Heterogeneous Trends of Precipitation Extremes in Recent Two Decades over East Africa

近20年东非极端降水变化趋势的空间异质性

+ Author Affiliations + Find other works by these authors
  • Corresponding author: Gensuo JIA, jiong@tea.ac.cn
  • Funds:

    Supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (CAS)—CAS Big Earth Data Science Engineering Program (XDA19030401) and National Key Research and Development Program of China (2016YFA0600303)

  • doi: 10.1007/s13351-021-1028-8

PDF

  • East Africa is so vulnerable to the impacts of precipitation extremes varying from frequent floods to prolonged droughts. However, systematic regional assessment of precipitation extremes across seasons has received little attention, and most previous studies of precipitation extremes have employed many indices and sparse gauge observations giving marginalized details. In this study, we use three precipitation extreme indices to examine the intensity of the highest single-day rainfall record (rx1day), prevalence of very heavy rainfalls (r30mm), and persistence of successive wet days (cwd) at both annual and seasonal scales over recent two decades (1998–2018) based on the Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis data. The results show that the most intensive and frequent precipitation extremes are noticeable from January to May across the areas extending from Madagascar to the Tanzanian coastal zone. These areas also exhibit patches of significant increasing trends in frequency, duration, and intensity of precipitation extremes annually and seasonally. However, significant declines in frequency and intensity of precipitation extremes are observed from western Ethiopia to Congo–Uganda, especially in June–September. The October–December season witnesses higher interannual variability amounting to overall weak and less significant trends. Further subregional assessment shows overall declining intensity and frequency of precipitation extremes in northern part of the study areas. Mean wetness duration increased, with persistence of moderate wet days and slight reduction of severe events. This study unveils higher susceptibility of the East African region to the widely observed hotspots of precipitation extremes posing threats to food security, water resource, and human well-being. The region should consider upscaling irrigation schemes in addition to planning resilient and supportive infrastructures to withstand the upsurging precipitation extremes, especially along the coastal zone.
    从频繁的洪水到长期的干旱,东非极易受到极端降水事件的影响。然而,对东非极端降水事件的已有研究大多采用了稀疏的地面观测资料,且缺乏对季节尺度极端降水事件的系统评估。本研究使用热带降雨测量任务(TRMM)的降水资料研究了近20年年际和季节尺度东非极端降水特征。结果表明,1–5月,马达加斯加到坦桑尼亚沿海地区极端降水频繁,且极端降水的频率、持续时间和强度呈增加趋势。6–9月,埃塞俄比亚西部到刚果–乌干达地区极端降水的频率和强度显著下降。 10–12月,极端降水的年际变率较高,总体趋势不显著。东非地区应考虑扩大灌溉计划,同时规划支持性的基础设施,以抵御频发的极端降水事件,尤其在对极端降水威胁更加敏感的沿海地区。
  • 加载中
  • Fig. 1.  Map of the study domain showing the topographical elevation (m) above the mean sea level, major rivers and lakes, and gauge stations used for data validation. Enclosed is the African continent map showing the location of the study domain (red-colored rectangle).

    Fig. 2.  Monthly rainfall climatology (left panels) and time series of rainfall from gauge stations (red line) and TRMM 3B42 product (blue line) during 1998–2009 (right panels) over (a, b) Sudan, (c, d) Uganda, (e, f) Kenya, and (g, h) Tanzania averaged for all stations over a respective country shown in Fig. 1. The correlation coefficient (r) and its significance (p) between the two datasets are shown on the top right of each panel.

    Fig. 3.  Taylor diagrams showing statistical comparison between the rain gauge observations (reference data; REF point) and TRMM-3B42 satellite-based precipitation product (blue marker) in reproducing the (a) annual rainfall cycle and (b) monthly rainfall evolution over (1) Sudan, (2) Uganda, (3) Kenya, and (4) Tanzania during 1998–2009. The angular axis (dotted line) shows the correlation coefficient (r), the radial axis (dashed concentric arc) shows the standardized deviation (SD), and the solid concentric semicircle denotes the centered root mean square error (RMSE) of TRMM against gauge observations. The gauge reference data point centered at REF is with r = 1, SD = 1, and RMSE = 0; and the closer the other point to it, the higher the agreement between the two datasets.

    Fig. 4.  (a–c) Annual mean, (d–f) trend, and (g–i) anomaly time series of the extreme rainfall intensity (rx1day; left panels), frequency (r30mm; middle panels), and duration of consecutive wet days (cwd; right panels). The anomaly time series is for the area weighted average over the Northern Subregion [NSR: 15°N–0°, 20°–41°E; red box in (d)] in red line and Southern Subregion [SSR: 0°–15°S, 20°–41°E; blue box in (d)] in blue line, and the overall trend of each line is shown by a dashed line of its corresponding color. The slopes/magnitudes of the trend lines are shown in Table 2.

    Fig. 5.  The extreme rainfall intensity (rx1day) (a–d) seasonal mean and (e–h) trend during the study period (1998–2018) for JF (first column), MAM (second column), JJAS (third column), and OND (fourth column). (i–l) show the anomaly time series of rx1day (mm day−1) over the Northern Subregion [NSR: 15°N–0°, 20°–41°E; red box in (e)] in red line and Southern Subregion [SSR: 0–15°S, 20°–41°E; blue box in (e)] in blue line for (i) JF, (j) MAM, (k) JJAS, and (l) OND, and the overall trend of each line is shown by a dashed line of its corresponding color. The slopes/magnitudes of the trend lines are shown in Table 2.

    Fig. 6.  As in Fig. 5, but for the frequency of extreme rainfall events (r30mm).

    Fig. 7.  As in Fig. 5, but for the extreme duration of consecutive wet days (cwd).

