Professional Documents
Culture Documents
Asl 466 Archive
Asl 466 Archive
1 61
2 62
3 Do CMIP5 simulations of Indian summer monsoon 63
4 64
5 rainfall differ from those of CMIP3? 65
6 66
7 K. Shashikanth,1,2 Kaustubh Salvi,1 Subimal Ghosh1,3* and K. Rajendran4 67
AQ1 8
1 Department • of Civil Engineering, Indian Institute of Technology Bombay, Mumbai 400076, India 68
2 Department of Civil Engineering, University College of Engineering, Osmania University, Hyderabad 500007, India
9 3 Interdisciplinary Programme in Climate Studies, Indian Institute of Technology Bombay, Mumbai 400076, India
69
10 4 CSIR Fourth Paradigm Institute (CSIR-4PI), Bangalore 560037, India 70
11 71
12 72
13 *Correspondence to: Abstract 73
14 Subimal Ghosh, Interdisciplinary 74
To understand the improvements in the simulations of Indian summer monsoon rainfall
Programme in Climate Studies,
15 Indian Institute of Technology (ISMR) by Coupled Model Intercomparison Project 5 (CMIP5) over CMIP3, a comparative 75
16 Bombay, Mumbai 400076, India. study is performed with the original and statistically downscaled outputs of five General 76
17 E-mail: subimal@civil.iitb.ac.in Circulation Models (GCMs). We observe that multi-model average of original CMIP5 77
18 simulations do not show visible improvements in bias, over CMIP3. We also observe that 78
19 CMIP5 original simulations have more multi-model uncertainty than those of CMIP3. The 79
20 statistically downscaled simulations show similar results in terms of bias; however, the 80
Received: 6 June 2013
uncertainty in CMIP5 downscaled rainfall projections is lower than that of CMIP3.
21 Revised: 26 August 2013 81
22 Accepted: 4 September 2013
Keywords: Indian summer monsoon rainfall; statistical downscaling; CMIP
82
23 83
24 84
25 85
26 86
27 1. Introduction and adaptation group for making policies to climate 87
28 change. 88
29 Indian summer monsoon rainfall (ISMR) provides Projections of ISMR with regional climate models 89
30 more than 80% of the total annual rainfall in the have been performed by various researchers mostly 90
31 country and directly relates to water resources, agri- with the use of dynamic downscaling. They include 91
32 culture, ecosystem, health and food security (Webster ‘PRECIS’ (Rupa kumar et al., 2006; Krishna kumar 92
33 et al., 1998; Turner and Annamalai, 2012). Climate et al., 2011), ‘RegCM3’ (Moetasim et al., 2009), and 93
34 change may have significant impacts on ISMR the global atmosphere–ocean model ECHAM5 as the 94
35 (June to September average rainfall), which will in driving model and the ‘COSMO-CLM’ as the regional 95
36 turn affect different sectors, and hence assessing model (Dobler and Ahrens, 2011). High-resolution 96
37 the impacts of climate change on monsoon rainfall Meteorological Research Institute (MRI) model of 97
38 is a high-priority task. Understanding the possible 20 km resolution has also been used by Rajendran 98
39 changes in Indian monsoon rainfall due to global and Kitoh (2008) for ISMR simulations. Use of 99
40 warming has already been started since 2007 (Rupa statistical downscaling for entire country has also been 100
41 Kumar et al., 2006; Kripalani et al., 2007; Moetasim performed recently by Salvi et al., (2013); however, 101
42 et al., 2009; Krishna Kumar et al., 2011; Salvi et al., all these studies have been performed with CMIP3 102
43 2013) with Coupled Model Intercomparison Project simulations. Majority of them show moderate increase 103
44 3 (CMIP3) simulations. With the recent release of in ISMR, except Moetasim et al. (2009), which shows 104
45 CMIP5 (Taylor et al., 2012) simulations, which suppression of monsoon. It will be important to 105
46 have undergone several improvements in terms of understand whether the use of CMIP5 (most recent 106
