doi: 10.1038/s41467-019-09776-9.
Enhanced future changes in wet and dry extremes over Africa at convection-permitting scale
Affiliations
- PMID:31015416
- PMCID: PMC6478940
- DOI: 10.1038/s41467-019-09776-9
Item in Clipboard
Enhanced future changes in wet and dry extremes over Africa at convection-permitting scale
Elizabeth J Kendon et al. Nat Commun..
Display options
Format
Display options
Format
Abstract
African society is particularly vulnerable to climate change. The representation of convection in climate models has so far restricted our ability to accurately simulate African weather extremes, limiting climate change predictions. Here we show results from climate change experiments with a convection-permitting (4.5 km grid-spacing) model, for the first time over an Africa-wide domain (CP4A). The model realistically captures hourly rainfall characteristics, unlike coarser resolution models. CP4A shows greater future increases in extreme 3-hourly precipitation compared to a convection-parameterised 25 km model (R25). CP4A also shows future increases in dry spell length during the wet season over western and central Africa, weaker or not apparent in R25. These differences relate to the more realistic representation of convection in CP4A, and its response to increasing atmospheric moisture and stability. We conclude that, with the more accurate representation of convection, projected changes in both wet and dry extremes over Africa may be more severe.
Conflict of interest statement
The authors declare no competing interests.
Figures
Wet season index. Three-month period with the highest mean precipitation ina,d TRMM, CMORPH andb,e R25 model, for present day and future, andc,f CP4A, for present day and future. The black lines indicate the 500 -, 1000-, 2000-, 3000- and 4000 -m height contours. For Figs. 2–6, 10 and Supplementary Figures 11–16, wet season as observed in TRMM is used as a common definition for all datasets
Wet season mean precipitation.a TRMM observations, differences with respect to TRMM forb the R25 model,c CP4A model andd CMORPH observations, and percentage differences between 2100 and present day fore the R25 model andf CP4A model. The median of future percentage changes across Africa (land points only) is indicated ine,f. Dataset differences and future changes are masked in white, where differences are not significant at the 5% level compared with year-to-year variability. The wet season is the 3-month period with the highest mean precipitation in TRMM, defined on a grid-point basis. The black lines indicate the 500 -, 1000 -, 2000-, 3000 - and 4000 -m height contours
Wet season 3-hourly precipitation occurrence.a TRMM observations, differences with respect to TRMM forb the R25 model,c CP4A model andd CMORPH observations, and percentage differences between 2100 and present day fore the R25 model andf CP4A model. Precipitation occurrence is defined as the frequency of wet values (>0.1 mm h−1). The median of future percentage changes across Africa (land points only) is indicated ine,f. Dataset differences and future changes are masked in white, where differences are not significant at the 5% level compared with year-to-year variability. The wet season is the 3-month period with the highest mean precipitation in TRMM, defined on a grid-point basis. The black lines indicate the 500 -, 1000 -, 2000-, 3000 - and 4000 -m height contours
Wet season mean 3-hourly precipitation intensity.a TRMM observations, differences with respect to TRMM forb the R25 model,c CP4A model andd CMORPH observations, and percentage differences between 2100 and present day fore the R25 model andf CP4A model. Mean precipitation intensity is defined as the mean of wet values (>0.1 mm h−1). The median of future percentage changes across Africa (land points only) is indicated ine,f. Dataset differences and future changes are masked in white, where differences are not significant at the 5% level compared to year-to-year variability. The wet season is the 3-month period with the highest mean precipitation in TRMM, defined on a grid-point basis. The black lines indicate the 500 -, 1000 -, 2000-, 3000 - and 4000 -m height contours
Wet season extreme precipitation intensity.a TRMM observations, differences with respect to TRMM forb the R25 model,c CP4A model andd CMORPH observations, and percentage differences between 2100 and present day fore the R25 model andf CP4A model. Extreme precipitation intensity is defined as the 99th percentile of wet values (>0.1 mm h−1), for 3-hourly precipitation. The median of future percentage changes across Africa (land points only) is indicated ine,f. Dataset differences and future changes are masked in white, where differences are not significant at the 5% level compared with year-to-year variability. The wet season is the 3-month period with the highest mean precipitation in TRMM, defined on a grid-point basis. The black lines indicate the 500 -, 1000 -, 2000-, 3000 - and 4000 -m height contours
