society 93, 485-498, doi:10.1175/bams- d- 11-00094.1 (2012). 8 Meinshausen, M. et al. The RCP greenhouse gas concentrations and their extensions from 1765 to 2300. Climatic Change 109, 213-241, doi:10.1007/s10584-011- 0156- z (2011). 9 Riahi, K. et al. RCP 8.5 A scenario of comparatively high greenhouse gas emissions. Climatic Change 109, 33-57, doi:10.1007/s10584-011- 0149- y (2011). 10 Thomson, A. et al. RCP4.5: a pathway for stabilization of radiative forcing by 2100. Climatic Change 109, 77-94, doi:10.1007/s10584-011- 0151-4 (2011). 11 Marcella, M. P. & Eltahir, E. A. B. Effects of mineral aerosols on the summertime climate of southwest Asia: Incorporating subgrid variability in a dust emission scheme. Journal of Geophysical Research 115, doi:10.1029/2010jd014036 (2010). 12 Marcella, M. P. & Eltahir, E. A. B. Modeling the Hydroclimatology of Kuwait: The Role of Subcloud Evaporation in Semiarid Climates. Journal of Climate 21, 2976-2989, doi:10.1175/2007jcli2123.1 (2008). 13 Marcella, M. P. & Eltahir, E. A. B. The Hydroclimatology of Kuwait: Explaining the Variability of Rainfall at Seasonal and Interannual Time Scales. Journal of Hydrometeorology 9, 1095-1105, doi:10.1175/2008jhm952.1 (2008). 14 Marcella, M. P. & Eltahir, E. A. B. The role of lateral boundary conditions in simulations of mineral aerosols by a regional climate model of Southwest Asia. Climate Dynamics (2011). 15 Marcella, M. P. & Eltahir, E. A. B. Modeling the Summertime Climate of Southwest Asia: The Role of Land Surface Processes in Shaping the Climate of Semiarid Regions. Journal of Climate 25, 704-719, doi:10.1175/2011jcli4080.1 (2011). 16 Stackhouse, J., P. W. et al. The NASA/GEWEX Surface Radiation Budget Release 3.0: 24.5- Year Dataset. GEWEX News 21 (2011).

17 Jin, M. & Liang, S. An Improved Land Surface Emissivity Parameter for Land Surface Models Using Global Remote Sensing Observations. Journal of Climate 19, 2867 2881, doi:http://dx.doi.org/10.1175/jcli3720.1 (2006). 18 Dee, D. P. et al. The ERA- Interim reanalysis: configuration and performance of the data assimilation system. Quarterly Journal of the Royal Meteorological Society 137, 553-597, doi:10.1002/qj.828 (2011). 19 Harris, I., Jones, P. D., Osborn, T. J. & Lister, D. H. Updated high- resolution grids of monthly climatic observations the CRU TS3.10 Dataset. International Journal of Climatology 34, 623-642, doi:10.1002/joc.3711 (2014). 20 Meehl, G. A. et al. Climate System Response to External Forcings and Climate Change Projections in CCSM4. Journal of Climate 25, 3661-3683, doi:10.1175/jcli- d- 11-00240.1 (2012). 21 Giorgetta, M. A. et al. Climate and carbon cycle changes from 1850 to 2100 in MPI- ESM simulations for the Coupled Model Intercomparison Project phase 5. Journal of Advances in Modeling Earth Systems 5, 572-597, doi:10.1002/jame.20038 (2013). 22 Bentsen, M. et al. The Norwegian Earth System Model, NorESM1- M Part 1: Description and basic evaluation of the physical climate. Geoscientific Model Development 6, 687-720, doi:10.5194/gmd- 6-687- 2013 (2013). 23 Davies- Jones, R. An Efficient and Accurate Method for Computing the Wet- Bulb Temperature along Pseudoadiabats. Monthly Weather Review 136, 2764-2785, doi:10.1175/2007mwr2224.1 (2008). 24 White, M. A., Diffenbaugh, N. S., Jones, G. V., Pal, J. S. & Giorgi, F. Extreme heat reduces and shifts United States premium wine production in the 21st century. Proceedings of the National Academy of Sciences of the United States of America 103, 11217-11222, doi:10.1073/pnas.0603230103 (2006). 25 Ashfaq, M., Bowling, L. C., Cherkauer, K., Pal, J. S. & Diffenbaugh, N. S. Influence of climate model biases and daily- scale temperature and precipitation events on hydrological impacts assessment: A case study of the United States. Journal of Geophysical Research 115, doi:10.1029/2009jd012965 (2010).