47 physics and resolution, it is important to assess their generation climate models) will result improvements 107
48 performances with respect to CMIP3 simulations. in models, or change in future projections or decrease 108
49 Comparison of future projections between CMIP3 in uncertainties. This will provide information to 109
50 and CMIP5 (Knutti and Sedláček, 2012; Monerie the policy makers on the requirements of change in 110
51 et al., 2012) is specifically important for those sectors, planning, with the change in model generations. 111
52 where the policy makers have started developing Here, we first compare the bias present in the ISMR 112
53 adaptation options based on assessed impacts of simulations of CMIP3 and CMIP5 models for both 113
54 climate change with CMIP3 simulations (IPCC, coarse resolution and downscaled runs. Subsequently, 114
55 2007). Any major changes in the projections, with the uncertainties associated with the simulations and 115
56 CMIP5 models, with respect to CMIP3 simulations, projections (Mujumdar and Ghosh, 2008; Raje and 116
57 will change the possible impacts and subsequently Mujumdar, •2009; Ghosh, 2010; Schiermeier, 2010) 117 AQ2
58 the adaptation options. Furthermore, it is also impor- are being assessed for both CMIP3 and CMIP5 runs. 118
59 tant to assess the uncertainties associated with the Downscaling is performed with kernel regression- 119
60 projections which provide confidence to the impacts based statistical methods coupled with classification 120
1 61
2 62
3 63
4 64
5 65
6 66
7 67
8 68
9 69
10 70
11 71
12 72
13 73
14 74
15 75
16 76
17 77
18 78
19 79
20 80
21 81
22 82
23 83
24 84
25 85
26 86
27 87
28 88
29 89
30 90
31 91
32 92
33 93
34 94
35 95
36 96
37 97
38 98
39 99
40 100
41 101
42 102
43 103
44 104
45 105
46 106
47
AQ3 Figure 1. •Multi-model average of precipitation (during Indian summer monsoon months, June, July, August and September) 107
48 simulated by CMIP3 and CMIP5 models. Both CMIP3 (b) and CMIP5 (c) multi-model averaged mean monsoon simulations fail 108
49 to produce the spatial variability present in the observed data (a), which is mostly at western coast and north east India due to 109
orographic impacts. Low-spatial resolutions of GCMs are the possible reasons. Similar performances are also observed in terms
50 of standard deviation when CMIP3 (e) and CMIP5 (f) simulations are compared with observed (d). The scatter plots of mean (g)
110
51 and standard deviation (h) between CMIP3 and CMIP5 simulations show no significant changes with new generation models. 111
52 112
53 113
54 corresponds to each grid point), also show similar regression-based downscaling approach (Kannan and 114
55 performances, though the simulations of CMIP5 have Ghosh, 2013; Salvi et al., 2013) for the same. The 115
56 slightly higher values than CMIP3. simulated results, for both CMIP3 and CMIP5, show 116
57 Poor simulations of orographic impacts and spa- resemblances in spatial patterns of mean (Figure 2(a) 117
58 tial variability of Indian rainfall, by coarse resolution and (b)) and standard deviation (Figure 2(d) and 118
59 GCMs, motivates us to use statistical downscaling, for (e)) with respect to the observed data, resulting low 119
60 finer resolution ISMR projections. We apply kernel bias. The errors in mean (Figure 2(c)) and standard 120
1 61
2 62
3 63
4 64
5 65
6 66
7 67
8 68
9 69
10 70
11 71
12 72
13 73
14 74
15 75
16 76
17 77
18 78
19 79
20 80
21 81
22 82
Figure 2. Multi-model average of downscaled Indian summer monsoon rainfall. Both CMIP3 and CMIP5 statistically downscaled
23 simulations capture the orographic impacts on the spatial variability of ISMR in mean (a, b) and standard deviation (d, e) 83
24 respectively. The scatter plots of errors in mean (c) and standard deviation (f) for CMIP3 vs CMIP5 show no improvements in 84