    Table 1.  List of the Precipitation Extreme Indices (PEIs) used in this study, which are selected from the Expert Team on Climate Change Detection and Indices (ETCCDI)

    IDNameDefinition/DescriptionUnit
    rx1dayExtreme rainfall intensityHighest 1-day precipitation totalmm
    r30mmVery heavy rain dayCount of days when precipitation ≥ 30 mmday
    cwdConsecutive wet dayMaximum number of consecutive wet day (with precipitation ≥ 1 mm)day
    Download: Download as CSV

    Table 2.  Seasonal- and annual-mean trend slopes/magnitudes of the PEIs averaged over the Northern Subregion (NSR) and Southern Subregion (SSR) during 1998–2018. Refer to the bottom panels of Figs. 4-7 to see the trend lines and additional information

    Index JF MAM JJAS OND Annual
    NSRSSRNSRSSRNSRSSRNSRSSRNSRSSR
    rx1day−0.192−0.138−0.2230.3880.5440.300−0.327−0.0220.6370.357
    r30mm−0.001 0.001−0.002−0.0040.010−0.002−0.004 0.0050.064 0.002
    cwd−0.010−0.019 0.023 0.018 0.027−0.014−0.005 0.024 0.025 0.022
    Note: Trends significant at the 5% level are shown in boldface.
    Download: Download as CSV
  • [1]

    Aguilar, E., A. Aziz Barry, M. Brunet, et al., 2009: Changes in temperature and precipitation extremes in western central Africa, Guinea Conakry, and Zimbabwe, 1955–2006. J. Geophys. Res. Atmos., 114, D02115. doi: 10.1029/2008JD011010.
    [2]

    Alexander, L. V., 2016: Global observed long-term changes in temperature and precipitation extremes: A review of progress and limitations in IPCC assessments and beyond. Weather Clim. Extremes, 11, 4–16. doi: 10.1016/j.wace.2015.10.007.
    [3]

    Alex-Ogwang, B., H. S. Chen, X. Li, et al., 2014: The influence of topography on east African October to December climate: Sensitivity experiments with RegCM4. Adv. Meteor., 2014, 1–14. doi: 10.1155/2014/143917.
    [4]

    Berhane, A., G. Hadgu, W. Worku, et al., 2020: Trends in extreme temperature and rainfall indices in the semi-arid areas of Western Tigray, Ethiopia. Environ. Syst. Res., 9, 3. doi: 10.1186/s40068-020-00165-6.
    [5]

    Blunden, J., D. S. Arndt, and M. O. Baringer, 2011: State of the climate in 2010. Bull. Amer. Meteor. Soc., 92, S1–S236. doi: 10.1175/1520-0477-92.6.S1.
    [6]

    Bushesha, M. S., and J. A. Mbura, 2015: Identification of reasons for and socio-economic impacts of persistent floods in Dar es Salaam. World J. Soc. Sci. Res., 2, 180. doi: 10.22158/wjssr.v2n2p180.
    [7]

    Cattani, E., A. Merino, and V. Levizzani, 2016: Evaluation of monthly satellite-derived precipitation products over East Africa. J. Hydrometeor., 17, 2555–2573. doi: 10.1175/JHM-D-15-0042.1.
    [8]

    Cattani, E., A. Merino, J. A. Guijarro, et al., 2018: East Africa rainfall trends and variability 1983–2015 using three long-term satellite products. Remote Sens., 10, 931. doi: 10.3390/rs10060931.
    [9]

    Chambers, J. Q., and D. A. Roberts, 2014: Drought in the Congo Basin. Nature, 509, 36–37. doi: 10.1038/nature13330.
    [10]

    Chang’a, L. B., A. L. Kijazi, K. B. Mafuru, et al., 2020: Understanding the evolution and socio-economic impacts of the extreme rainfall events in March–May 2017 to 2020 in East Africa. Atmos. Climate Sci., 10, 553–572. doi: 10.4236/acs.2020.104029.
    [11]

    Chen, M. Y., W. Shi, P. P. Xie, et al., 2008: Assessing objective techniques for gauge-based analyses of global daily precipitation. J. Geophys. Res. Atmos., 113, D04110. doi: 10.1029/2007JD009132.
    [12]

    Chikoore, H., J. H. Vermeulen, and M. R. Jury, 2015: Tropical cyclones in the Mozambique Channel: January–March 2012. Nat. Hazards, 77, 2081–2095. doi: 10.1007/s11069-015-1691-0.
    [13]

    Cuthbert, M. O., R. G. Taylor, G. Favreau, et al., 2019: Observed controls on resilience of groundwater to climate variability in sub-Saharan Africa. Nature, 572, 230–234. doi: 10.1038/s41586-019-1441-7.
    [14]

    Dinku, T., P. Ceccato, E. Grover-Kopec, et al., 2007: Validation of satellite rainfall products over East Africa’s complex topography. Int. J. Remote Sens., 28, 1503–1526. doi: 10.1080/01431160600954688.
    [15]

    Dyer, E. L. E., D. B. A. Jones, J. Nusbaumer, et al., 2017: Congo Basin precipitation: Assessing seasonality, regional interactions, and sources of moisture. J. Geophys. Res. Atmos., 122, 6882–6898. doi: 10.1002/2016JD026240.
    [16]

    Enyew, B. D., and G. J. Steeneveld, 2014: Analysing the impact of topography on precipitation and flooding on the Ethiopian highlands. J. Geol. Geosci., 3, 1000173. doi: 10.4172/2329-6755.1000173.
    [17]

    Fessehaye, M., Y. Brugnara, M. J. Savage, et al., 2019: A note on air temperature and precipitation variability and extremes over Asmara: 1914–2015. Int. J. Climatol., 39, 5215–5227. doi: 10.1002/joc.6134.
    [18]

    Fitchett, J. M., and S. W. Grab, 2014: A 66-year tropical cyclone record for South-East Africa: temporal trends in a global context. Int. J. Climatol., 34, 3604–3615. doi: 10.1002/joc.3932.
    [19]

    Funk, C., M. D. Dettinger, J. C. Michaelsen, et al., 2008: Warming of the Indian Ocean threatens eastern and southern Afri-can food security but could be mitigated by agricultural development. Proc. Natl. Acad. Sci. USA, 105, 11,081–11,086. doi: 10.1073/pnas.0708196105.
    [20]

    Funk, C., A. Hoell, S. Shukla, et al., 2016: The East African monsoon system: Seasonal climatologies and recent variations. The Monsoons and Climate Change: Observations and Modeling, L. M. V. de Carvalho, and C. Jones, Eds., Springer International Publishing, Cham, 163–185, doi: 10.1007/978-3-319-21650-8_8.
    [21]