Table SI1: Horizontal resolutions of the atmospheric and ocean components of the three GCMs used in this study. GCM Atmosphere Ocean CCSM4 0.9 x1.25 1.11 x0.27 0.54 MPI- ESM T63/1.875 0.4 x0.4 NorESM 1.875 x2.5 1 x1

Figure SI1: Map of the 25- km MRCM domain including land cover (colors) and topography (contour lines). The MRCM grid, centered at 24 N and 47 E on a Lambert Conformal projection, consists of 144 points in the x- direction and 130 points in the y- direction.

Figure SI2: Map of annual surface albedo for the period 1984-2007: (a) NASA/GEWEX SRB and (b) MRCM with albedo improvements (land only). Averages for land excluding the buffer zone (LND) and the Arabian Peninsula (AP) are indicated in the bottom right corner of each plot.

Future Temperature in Southwest Asia Projected to Exceed a Threshold for Human Adaptability Figure SI3: JAS 30- year climatology of mean TW for: ERA- Interim (left column), MRCM forced by ERA- Interim data (middle column) and their difference (right column). Data for the difference plots are interpolated to the coarser ERA- Interim grid. Averages for the domain excluding the buffer zone (DOM), land excluding the buffer zone (LND) and the Arabian Peninsula (AP) are indicated in the bottom right corner of each plot.

Figure SI4: Map of JAS SST for each of the three GCMs interpolated onto the MRCM grid for each GHG scenario. The name of the GCM is indicated in the top left of each panel and the GHG concentrations in the top right.

Figure SI5: Simulated bias of the average annual maximum of TWmax for each GCM compared to the ERA- Interim Reanalysis data. The GCM being compared is indicated in the top right for each panel. Average biases for the domain excluding the buffer zone (DOM), land excluding the buffer zone (LND) and the Arabian Peninsula (AP) are indicated in the bottom right corner of each plot. TWmax is the maximum daily value averaged over a 6- hour window.

36# 34# Dhahran# Jeddah# Tabuk# Abu#Dhabi# Jazan# Riyadh# 32# 30# Annual&Maximum&TW max &( o C)& 28# 26# 24# 22# 20# 18# 16# 1980# 1985# 1990# 1995# 2000# 2005# 2010# Year& Figure SI6: Time series of the annual maximum of TWmax for six meteorological stations in the Southwest Asia region: two stations adjacent to the Persian Gulf (Dhahran and Abu Dhabi); two adjacent to the Red Sea (Jeddah and Jazan); and two in the desert interior (Riyadh and Tabuk). Trends for each of the times series are significant at the 95% level according to the Mann- Kendall test (not shown). TWmax is the maximum daily value averaged over a 6- hour window.

Figure SI7: Ensemble average of the JAS 6- hour 30- year mean TW (top row) and T (bottom row) temperatures for each GHG scenario: historical (left column), RCP 4.5 (middle column) and RCP 8.5 (right column). Averages for the domain excluding the buffer zone (DOM), land excluding the buffer zone (LND) and the Arabian Peninsula (AP) are indicated in the bottom right corner of each plot.

Future Temperature in Southwest Asia Projected to Exceed a Threshold for Human Adaptability Figure SI8: Time series of the annual maximum of Tmax for each ensemble member and GHG scenario. Blue, green and red lines represent historical (1976-2005), RCP 4.5 (2071-2100) and RCP 8.5 (2071-2100), respectively. Tmax is the maximum daily value averaged over a 6- hour window. The background image is obtained from NASA Visible Earth.

Future Temperature in Southwest Asia Projected to Exceed a Threshold for Human Adaptability Figure SI9: Histogram of the summer (JAS) of Tmax for each GHG scenario s ensemble: historical (blue), RCP 4.5 (green) and RCP 8.5 (red). The histogram bin interval is 1.0 C and the values on the y- axis indicate the number of exceedances. Values indicated within each plot represent the 50th and 95th percentile event thresholds. Tmax is the maximum daily value averaged over a 6- hour window. The background image is obtained from NASA Visible Earth.