25 terms of bias in CMIP5 simulations over CMIP3. However, downscaling reduces the error in the rainfall simulations because of 85
26 bias correction and kernel regression-based training method with observed data. The predictors used in kernel regression-based 86
27 statistical downscaling are U wind, V wind, temperature both at surface and 500 hPa, specific humidity at 500 hPa and MSLP. 87
28 88
29 deviation (Figure 2(e)) for CMIP3 and CMIP5 are pre- Woldemeskel et al., 2012). Here, we compare the 89
30 sented in scatter plots for comparison. It is true that the uncertainties present in CMIP3 and CMIP5 model sim- 90
31 errors have been reduced significantly after downscal- ulations of ISMR. Uncertainty across the models has 91
32 ing (if compared with original simulations), however, recently been quantified by a metric known as, square 92
33 significant improvements in CMIP5 downscaled simu- root error variance by Woldemeskel et al. (2012), 93
34 lations over CMIP3 are not really been observed. The however, this is scale dependent. To make the skill 94
35 error plot for mean projections of rainfall for indi- score scale free, here we use coefficient of variation 95
36 vidual GCMs for both CMIP3 and CMIP5 shows no (CV) which has also been recommended by Watan- 96
37 significant differences for testing period 1981–2000 abe et al. (2012). CV is the ratio of standard deviation 97
38 (Figures S2 and S3). to mean and hence it is dimensionless. Higher CV 98
39 To understand the possible changes in future, we demotes high uncertainty across models. The uncer- 99
40 consider the worst case scenarios A2 for CMIP3 tainty estimates for both original and downscaled 100
41 and RCP 85 for CMIP5 models. The coarse reso- projections for CMIP3 and CMIP5 are presented in 101
42 lution original projections by CMIP3 (Figure 3(a)) Figure 4. The uncertainties for CMIP5 simulations 102
43 and CMIP5 (Figure 3(b)) models show spatially uni- (Figure 4(b)) are observed to be higher than those 103
44 form increasing changes which are also similar to of CMIP3 (Figure 4(a)). However, the downscaled 104
45 each other. The downscaled projections for future, products show completely different results. The uncer- 105
46 with CMIP3 (Figure 3(c)) and CMIP5 (Figure 3(d)) tainties in CMIP5 downscaled outputs (Figure 4(e)) 106
47 models, show spatially non-uniform changes, where are significantly less than CMIP3 simulations (Figure 107
48 the impacts of orography due to Western Ghats and 4(d)). This is also tested with other uncertainty estima- 108
49 Himalayas are distinctly visible (Appendix S1). The tor, where we have considered the differences between 109
50 west coast and the north east India are expected to maximum and minimum changes (Figure S6). Except 110
51 receive increased rainfall during monsoon. The spatial few small patches, the uncertainties in downscaled 111
52 patterns of changes simulated by downscaled CMIP5 ISMR simulations have been reduced by CMIP5 mod- 112
53 are observed to be similar to downscaled CMIP3. The els (Figure 4(f)). This is probably because of similar 113
54 future projected changes in mean rainfall for individ- simulations of most of the predictors by CMIP5 mod- 114
55 ual GCMs with CMIP3 and CMIP5 simulations show els. For further analysis, we have computed the uncer- 115
56 almost similar results except a few locations (Figures tainties of CMIP3 and CMIP5 predictors (Figure S7). 116
57 S4 and S5). We observed that there is a mixed result, though major- 117
58 Climatic projections are characterized by the uncer- ity of the predictors from CMIP5 has lower uncertainty 118
59 tainty associated with the use of multiple models compared to CMIP3, which makes the downscaling 119
60 (Ghosh and Mujumdar, 2009; Watanabe et al., 2012; algorithms, suitable for uncertainty reduction. It is 120
1 61
2 62
3 63
4 64
5 65
6 66
7 67
8 68
9 69