    Funk, C. C., G. Eilerts, J. Verdin, et al., 2011: A climate trend analysis of Sudan. Fact Sheet 2011-3072. U.S. Geological Survey, Reston, VA, 6 pp, doi: 10.3133/fs20113072.
    [22]

    Funk, C. C., J. Rowland, G. Eilerts, et al., 2012a: A climate trend analysis of Ethiopia. Fact Sheet 2012-3053. U.S. Geological Survey, Reston, VA, 6 pp, doi: 10.3133/fs20123053.
    [23]

    Funk, C. C., J. Rowland, G. Eilerts, et al., 2012b: A climate trend analysis of Uganda. Fact Sheet 2012-3062. U.S. Geological Survey, Reston, VA, 4 pp, doi: 10.3133/fs20123062.
    [24]

    Gebrechorkos, S. H., S. Hülsmann, and C. Bernhofer, 2019: Changes in temperature and precipitation extremes in Ethiopia, Kenya, and Tanzania. Int. J. Climatol., 39, 18–30. doi: 10.1002/joc.5777.
    [25]

    Gebrechorkos, S. H., S. Hülsmann, and C. Bernhofer, 2020: Analysis of climate variability and droughts in East Africa using high-resolution climate data products. Glob. Planet. Change, 186, 103130. doi: 10.1016/j.gloplacha.2020.103130.
    [26]

    Harrison, L., C. Funk, and P. Peterson, 2019: Identifying changing precipitation extremes in Sub-Saharan Africa with gauge and satellite products. Environ. Res. Lett., 14, 085007. doi: 10.1088/1748-9326/ab2cae.
    [27]

    Hastenrath, S., D. Polzin, and C. Mutai, 2010: Diagnosing the droughts and floods in equatorial East Africa during boreal autumn 2005–08. J. Climate, 23, 813–817. doi: 10.1175/2009JCLI3094.1.
    [28]

    Hession, S. L., and N. Moore, 2011: A spatial regression analysis of the influence of topography on monthly rainfall in East Africa. Int. J. Climatol., 31, 1440–1456. doi: 10.1002/joc.2174.
    [29]

    Huffman, G. J., D. T. Bolvin, E. J. Nelkin, et al., 2007: The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. J. Hydrometeor., 8, 38–55. doi: 10.1175/JHM560.1.
    [30]

    Jiang, S. S., Z. X. Zhang, Y. H. Huang, et al., 2017: Evaluating the TRMM multisatellite precipitation analysis for extreme precipitation and streamflow in Ganjiang River basin, China. Adv. Meteor., 2017, 2902493. doi: 10.1155/2017/2902493.
    [31]

    Jiang, Y., L. M. Zhou, C. J. Tucker, et al., 2019: Widespread increase of boreal summer dry season length over the Congo rainforest. Nat. Climate Change, 9, 617–622. doi: 10.1038/s41558-019-0512-y.
    [32]

    Kendall, M. G., 1970: Rank Correlation Methods. 4th ed. Charles Griffin, London, 210 pp.
    [33]

    Kilavi, M., D. MacLeod, M. Ambani, et al., 2018: Extreme rainfall and flooding over central Kenya including Nairobi city during the long-rains season 2018: Causes, predictability, and potential for early warning and actions. Atmosphere, 9, 472. doi: 10.3390/atmos9120472.
    [34]

    Kiunsi, R., 2013: The constraints on climate change adaptation in a city with a large development deficit: the case of Dar es Salaam. Environ. Urban., 25, 321–337. doi: 10.1177/0956247813489617.
    [35]

    Kucera, P. A., E. E. Ebert, F. J. Turk, et al., 2013: Precipitation from space: Advancing earth system science. Bull. Amer. Meteor. Soc., 94, 365–375. doi: 10.1175/BAMS-D-11-00171.1.
    [36]

    Liebmann, B., I. Bladé, G. N. Kiladis, et al., 2012: Seasonality of African precipitation from 1996 to 2009. J. Climate, 25, 4304–4322. doi: 10.1175/JCLI-D-11-00157.1.
    [37]

    Liebmann, B., M. P. Hoerling, C. Funk, et al., 2014: Understanding recent eastern Horn of Africa rainfall variability and change. J. Climate, 27, 8630–8645. doi: 10.1175/JCLI-D-13-00714.1.
    [38]

    Liebmann, B., I. Bladé, C. Funk, et al., 2017: Climatology and interannual variability of boreal spring wet season precipitation in the eastern Horn of Africa and implications for its recent decline. J. Climate, 30, 3867–3886. doi: 10.1175/JCLI-D-16-0452.1.
    [39]

    Lyon, B., 2014: Seasonal drought in the Greater Horn of Africa and its recent increase during the March–May long rains. J. Climate, 27, 7953–7975. doi: 10.1175/JCLI-D-13-00459.1.
    [40]

    Lyon, B., and D. G. Dewitt, 2012: A recent and abrupt decline in the East African long rains. Geophys. Res. Lett., 39, L02702. doi: 10.1029/2011GL050337.
    [41]

    Lyon, B., and N. Vigaud, 2017: Unraveling East Africa’s climate paradox. Climate Extremes: Patterns and Mechanisms, S. Y. S. Wang, J. H. Yoon, C. C. Funk, et al., Eds., American Geophysical Union (AGU), Hoboken, 265–281, doi: 10.1002/9781119068020.ch16.
    [42]

    Mafuru, K. B., and G. R. Tan, 2018: Assessing prone areas to heavy rainfall and the impaction of the upper warm temperature anomaly during March–May rainfall season in Tanzania. Adv. Meteor., 2018, 8353296. doi: 10.1155/2018/8353296.
    [43]

    Mahbod, M., A. Shirvani, and F. Veronesi, 2019: A comparative analysis of the precipitation extremes obtained from tropical rainfall-measuring mission satellite and rain gauges datasets over a semiarid region. Int. J. Climatol., 39, 495–515. doi: 10.1002/joc.5824.
    [44]