Present-day extreme precipitation and the frequency of exceedance in future, for the wet season. Present-day extreme precipitation threshold ina TRMM observations, differences with respect to TRMM forb the R25 model,c CP4A model andd CMORPH observations and the ratio of the future compared with the present-day frequency of exceedance of this threshold fore the R25 model andf CP4A model. Extreme precipitation threshold is defined as the 99.9th percentile of 3-hourly precipitation in the wet season in the present-day. The median of future/present exceedance ratio across Africa (land points only) is indicated ine andf. Differences and future changes are masked in white, where differences are not significant at the 5% level compared with year-to-year variability. For all datasets, the wet season is the 3-month period with the highest mean precipitation in TRMM, defined on a grid-point basis. The black lines indicate the 500 -, 1000 -, 2000-, 3000 - and 4000 -m height contours. Definition of Africa sub-regions for subsequent analysis is shown ina
Fractional contribution of 3-hourly precipitation intensity bins to the total precipitation.a,b Fractional contribution (%) for present day for TRMM and CMORPH observations and R25 and CP4A models;c,d difference in fractional contribution between 2100 and present day for R25 and CP4A models for Sahel July–August–September (JAS) and E-Africa March–April–May (MAM). For CMORPH, the original version 1 of the data (CMORPH-v1, grey dashed) is shown as well as the bias-corrected data (CMORPH, grey solid). Regions are as shown in Fig. 6. All 3-hourly data in the given season, in the 10-year period, from all land points in the sub-region are used to calculate the fractional contribution. Stars indicate where future changes are significant at the 1% level compared with year-to-year variability, assessed using bootstrap resampling
Probability distribution of dry spell duration.a,b Present-day duration of dry spells for TRMM and CMORPH observations and R25 and CP4A models;c,d differences in dry spell distribution between 2100 and present day for R25 and CP4A models for Sahel July–August–September (JAS) and Gulf-of-Guinea April–May–June (AMJ). Dry spells are defined as days with <1 mm of rainfall. For CMORPH, the original version 1 of the data (CMORPH-v1, grey dashed) is shown as well as the bias-corrected data (CMORPH, grey solid). Regions are as shown in Fig. 6. All daily data in the given season, in the 10-year period, from all land points in the sub-region are used to calculate the distribution. Stars indicate where future changes are significant at the 1% level compared with year-to-year variability, assessed using bootstrap resampling
Hovmöller plots of 3-hourly precipitation. Precipitation (mm h−1) averaged over the latitude band 5–15° N for July–August–September (JAS) 2001 for TRMM, R25 and CP4A and for the equivalent season in the CP4A future simulation
Scaling between future changes in extreme 3-hourly precipitation intensity and the dew point temperature.a,b Scaling coefficient given by future change in logarithm of extreme precipitation intensity divided by future change in mean dew point temperature, for the wet season, for CP4A and R25. The median scaling across Africa is indicated in the panel titles; colours correspond with the scaling coefficient divided by the Clausius–Clapeyron relationship of 6.2%K−1 (such that a value of 1 corresponds to CC scaling).c,d Joint probability distribution of change in logarithm of extreme precipitation intensity versus change in mean dew point temperatureTd, for the wet season, across Africa, for CP4A and R25. Cyan lines show the relationship for 0.5, 1 and 2 times CC-scaling; red lines show the average relationship obtained from fitting a Lowess regression line. Extreme precipitation intensity is defined as the 95th percentile of wet values (>0.1 mm h−1), for daily maximum 3-hourly precipitation, and is set to missing (and masked in grey) over sea points and where less than 5% of the data is wet. The wet season is the 3-month period with the highest mean precipitation in TRMM, defined on a grid-point basis
References
- Christensen, J. H. et al. Chapter 11: Regional climate projections. In: Solomon, et al. (eds.)Climate change 2007: The physical science basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. (Cambridge University Press, United Kingdom and New York, NY, USA. 46pp 2007).
- Dosio A, Panitz HJ. Climate change projections for CORDEX-Africa with COSMO-CLM regional climate model and differences with the driving global climate models. Clim. Dyn. 2016;46:1599–1625. doi: 10.1007/s00382-015-2664-4. - DOI
- Martin GM, Klingaman NP, Moise AF. Connecting spatial and temporal scales of tropical precipitation in observations and the MetUM-GA6. Geosci. Model Devel. 2017;10:105–126. doi: 10.5194/gmd-10-105-2017. - DOI
- Stratton R. A., Stirling A. J. Improving the diurnal cycle of convection in GCMs. Quarterly Journal of the Royal Meteorological Society. 2011;138(666):1121–1134. doi: 10.1002/qj.991. - DOI
- Stephens GL, et al. Dreary state of precipitation in global models. J. Geophys. Res. 2010;115:D24211,. doi: 10.1029/2010JD014532. - DOI
Publication types
LinkOut - more resources
Full Text Sources