Performance of three Selected Convective Schemes for Predicting Indian Summer Monsoon Rainfall using RegCM4.4 A.K.M. Saiful Islam Professor Institute of Water and Flood Management Bangladesh University of Engineering and Technology
Monsoon and Bangladesh Monsoon plays an important role to our economy especially to the agriculture. 80% of the rainfall occurred during the monsoon seasons (June-September). However, excess rainfall during the monsoon causes moderate to severe flooding to the country. Therefore prediction of monsoon using numerical climate or weather modeling is vital for the country.
Indian Summer Monsoon (Southwest monsoon) The southwestern summer monsoons occur from July through September. The Thar Desert and adjoining areas of the northern and central Indian subcontinent heats up considerably during the hot summers. This causes a low pressure area over the northern and central Indian subcontinent. To fill this void, the moisture-laden winds from the Indian Ocean rush in to the subcontinent. These winds, rich in moisture, are drawn towards the Himalayas. The Himalayas act like a high wall, blocking the winds from passing into Central Asia, and forcing them to rise. As the clouds rise their temperature drops and precipitation occurs. Some areas of the subcontinent receive up to 10,000 mm (390 in) of rain annually. Onset dates and prevailing wind currents of the southwest summer monsoons in India.
South Asian Monsoon circulations Winter Summer Source: Meteorology Today
Climate Change effects Monsoon? IPCC s AR5 (WG-I) Overall, monsoonal rainfall is projected to become more intense in future, and to affect larger areas, because atmospheric moisture content increases with temperature. However, the localized effects of climate change on regional monsoon strength and variability are complex and more uncertain
Bengal Delta: Ganges, Brahmaputra and Meghna (GBM) Basins Bangladesh is a delta formed by the three major rivers, namely the Ganges, Brahmaputra and Meghna with a total area of just over 1.7 million km 2, distributed between India (64%), China (18%), Nepal (9%), Bangladesh (7%) and Bhutan (3%)
Floods in Bangladesh Floods is a regular phenomenon for Bangladesh. Every year one-third areas of Bangladesh are flooded by a water carried from GBM basins during monsoon seasons. Major floods occurred in 1988, 1998, 2004 and 2007.
Flood Hydrographs of major rivers in Bangladesh Delta and Climate Observatory
Flood Forecasting and Warning Center (FFWC) of Bangladesh Flood Warning Map of Today : 10 July 2014
Downscaling regional climate information for impact assessment studies -> REGCM Source: Giorgi et al. 2014 Global model (AOGCM) Time-slice AGCM, VARGCM Regional Model (RCM) Statistical Downscaling Impacts Storms Flood Drought Water Resources Energy Agriculture Landuse Change Pollution Health Fisheries Ecosystems
Nested Regional Climate Modeling: Technique and Strategy Motivation: The resolution of GCMs is still too coarse to capture regional and local climate processes Technique:A Regional Climate Model (RCM) is nested within a GCM in order to locally increase the model resolution. Initial conditions (IC) and lateral boundary conditions (LBC) for the RCM are obtained from the GCM ( One-way Nesting ) or analyses of observations (perfect LBC). Strategy: The GCM simulates the response of the general circulation to the large scale forcings, the RCM simulates the effect of sub- GCM-grid scale forcings and provides fine scale regional information Technique borrowed from NWP Source: Giorgi et al. 2014
RCM Nesting procedure Initial and lateral Boundary conditions can be from analyses of observations or from GCMs Inner Domain Fine Scale Buffer Zone (LBC Relaxation) Large Scale Forcing Fields Source: Giorgi et al. 2014