    Mann, H. B., 1945: Nonparametric tests against trend. Econome-trica, 13, 245–259. doi: 10.2307/1907187.
    [45]

    Mavume, A., L. Rydberg, M. Rouault, et al., 2009: Climatology and landfall of tropical cyclones in the South-West Indian Ocean. West. Indian Ocean J. Mar. Sci., 8, 15–36. doi: 10.4314/wiojms.v8i1.56672.
    [46]

    Meier-Fleischer, K., M. Böttinger, and M. Haley, 2017: NCL User Guide. High Quality Graphics with NCL 6.4.0. Deutsches Klimarechenzentrum (DKRZ), Hamburg, Germany, 294 pp, doi: 10.5065/D6WD3XH5.
    [47]

    Mekasha, A., K. Tesfaye, and A. J. Duncan, 2014: Trends in daily observed temperature and precipitation extremes over three Ethiopian eco-environments. Int. J. Climatol., 34, 1990–1999. doi: 10.1002/joc.3816.
    [48]

    Mequaninta, F., R. Mitikub, and A. Shimelesc, 2016: Observed and future climate variability and extremes over East Shoa Zone, Ethiopia. J. Climatol. Wea. Forecasting, 4, 1000183. doi: 10.4172/2332-2594.1000183.
    [49]

    Michaelides, S., V. Levizzani, E. Anagnostou, et al., 2009: Precipitation: Measurement, remote sensing, climatology and modeling. Atmos. Res., 94, 512–533. doi: 10.1016/j.atmosres.2009.08.017.
    [50]

    Niang, I., O. C. Ruppel, M. A. Abdrabo, et al., 2014: Africa Climate Change 2014: Impacts, Adaptation and Vulnerability. Part B: Regional Aspects. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Pa-nel on Climate Change, V. R. Barros, C. B. Field, D. J. Dokken, et al., Eds., Cambridge University Press, Cambridge, 1199–1266, doi: 10.1017/CBO9781107415386.002.
    [51]

    Nicholson, S. E., 1996: A review of climate dynamics and climate variability in eastern Africa. The Limnology, Climatology and Paleoclimatology of the East African Lakes, T. C. Johnson, E. O. Odada, and K. T. Whittaker, Eds., Routledge, London, 25–56, doi: 10.1201/9780203748978-2.
    [52]

    Nicholson, S. E., 2015: Long-term variability of the East African ‘short rains’ and its links to large-scale factors. Int. J. Climatol., 35, 3979–3990. doi: 10.1002/joc.4259.
    [53]

    Nicholson, S. E., 2016a: An analysis of recent rainfall conditions in eastern Africa. Int. J. Climatol., 36, 526–532. doi: 10.1002/joc.4358.
    [54]

    Nicholson, S. E., 2016b: The Turkana low-level jet: mean climatology and association with regional aridity. Int. J. Climatol., 36, 2598–2614. doi: 10.1002/joc.4515.
    [55]

    Nicholson, S. E., 2017: Climate and climatic variability of rainfall over eastern Africa. Rev. Geophys., 55, 590–635. doi: 10.1002/2016RG000544.
    [56]

    Nicholson, S. E., 2018: The ITCZ and the seasonal cycle over equatorial Africa. Bull. Amer. Meteor. Soc., 99, 337–348. doi: 10.1175/BAMS-D-16-0287.1.
    [57]

    Nicholson, S. E., B. Some, J. McCollum, et al., 2003: Validation of TRMM and other rainfall estimates with a high-density gauge dataset for West Africa. Part II: Validation of TRMM rainfall products. J. Appl. Meteor., 42, 1355–1368. doi: 10.1175/1520-0450(2003)042<1355:VOTAOR>2.0.CO;2.
    [58]

    Omondi, P. A., J. L. Awange, E. Forootan, et al., 2014: Changes in temperature and precipitation extremes over the Greater Horn of Africa region from 1961 to 2010. Int. J. Climatol., 34, 1262–1277. doi: 10.1002/joc.3763.
    [59]

    Ongoma, V., H. S. Chen, C. J. Gao, et al., 2018a: Future changes in climate extremes over Equatorial East Africa based on CMIP5 multimodel ensemble. Nat. Hazards, 90, 901–920. doi: 10.1007/s11069-017-3079-9.
    [60]

    Ongoma, V., H. S. Chen, and G. W. Omony, 2018b: Variability of extreme weather events over the equatorial East Africa, a case study of rainfall in Kenya and Uganda. Theor. Appl. Climatol., 131, 295–308. doi: 10.1007/s00704-016-1973-9.
    [61]

    Onyango, A. O., H. M. Xu, and Z. H. Lin, 2020: Diurnal cycle of rainfall over Lake Victoria Basin during the long-rain season based on TRMM satellite estimate. Int. J. Climatol., 40, 4622–4637. doi: 10.1002/joc.6479.
    [62]

    Ozer, P., and A. Mahamoud, 2013: Recent extreme precipitation and temperature changes in Djibouti City (1966–2011). J. Climatol., 2013, 928501. doi: 10.1155/2013/928501.
    [63]

    Peterson, T. C., and M. J. Manton, 2008: Monitoring changes in climate extremes: A tale of international collaboration. Bull. Amer. Meteor. Soc., 89, 1266–1271, doi: 10.1175/1520-0477-89.9.1251.
    [64]

    Schmocker, J., H. P. Liniger, J. N. Ngeru, et al., 2016: Trends in mean and extreme precipitation in the Mount Kenya region from observations and reanalyses. Int. J. Climatol., 36, 1500–1514. doi: 10.1002/joc.4438.
    [65]

    Schott, F. A., S. P. Xie, and J. P. Jr. McCreary, 2009: Indian Ocean circulation and climate variability. Rev. Geophys., 47, RG1002. doi: 10.1029/2007RG000245.
    [66]

    Schulzweida, U., 2019: CDO User Guide (1.9.8). Zenodo, 222 pp, doi: 10.5281/ZENODO.3539275.
    [67]

    Sen, P. K., 1968: Estimates of the regression coefficient based on Kendall’s Tau. J. Amer. Stat. Assoc., 63, 1379–1389. doi: 10.1080/01621459.1968.10480934.
    [68]