10 70
11 71
12 72
13 73
14 74
15 75
16 76
17 77
18 78
19 79
20 80
21 81
22 82
23 83
24 84
25 85
26 86
27 Figure 3. Future changes simulated by original and downscaled GCM simulations of ISMR. (a, b) Changes simulated by multi-model 87
average of original GCM outputs of CMIP3 and CMIP5, respectively. The changes are spatially uniform and increasing. However,
28 the downscaling products of CMIP3 (c) and CMIP5 (d) show spatially heterogeneous changes and the impacts of orography are
88
29 clearly visible. There is no significant difference between CMIP3 and CMIP5 simulated changes. 89
30 90
31 91
32 true that the quantified uncertainty resulting from a ISMR, raise the points against the need of using 92
33 downscaling method depends on the selection predic- the new generation simulations. This is addressed 93
34 tors and the uncertainties present in their simulations. with the statistical downscaling approach, where, we 94
35 The present set of predictors, used in downscaling, observe, that downscaled products of CMIP5 models 95
36 simulate ISMR well, with low uncertainty and hence also show similar bias as CMIP3, similar pattern 96
37 recommended for impacts analysis and policy making. of changes but with lower uncertainties. The low 97
38 To understand the impacts of selection of predictors, uncertainty in the changes projected by downscaled 98
39 we perform the same analysis with three combinations CMIP5 simulations will give more confidence to the 99
40 of predictors namely (1) surface variables only (Figure policy makers for adaptation strategies in reducing 100
41 S8), (2) 500 hpa variables with MSLP (Figure S9) and, the risks due to climate change. Here we consider 101
42 (3) all combined (Figure 4). We observed that second five GCMs for comparison of CMIP3 and CMIP5 102
43 and third combinations are most effective in reduc- simulations, which is a small subset of almost 23 103
44 ing the uncertainty of downscaled CMIP5 simulations CMIP3 and 32 CMIP5 models. Hence, conclusion 104
45 (Figure 4 and Figure S9). derived from only five models may not be definitive 105
46 and may change with inclusion of other models. This 106
47 may be considered as the limitation of this study. 107
48 4. Concluding remarks Use of multiple downscaling model may result in 108
49 downscaling uncertainty (Dibike and Coulibaly, 2005; 109
50 Our results highlight that, even with the improvements King et al., 2012), which has not been considered in 110
51 in understanding of climate physics of CMIP5 (Taylor this work, and modeling of that is a potential area of 111
52 et al., 2012) data products, the simulations of ISMR future research. 112
53 with coarse resolution climate models have become 113
54 worse, with similar bias, but with higher uncertainties. 114
55 Probably this points the necessity of fine resolution Supporting information 115
56 models specifically to capture the orographic effects, 116
57 which is responsible for huge spatial variability. The The following supporting information is available: 117
58 fine resolution downscaling model captures the spatial 118
Appendix S1. APHRODITE information.
59 pattern of ISMR, due to orography. Similar bias with 119
60 higher uncertainty in CMIP5 original simulations of Figure S1. Algorithms for statistical downscaling. 120
1 61
2 62
3 63
4 64
5 65
6 66
7 67
8 68
9 69
10 70
11 71
12 72
13 73
14 74
15 75
16 76
17 77
18 78
19 79
20 80
21 81
22 82
23 83
24 84
25 85
26 Figure 4. Uncertainty in the changed projections across the models. The uncertainties are measured as the coefficient of variations 86
27 and are presented spatially for CMIP3 (a) and CMIP5 (b) original GCMs. The differences between them (c) show increase in 87
uncertainty in most of the locations. Similar plots for downscaled projections are presented in (d) and (e) for CMIP3 and CMIP5.