Modeling of Climate Systems
3D Climate Models
The ICTP regional climate model system RegCM4 (Giorgi et al. 2012, CR SI 2012) Source: Giorgi et al. 2014 Dynamics: Hydrostatic (Giorgi et al. 1993a,b) Non-hydrostatic in progress Radiation: CCM3 (Kiehl 1996) NNRD (Solmon) Large-Scale Precipitaion: SUBEX (Pal et al 2000) Explicit microphysics (Nogherotto) Cumulus convection: Grell (1993) Anthes-Kuo (1977) MIT (Emanuel 1991) Mixed convection Tiedtke Planetary boundary layer: Modified Holtslag, Holtslag (1990) UW-PBL (O Brien et al. 2011) Land Surface: BATS (Dickinson et al 1993) SUB-BATS (Giorgi et al 2003) CLM3.5 (Steiner et al. 2009) CLM4.5 (Oleson et al. 2012) Ocean Fluxes BATS (Dickinson et al 1993) Zeng (Zeng et al. 1998) Diurnal SST Configuration Adaptable to any region Tropical belt configuration
Source: Giorgi et al. 2014 The ICTP regional climate model system RegCM4, coupled components Coupled ocean MIT ocean model (Artale et al. 2010) ROMS (Ratnam et al. 2009) Interactive lake 1D thermal lake mode (Hostetler et al. 1994; Small et al. 1999) Interactive biosphere Available in CLM but never tested Interactive hydrology CHYM hydrological model available in off line mode Aerosols: OC-BC-SO4 (Solmon et al 2005) Dust (Zakey et al 2006) Sea Salt (Zakey et al. 2009) Gas phase chemistry: Various schemes and solvers tested CBMZ + Sillmann solver implemented (Shalaby et al. 2012)
Objectives Predicting monsoon using climate model is a challenging tasks. The major issue of capturing South Asian Monsoon using Regional Climate Model, RegCM4, lies on the selecting appropriate convictive Schemes. Therefore, this study is focusing on finding performances on three convective schemes GRELL, EMANUEL and TIEDTKE. We also investigate performance of Land Surface Scheme CLM4.5 with widely used BATS scheme as Land Surface Scheme has providing feedbacks to the convective schemes.
Regional Climate Modeling Control Experiments using RegCM 4.4 Model Domain: South Asia Domain (60 E to 100 E and 0N to 40 N) Boundary Conditions: Era Interim simulation Horizontal Resolution: ~50 km Vertical Resolutions: 18 levels in sigma-p coordinates Integration: 1997 1998 Analysis Period: Monsoon Season (June-September)
Modeling Schemes and Parameters Radiation Scheme: NCAR CCM3 Convective Schemes: 1. Grell over land and Emanuel scheme over ocean; 2. TDK over land and ocean; 3. Grell over land and TDK scheme over ocean; Land surface: 1. BATS and 2.CLM4.5 Boundary layer: UW PBL scheme Large-scale precipitation: SUBEX scheme Ocean Flux scehme : Zeng Time Step : 75 seconds
RegCM4.4 Experiments Sl. Physical Schemes LS 1 Grell (land) & Emanuel (ocean) CLM4.5 2. Grell (land) & Emanuel (ocean) BATS 3. Grell (land) & Tiedtke (ocean) CLM4.5 4. Tiedtke (land) & Tiedtke (ocean) CLM4.5 5. Tiedtke (land) & Tiedtke (ocean) BATS
Mean Seasonal precipitation (mm/d) Grell (L) Emnuel (O) Grell (L) Tiedtke (O) Tiedtke CLM4.5 Grell (L) Emnuel (O) Tiedtke CRU BATS
Precipitation Bias (RegCM-CRU) (mm/d) Grell (L) Emnuel (O) Grell (L) Tiedtke (O) Tiedtke CLM4.5 Grell (L) Emnuel (O) Tiedtke BATS
Conclusions Tiedtke is very sensitive to model domain comparing to other schemes. BATS produces dry biases comparing to CLM4.5 We found Grell over land and Tiedtke over ocean shows good performance though it needs more investigation. Future plans- Comparing model runs with other Gridded observed data sets (APHRODITE, GPCP, TRMM) Experiments will be made for other recent flood years (e.g. 2004, 2007). Use of statistical indicators (RMSE, SD, R 2 ) to assess relative performance of various convective schemes.