    Seneviratne, S. I., N. Nicholls, D. Easterling, et al., 2012: Changes in climate extremes and their impacts on the natural physical environment. Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation: Special Report of the Intergovernmental Panel on Climate Change, C. B. Field, V. Barros, T. F. Stocker, et al., Eds., Cambridge University Press, Cambridge, UK, 109–230, doi: 10.1017/CBO9781139177245.006.
    [69]

    Shiferaw, A., T. Tadesse, C. Rowe, et al., 2018: Precipitation extremes in dynamically downscaled climate scenarios over the Greater Horn of Africa. Atmosphere, 9, 112. doi: 10.3390/atmos9030112.
    [70]

    Tan, M. L., 2019: Assessment of TRMM product for precipitation extreme measurement over the Muda River Basin, Malaysia. HydroResearch, 2, 69–75. doi: 10.1016/j.hydres.2019.11.004.
    [71]

    Taylor, K. E., 2001: Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res. Atmos., 106, 7183–7192. doi: 10.1029/2000JD900719.
    [72]

    Taylor, R. G., M. C. Todd, L. Kongola, et al., 2013: Evidence of the dependence of groundwater resources on extreme rainfall in East Africa. Nat. Climate Change, 3, 374–378. doi: 10.1038/nclimate1731.
    [73]

    Theil, H., 1992: A rank-invariant method of linear and polynomial regression analysis. Henri Theil’s Contributions to Economics and Econometrics: Econometric Theory and Methodology, B. Raj, and J. Koerts, Eds., Springer, Dordrecht, 345–381, doi: 10.1007/978-94-011-2546-8_20.
    [74]

    Vincent, L. A., E. Aguilar, M. Saindou, et al., 2011: Observed trends in indices of daily and extreme temperature and precipitation for the countries of the western Indian Ocean, 1961–2008. J. Geophys. Res. Atmos., 116, D10108. doi: 10.1029/2010JD015303.
    [75]

    Wainwright, C. M., J. H. Marsham, R. J. Keane, et al., 2019: ‘Eastern African Paradox’ rainfall decline due to shorter not less intense Long Rains. npj Climate Atmos. Sci., 2, 34. doi: 10.1038/s41612-019-0091-7.
    [76]

    Wainwright, C. M., D. L. Finney, M. Kilavi, et al., 2021: Extreme rainfall in East Africa, October 2019–January 2020 and context under future climate change. Weather, 76, 26–31. doi: 10.1002/wea.3824.
    [77]

    Walker, D. P., J. H. Marsham, C. E. Birch, et al., 2020: Common mechanism for interannual and decadal variability in the East African long rains. Geophys. Res. Lett., 47, e2020GL089182. doi: 10.1029/2020GL089182.
    [78]

    Wenhaji Ndomeni, C., E. Cattani, A. Merino, et al., 2018: An observational study of the variability of East African rainfall with respect to sea surface temperature and soil moisture. Quart. J. Roy. Meteor. Soc., 144, 384–404. doi: 10.1002/qj.3255.
    [79]

    Williams, A. P., and C. Funk, 2011: A westward extension of the warm pool leads to a westward extension of the Walker circulation, drying eastern Africa. Climate Dyn., 37, 2417–2435. doi: 10.1007/s00382-010-0984-y.
    [80]

    Williams, A. P., C. Funk, J. Michaelsen, et al., 2012: Recent summer precipitation trends in the Greater Horn of Africa and the emerging role of Indian Ocean sea surface temperature. Climate Dyn., 39, 2307–2328. doi: 10.1007/s00382-011-1222-y.
    [81]

    Yang, W. C., R. Seager, M. A. Cane, et al., 2015: The annual cycle of East African precipitation. J. Climate, 28, 2385–2404. doi: 10.1175/JCLI-D-14-00484.1.
    [82]

    Zhang, Q., V. P. Singh, J. F. Li, et al., 2011: Analysis of the periods of maximum consecutive wet days in China. J. Geophys. Res. Atmos., 116, D23106. doi: 10.1029/2011JD016088.
    [83]

    Zhang, X. B., L. Alexander, G. C. Hegerl, et al., 2011: Indices for monitoring changes in extremes based on daily temperature and precipitation data. WIREs Climate Change, 2, 851–870. doi: 10.1002/wcc.147.
    [84]

    Zhang, Y., F. Jiang, W. Wei, et al., 2012: Changes in annual maximum number of consecutive dry and wet days during 1961–2008 in Xinjiang, China. Nat. Hazards Earth Syst. Sci., 12, 1353–1365. doi: 10.5194/nhess-12-1353-2012.
    [85]

    Zhou, L. M., Y. H. Tian, R. B. Myneni, et al., 2014: Widespread decline of Congo rainforest greenness in the past decade. Nature, 508, 86–90. doi: 10.1038/nature13265.
    [86]

    Zolina, O., C. Simmer, S. K. Gulev, et al., 2010: Changing structure of European precipitation: Longer wet periods leading to more abundant rainfalls. Geophys. Res. Lett., 37, L06704. doi: 10.1029/2010GL042468.
  • Gensuo JIA and Zacharia Florence MTEWELE.pdf

  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Heterogeneous Trends of Precipitation Extremes in Recent Two Decades over East Africa

    Corresponding author: Gensuo JIA, jiong@tea.ac.cn
  • 1. Key Laboratory of Regional Climate–Environment for Temperate East Asia, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China
  • 2. University of Chinese Academy of Sciences, Beijing 100049, China
  • 3. Physics Department, University of Dar es Salaam, Dar es Salaam P. O. Box 35063, Tanzania
Funds: Supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (CAS)—CAS Big Earth Data Science Engineering Program (XDA19030401) and National Key Research and Development Program of China (2016YFA0600303)