28 For downscaled products uncertainty has been reduced significantly in CMIP5 at almost all locations (f). 88
29 89
30 90
31 Figure S2. Error in mean for CMIP3 simulations of (a) BCCR (CMIP5–152 CMIP3) (c). They show similar results as Fig. 91
(NCC), (b) CCCma, (c) MIROC, (d) MPI and (e) MRI GCMs. 4 (more uncertainty in downscaled CMIP3 153 simulations).
32 92
33 Figure S3. Error in mean for CMIP5 simulations of (a) BCCR 93
34 (NCC), (b) CCCma, (c) MIROC, (d) MPI and (e) MRI GCMs. Acknowledgements 94
35 Figure S4. Projected Changes for A2 (CMIP3) scenario with We thank the World Climate Research Programme’s working 95
36 (a) BCCR (NCC), (b) CCCma, (c) MIROC, (d) MPI and (e) Group on coupled modeling, which is responsible for CMIP, 96
37 MRI GCMs. and the climate modeling for making available their model 97
38 outputs. We would like to thank also APHRODITE, Japan for 98
Figure S5. Projected Changes for RCP85 (CMIP5) scenario making available 25 km resolution observed data.
39 99
with (a) BCCR (NCC), (b) 117 CCCma, (c) MIROC, (D) MPI
40 and (e) MRI GCMs.
100
41 101
42 Figure S6. Uncertainties in CMIP3 and CMIP5 raw [CMIP3 References 102
43 (a), CMIP5 (b) and their differences (c)] and downscaled 103
simulations [CMIP3 (d), CMIP5 (e) and their differences Dibike YB, Coulibaly P. 2005. Hydrologic impact of climate change in
44 the Saguenay watershed: comparison of downscaling methods and 104
(f)] computed with the differences between maximum and
45 hydrologic models. Journal of Hydrology 307: 144–163. 105
minimum changes. They are consistent with the co-efficient
46 of variation plots in Fig. 4.
Dobler A, Ahrens B. 2011. Four climate change scenarios for 106
47 the Indian summer monsoon by the regional climate model 107
Figure S7. Differences (CMIP5–CMIP3) in uncertainty (coef- COSMO-CLM. Journal of Geophysical Research 116: D24104,
48 DOI: 10.1029/2011JD016329.
108
49 ficient of variation across models) between CMIP3 and CMIP5 109
Ghosh S. 2010. SVM-PGSL coupled approach for statistical down-
50 simulations computed for the predictors: (a) humidity, (b) mean scaling to predict rainfall from GCM output. Journal of Geophysical 110
sea level pressure, (c) temperature at 500 hPa, (d) surface air
51 Research 115: D22102, DOI: 10.1029/2009JD013548. 111
temperature, (e) U wind at 500 hPa, (f) surface U wind, (g) V Ghosh S, Mujumdar PP. 2009. Climate change impact assessment:
52 wind at 500 hPa and (h) surface V wind.
112
uncertainty modeling with imprecise probability. Journal of Geo-
53 physical Research 114: D18113, DOI: 10.1029/2008JD011648.
113
54 Figure S8. Uncertainties in CMIP3 (a) and CMIP5 (b) down- IPCC. 2007. Climate Change 2007: The Physical Science Basis. 114
55 scaled simulations in terms of CV plot using the surface level Contribution of Working Group I to the Fourth Assessment Report 115
56 predictors and their differences (CMIP5–CMIP3) (c). They of the Intergovernmental Panel on Climate Change, Solomon S, Qin 116
show similar level of uncertainty. D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller
57 117
HL (eds). Cambridge University Press: Cambridge, UK/New York,
58 Figure S9. Uncertainties in CMIP3 (a) and CMIP5 (b) NY.
118
59 downscaled simulations in terms of 151 CV plot using the Johnson F, Sharma A. 2012. A nesting model for bias correction 119
60 500 hpa level predictors with MSLP and their differences of variability at multiple time scales in general circulation model 120
1 precipitation simulations. Water Resources Research 48(1): 1–16, Mujumdar PP, Ghosh S. 2008. Modeling GCM and Scenario 61
2 DOI: 10.1029/2011WR010464. uncertainty using a possibilistic approach: application to the 62
Kalnay E, Kanamitsu M, Kistler R, Collins W, Deavan D, Gandin L, Mahanadi River, India. Water Resources Research 44: W06407,
3 Iredell M, Saha S, White G, Wollen J, Zhu Y, Chelliah M, Ebisuzaki DOI: 10.1029/2007WR006137.
63
4 W, Higgins W, Janowiak J, Mo KC, Ropelewski C, Wang J, Leetmaa Raje D, Mujumdar PP. 2009. A conditional random field-based 64
5 A, Reynolds R, Jenne R, Joseph D. 1996. The NCEP/NCAR 40- downscaling method for assessment of climate change impact on 65
6 years reanalysis project. Bulletin of the American Meteorological multisite daily precipitation in the Mahanadi basin. Water Resources 66
7 Society 77(3): 437471. Research 45: W10404, DOI: 10.1029/2008WR007487. 67
Kannan S, Ghosh S. 2011. Prediction of daily rainfall state in a river Rajendran K, Kitoh A. 2008. Indian summer monsoon in future climate
8 basin using statistical downscaling from GCM output. Stochastic projection by a super high resolution global model. Current Science
68
9 Environmental Research and Risk Assessment 25: 457–474, DOI: 95(11): 1560–1569. 69
10 10.1007/s00477-010-0415-y. Rupa Kumar KR, Sahai AK, Kumar KK, Patwardhan SK, Mishra PK, 70
11 Kannan S, Ghosh S. 2013. A nonparametric kernel regression model Revadekar JV, Kamala K. 2006. High-resolution climate change 71
12 for downscaling multisite daily precipitation in the Mahanadi basin. scenarios for India for the 21st century. Current Science 90(3). 72
Water Resources Research 49, DOI: 10.1002/wrcr.20118. Salvi K, Kannan S, Ghosh S. 2013. High-resolution multisite
13 King L, Irwin S, Sarwar R, McLeod AI, Simonovic SP. 2012. daily rainfall projections in India with statistical downscaling
73
14 The effects of climate change on extreme precipitation events in for climate change impacts assessment. Journal of Geophysical 74
15 the Upper Thames River Basin: A comparison of downscaling Research – Atmospheres 118, DOI: 10.1002/jgrd.50280. 75
16 approaches. Canadian Water Resources Journal 37(3): 253–274. Schiermeier Q. 2010. The real holes in the climate science. Nature 76
17 Knutti R, Sedláček J. 2012. Robustness and uncertainties in the 463: 284–287. 77
new CMIP5 climate. Nature Climate Change 2: 587–595, DOI: Taylor , Karl E, Ronald J, Stouffer , Gerald A, Meehl . 2012. An
18 10.1038/nclimate1495. overview of CMIP5 and the experiment design. Bulletin of the Amer-
78
19 Kripalani RH, Oh JH, Kulkarni A, Sabade SS, Chaudhari HS. 2007. ican Meteorological Society 93: 485–498, DOI: 10.1175/BAMS-D- 79
20 South Asian summer monsoon precipitation variability: Coupled cli- 11-00094.1. 80
21 mate model simulations and projectiounder IPCC AR4. Theoretical Turner AG, Annamalai H. 2012. Climate change and the South Asian 81
22 and Applied Climatology 90(3): 133–159. summer monsoon, model projections. Nature Climate Change, DOI: 82
Krishna Kumar K, Patwardhan SK, Kulkarni A, Kamala K, Rao KK, 10.1038/nclimate1716.
23 Jones R. 2011. Simulated projections for summer monsoon climate Watanabe S, Kanae S, Seto S, Yeh PJ-F, Hirabayashi Y, Oki T. 2012.