Abstract: East Africa is so vulnerable to the impacts of precipitation extremes varying from frequent floods to prolonged droughts. However, systematic regional assessment of precipitation extremes across seasons has received little attention, and most previous studies of precipitation extremes have employed many indices and sparse gauge observations giving marginalized details. In this study, we use three precipitation extreme indices to examine the intensity of the highest single-day rainfall record (rx1day), prevalence of very heavy rainfalls (r30mm), and persistence of successive wet days (cwd) at both annual and seasonal scales over recent two decades (1998–2018) based on the Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis data. The results show that the most intensive and frequent precipitation extremes are noticeable from January to May across the areas extending from Madagascar to the Tanzanian coastal zone. These areas also exhibit patches of significant increasing trends in frequency, duration, and intensity of precipitation extremes annually and seasonally. However, significant declines in frequency and intensity of precipitation extremes are observed from western Ethiopia to Congo–Uganda, especially in June–September. The October–December season witnesses higher interannual variability amounting to overall weak and less significant trends. Further subregional assessment shows overall declining intensity and frequency of precipitation extremes in northern part of the study areas. Mean wetness duration increased, with persistence of moderate wet days and slight reduction of severe events. This study unveils higher susceptibility of the East African region to the widely observed hotspots of precipitation extremes posing threats to food security, water resource, and human well-being. The region should consider upscaling irrigation schemes in addition to planning resilient and supportive infrastructures to withstand the upsurging precipitation extremes, especially along the coastal zone.

近20年东非极端降水变化趋势的空间异质性

从频繁的洪水到长期的干旱,东非极易受到极端降水事件的影响。然而,对东非极端降水事件的已有研究大多采用了稀疏的地面观测资料,且缺乏对季节尺度极端降水事件的系统评估。本研究使用热带降雨测量任务(TRMM)的降水资料研究了近20年年际和季节尺度东非极端降水特征。结果表明,1–5月,马达加斯加到坦桑尼亚沿海地区极端降水频繁,且极端降水的频率、持续时间和强度呈增加趋势。6–9月,埃塞俄比亚西部到刚果–乌干达地区极端降水的频率和强度显著下降。 10–12月,极端降水的年际变率较高,总体趋势不显著。东非地区应考虑扩大灌溉计划,同时规划支持性的基础设施,以抵御频发的极端降水事件,尤其在对极端降水威胁更加敏感的沿海地区。
    • Africa is one of the most vulnerable continents to extreme weather and climate, which is exacerbated by its high exposure and low coping capacity (Niang et al., 2014). East African countries have often been facing precipitation extremes alternating between droughts and floods (Hastenrath et al., 2010; Nicholson, 2016a). The drought conditions often persist for more than a single rainy season or even years; meanwhile, they may be preceded or followed by abrupt severe floods (Lyon and Dewitt, 2012; Nicholson, 2016a). For instance, the 2010/2011 drought conditions that hit much of the eastern African region were followed by severe floods over the same area (Nicholson, 2016a). Similarly, prolonged 2008 drought conditions were followed by flooding in early 2010 (Nicholson, 2016a), whereas the 1997/1998 severe floods during the El Niño episode preceded 1998/1999 drought conditions. This is perhaps the heart of economic crisis within the region as the livelihoods and socio-economic activities of the vast majority rely on rain-fed agricultural systems, pastoralism, and fishery, which are very sensitive to extreme rainfall. The alternating drought and flood episodes may thus keep on reversing the economic growth, and subsequently threatening the achievement of sustainable development among individuals, countries, and the region as a whole.

      Annual negative and positive trends of precipitation extreme indices have been found across East Africa although their significance was limited to only a few indices and stations. During 1981–2010, Djibouti experienced a significant decline in intensity and frequency of precipitation extremes (Ozer and Mahamoud, 2013). That was also the case for Eritrea during 1914–2015, which was further accompanied by shortening wetness persistence (Fessehaye et al., 2019). Insignificant decreasing trends in frequency, intensity, and duration of precipitation extremes were also pointed out for most stations over East Africa during 1961–2010 (Omondi et al., 2014). The reducing frequency of very heavy rainy days was also observed across some stations in Kenya (1961–2010) and Uganda (1980–2010) (Ongoma et al., 2018b). Nonetheless, these notable trend patterns were incoherent for some neighboring stations with opposite trends and rare statistical significance. Moreover, most of the studies focused only on annual patterns of precipitation extremes, leaving seasonal details little explored.

      Assessment of regional observed changes in precipitation extremes over East Africa in the fifth assessment report of the Intergovernmental Panel on Climate Change (IPCC) was constrained by the limited availability of literature (Seneviratne et al., 2012). This was also the case over most other African regions, in addition to lack of data and inconsistency among reported patterns in some regions. Afterwards, East Africa has witnessed a growing number of studies exploring historical changes, variability, trends, and projections of precipitation and temperature related extremes from local to regional coverage (Omondi et al., 2014; Cattani et al., 2018; Ongoma et al., 2018a, b; Shiferaw et al., 2018). Precipitation extreme indices designated by the World Meteorological Organization (WMO) Expert Team on Climate Change Detection and Indices (ETCCDI) have often been used in studying precipitation extremes across different parts of the world. Most of the previous studies using precipitation extreme indices in East Africa employed station gauge observations ranging from covering the whole East Africa (Omondi et al., 2014) to subregional and local scales [e.g., Uganda and Kenya (Ongoma et al., 2018b), Djibouti city (Ozer and Mahamoud, 2013), subregions in Ethiopia (Mekasha et al., 2014; Mequaninta et al., 2016; Berhane et al., 2020), and Asmara in Eritrea (Fessehaye et al., 2019)], while Schmocker et al. (2016) used both station observations and reanalysis datasets in their study over the Mount Kenya region.

      However, it is noteworthy that, there are limited gauge observations in the East African countries like in most other developing countries (Dinku et al., 2007; Gebrechorkos et al., 2019, 2020) due to (among others) limited funding allocated for the establishment, maintenance, and operation of such stations (Omondi et al., 2014). The poor spatial resolution of these in-situ observing stations within the region is a caveat that is always given when using the interpolated or other derived datasets from those inputs. This is pointed out by some of the available global observation station-related datasets like Climate Prediction Center (CPC) Global Unified Gauge-Based Analysis of Daily Precipitation (Chen et al., 2008). More-over, restrictions or individual countries’ policies on data sharing across different countries have somewhat been hampering regional studies aiming to employ gauge station observations across several countries. As such, most of the studies within the region that employed field observations focused over local instead of national and subregional scales (Ozer and Mahamoud 2013; Mekasha et al., 2014; Mequaninta et al., 2016; Schmocker et al., 2016; Ongoma et al., 2018b; Fessehaye et al., 2019; Berhane et al., 2020). Nevertheless, the daily field meteorological records were collected and compiled from all the 11 East African countries as a result of the workshops that enabled sharing of historical meteorological data from each country (Omondi et al., 2014). Remote sensing data, in particular satellite datasets, have also become of potential use in data-sparse regions, like East Africa in our case, as they provide both higher spatial and temporal resolutions and are freely available to the public (Michaelides et al., 2009; Cattani et al., 2016). These datasets enable even real-time monitoring and forecasting of precipitation extreme events and have become useful in climate analyses and application studies as they are accumulating over a sufficient temporal length (Michaelides et al., 2009; Kucera et al., 2013).

      In examining precipitation extremes over the recent two decades (1998–2018) in East Africa, we thus employed high-resolution Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis (TMPA, 3B42) data (Huffman et al., 2007). Our study provides a wide view of precipitation extremes across the whole East African region and neighboring areas, unlike most previous studies that focused over only local domains. The study gives a systematic assessment of precipitation extremes across seasons, which has received little attention, in addition to annual scale patterns. We have chosen three indices aiming at providing a detailed focus on duration, intensity, and frequency of precipitation extremes across the region.

    2.   Study domain
    • The East African region in this study extends over 25°N–15°S, 20°–55°E, across 11 countries including Burundi, Djibouti, Eritrea, Ethiopia, Kenya, Rwanda, Somalia, South Sudan, Sudan, Tanzania, and Uganda, which are also often addressed as the Greater Horn of Africa (Fig. 1). However, it should be noted that, based on regional political and economic integrations, the terms East Africa and Greater Horn of Africa may be carrying different definitions in terms of the number of member countries or areal coverage. Throughout this study, the term East Africa is used just to refer to the study domain as defined by the aforementioned latitudinal and longitudinal extents and not otherwise.

      Figure 1.  Map of the study domain showing the topographical elevation (m) above the mean sea level, major rivers and lakes, and gauge stations used for data validation. Enclosed is the African continent map showing the location of the study domain (red-colored rectangle).

      East Africa is well known for its distinguished highlands, the famous Great African Rift Valley, and Great African Lakes. These diverse topographical features along with the rift valleys, lakes, and oceans play significant roles in modulating the East African regional climatic conditions through local atmospheric circulations like mountain and valley breezes as well as lake and sea breezes apart from large scale or global variations brought about by teleconnections.

      The annual rainfall cycle over much of the East Afri-can region peaks either once or twice a year (Liebmann et al., 2012; Yang et al., 2015; Cattani et al., 2016). The northwestern (southern) areas’ rainfall regime peaks during boreal summer (winter), whereas equatorial to northeastern areas are characterized by bimodal regime peaking in boreal spring and autumn (Figs. 2a, c, e, g). Based on the rainfall seasonality in this region, the seasons selected for this study are January–February (JF), March–April–May (MAM), June–July–August–September (JJAS), and October–November–December (OND). JF is taken as a transition season between OND and MAM, during which rainfall is mainly confined over most of the southern areas of the study domain. These areas receive much of their rainfall from November to April (NDJFMA), thus accounting for both OND, JF, and MAM seasons. In the present study, JF was thus chosen to complement the rainfall regime in the further southern regions, where the NDJFMA season was not used, to avoid overlapping the same months in different seasons. Some studies with a larger spatial domain covering the whole or large part of East Africa have also used similar selection of seasons (Yang et al., 2015; Cattani et al., 2018; Wenhaji Ndomeni et al., 2018; Gebrechorkos et al., 2020).

      Figure 2.  Monthly rainfall climatology (left panels) and time series of rainfall from gauge stations (red line) and TRMM 3B42 product (blue line) during 1998–2009 (right panels) over (a, b) Sudan, (c, d) Uganda, (e, f) Kenya, and (g, h) Tanzania averaged for all stations over a respective country shown in Fig. 1. The correlation coefficient (r) and its significance (p) between the two datasets are shown on the top right of each panel.

    3.   Data and methods
    • This study primarily employed the satellite daily precipitation product version 7 of TRMM Multisatellite Precipitation Analysis (TMPA 3B42; Huffman et al., 2007). TMPA 3B42 is derived from the space-borne observations and adjusted with monthly rain gauges to improve understanding of the water cycle in the climate system (Huffman et al., 2007). TMPA datasets are a success of the international joint mission between NASA of the United States and the Japan Aerospace Exploration Agency (JAXA) aimed at providing the best precipitation estimates over the tropics where much of global precipitation occurs. The TMPA 3B42 data have high spatial (0.25° × 0.25°) and temporal (1-day) resolutions. Its spatial coverage extends from 50°S to 50°N, with a temporal span from 1998 to near-present (2019, last accessed August 2019). For this study, we extracted data over the spatial domain of 15°S–25°N, 20°–55°E enclosing the 11 countries of the East African region for a pe-riod of 21 yr from 1 January 1998 to 31 December 2018.

      We first reviewed and assessed the applicability of the TRMM precipitation product within the study domain and its neighboring areas. A validation study for monthly and seasonal mean rainfall fields from various TRMM products against high-density gauge datasets over West Africa showed that the final merged TRMM product is excellently agreeing with gauge observations within the region (Nicholson et al., 2003). Another validation study of 10-days and monthly rainfall totals from several satellite rainfall estimates against gauge observations over the eastern African complex terrains showed that TRMM products also performed relatively well and showed consistent results with the western African study (Dinku et al., 2007). Evaluation of precipitation variability from several satellite-derived monthly rainfall products over the East African region against rain gauge products revealed advantages of TMPA 3B42 product over other products (Cattani et al., 2016). This dataset was also used to study the diurnal rainfall variation over the Lake Victoria basin in East Africa following its high spatial and temporal resolution (Onyango et al., 2020). Moreover, several other studies across different parts of the world have employed this dataset in examining precipitation extremes, and pointed out that it relatively well agrees with gauge observations and performs better than many other satellite data products (Jiang et al., 2017; Harrison et al., 2019; Mahbod et al., 2019; Tan, 2019). For example, across sub-Saharan Africa, the performance of TRMM 3B42 ranked the highest among the various satellite products in consistency with gauge-based datasets and in reproduced precipitation extreme indices including the number of wet days (r1mm), precipitation total from wet days (prcptot), and maximum 1-day precipitation (rx1day) (Harrison et al., 2019). However, it is noted that the performance of the satellite-derived precipitation estimates within the East African region is partly challenged by the complex regional topographical features (Cattani et al., 2016).