83
24 over India by a high-resolution regional climate model ( PRECIS ). Intercomparison of bias-correction methods for monthly temperature 84
25 Current Science 101(3). and precipitation simulated by multiple climate models. Journal of 85
26 Li H, Sheffield J, Eric FW. 2010. Bias correction of monthly pre- Geophysical Research 117: D23114, DOI: 10.1029/2012JD018192. 86
27 cipitation and temperature fields from Intergovernmental Panel on Webster PJ, Magana VO, Palmer TN, Shukla J, Tomas RA, Yanai 87
Climate Change AR4 models using equidistant quantile match- M, Yasunari T. 1998. Monsoons: processes, predictability and the
28 ing. Journal of Geophysical Research, DOI: 10.1029/2009JD0 prospects for prediction. Journal of Geophysical Research 103:
88
29 12882. 14451–14510. 89
30 Mehrotra R, Sharma A. 2005. A nonparametric non homogeneous Wilby, RL, SP Charles, E Zorita, B Timbal, P Whetton, LO 90
31 hidden Markov model for downscaling of multisite daily rainfall Mearns. (2004). The guidelines for use of climate scenarios 91
32 occurrences. Journal of Geophysical Research 110: D16108, DOI: developed from statistical downscaling methods. Supporting 92
AQ4 10.1029/2004JD005677. • material of the Intergovernmental Panel on Climate Change
33 Mehrotra R, Sharma A. 2007. A semi-parametric model for stochastic (IPCC), pre-pared on behalf of Task Group on Data and Sce-
93
34 generation of multi-site daily rainfall exhibiting low-frequency nario Support for Impacts and Climate Analysis (TGICA). 94
35 variability. Journal of Hydrology 335: 180–193. http://ipccddc.cru.uea.ac.uk/guidelines/StatDownGuide.pdf. 95
36 Moetasim A, Shi Y, Tung W, Trapp RJ, Gao X, Pal JS, Diffenbaugh Woldemeskel FM, Sharma A, Sivakumar B, Mehrotra R. 2012. An 96
37 NS. 2009. Suppression of south Asian summer monsoon precipita- error estimation method for precipitation and temperature projections 97
tion in the 21st century. Geophysical Research Letters 36(1), DOI: for future climates. Journal of Geophysical Research 117: D22104,
38 10.1029/2008GL036500. DOI: 10.1029/2012JD018062.
98
39 Monerie P-A, Fontaine B, Roucou P. 2012. Expected future changes Yatagai A, Arakawa O, Kamiguchi K, Kawamoto H, Nodzu MI, 99
40 in the African monsoon between 2030 and 2070 using some CMIP3 Hamada A. 2009. A 44 year daily gridded precipitation data set for 100
41 and CMIP5 models under a medium-low RCP scenario. Journal of Asia based on a dense network of rain gauges. SOLA 5: 137–140, 101
42 Geophysical Research, DOI: 10.1029/2012JD017510. DOI: 10.2151/sola.2009-035. 102
43 103
44 104
45 105
46 106
47 107
48 108
49 109
50 110
51 111
52 112
53 113
54 114
55 115
56 116
57 117
58 118
59 119
60 120
IMPORTANT NOTE: Please mark your corrections and answers to these queries directly onto the proof
at the relevant place. DO NOT mark your corrections on this query sheet.
AQ1. Please check and confirm whether the authors and their respective affiliations are correctly identified.
AQ2. Citation “Mujumdar and Ghosh, 2009” has been changed to ”Mujumdar and Ghosh, 2008” to match
the reference list. Kindly confirm whether the change is fine.
AQ3. These figures (1, 2, 3, 4) are poor quality. Kindly resupply.
AQ4. References “Mehrotra and Sharma (2005)” and “Mehrotra and Sharma (2007)” have been provided in
the list but not cited in the text. So please cite in the text or delete from the list.