      Furthermore, we validated the TRMM data with meteorological station observations acquired from four countries within the East African region. These were monthly rain gauge observations for a total of 98 stations (Fig. 1) sourced from national hydrometeorological departments of Sudan (27), Uganda (12), Kenya (33), and Tanzania (26), covering the 1998–2009 period. Daily TRMM 3B42 precipitation product was first aggregated to monthly totals and then extracted for each corresponding gauge station. The monthly climatology of each station was evaluated for both datasets and then averaged for each country to reproduce annual rainfall cycles. Similarly, the monthly series of each country was obtained by averaging all stations for each dataset. The two datasets revealed strong agreement in reproducing the annual rainfall cycle and monthly rainfall variations (r > 0.97, p < 0.001; Fig. 2). Moreover, Taylor diagram was also used to graphically examine how closely the TRMM 3B42 product matches gauge observations (Taylor, 2001). The statistics show that the two datasets agree well on reproducing both the annual rainfall cycle and monthly rainfall evolution. This is marked with the nearest positioning of the TRMM 3B42 data point for every country to the reference point of gauge observations (Fig. 3). Almost similar spatial variability was detected with close normalized standard deviation (SD ≈ 1), low centered root mean square error (RMSE ≤ 0.25), and high correlation (r ≥ 0.97) between the two datasets.

      Figure 3.  Taylor diagrams showing statistical comparison between the rain gauge observations (reference data; REF point) and TRMM-3B42 satellite-based precipitation product (blue marker) in reproducing the (a) annual rainfall cycle and (b) monthly rainfall evolution over (1) Sudan, (2) Uganda, (3) Kenya, and (4) Tanzania during 1998–2009. The angular axis (dotted line) shows the correlation coefficient (r), the radial axis (dashed concentric arc) shows the standardized deviation (SD), and the solid concentric semicircle denotes the centered root mean square error (RMSE) of TRMM against gauge observations. The gauge reference data point centered at REF is with r = 1, SD = 1, and RMSE = 0; and the closer the other point to it, the higher the agreement between the two datasets.

    • The assessment of precipitation extremes in this study was based on three precipitation extreme indices (PEIs; Table 1): the highest single-day rainfall record (rx1day), prevalence of very heavy rainfall of at least 30 mm in a day (r30mm), and persistence of successive wet days (cwd), to examine the temporal and spatial changes in the intensity, frequency, and duration of precipitation extremes, respectively. These indices are mathematically independent, but their cause and implication can be linked to each other. The selected indices are a subset of the climate extreme indices designated by the WMO ETCCDI to enhance uniform monitoring, detection, and attribution of changes in climate extremes from daily datasets since the 1990s (Peterson and Manton, 2008). This allows easy comparison of the results from different areas around the world as they conform to one ano-ther (Zhang X. B. et al., 2011). For instance, the IPCC reports evaluate available studies based on temperature and precipitation extreme indices across the globe to formulate several factual statements about the global and regional observed and projected changes in climate extremes (Alexander, 2016). The climate extreme indices, together with additional sector-specific climate indices, are now operationally maintained by another re-established Expert Team on Sector-specific Climate Indices (ET-SCI) since 2018 (https://www.wcrp-climate.org/data-etccdi).

      IDNameDefinition/DescriptionUnit
      rx1dayExtreme rainfall intensityHighest 1-day precipitation totalmm
      r30mmVery heavy rain dayCount of days when precipitation ≥ 30 mmday
      cwdConsecutive wet dayMaximum number of consecutive wet day (with precipitation ≥ 1 mm)day

      Table 1.  List of the Precipitation Extreme Indices (PEIs) used in this study, which are selected from the Expert Team on Climate Change Detection and Indices (ETCCDI)

    • The selected PEIs were computed from daily precipitation data (TRMM 3B42) by using ClimPACT2, the R software package designated by the ETCCDI and is now operationally maintained by ET-SCI. This package returns the computed PEIs at annual and monthly timescales. The annually computed PEIs are directly used for analyses of annual precipitation extremes patterns. Meanwhile, the monthly computed PEIs are evaluated on the selected seasonal timescales by using the Climate Data Operators (CDO; Schulzweida, 2019) for seasonal analyses of precipitation extremes. In performing the monthly-to-seasonal conversion in CDO, we used two major operators depending on the nature of the index. For rx1day and cwd, which are given as the maximum value each month, the seasonal maximum operator was used to select the maximum value for a season among the months of a particular season in every yearly time step. For r30mm, which counts the number of very heavy rainfall days in a month with at least 30 mm, the seasonal sum operator was supposed to obtain the total frequency in a season. However, the selected seasons (JF, MAM, JJAS, and OND) for our study vary in length, which would lead to significant variation in magnitudes among different seasons. Instead, we used a seasonal mean operator, which sums and then divides by the number of months in a season to obtain the seasonal mean values, enabling comparison among different seasons.

    • The overall tendency of precipitation extremes during the study period was assessed by computing the trends of the PEIs on each grid point with the Theil–Sen estimator (Sen, 1968; Theil, 1992). This method estimates the trend by computing the median of slopes [Eq. (1)] among all data pairs in a time series at a particular grid point,

      $${M}_{i}=\frac{{x}_{j}-{x}_{k}}{j-k}, $$ (1)

      where $ {M}_{i} $ is the slope of the i data pair $ ({x}_{j},{x}_{k}) $ and $ j>k $.

      We used the Mann–Kendall (MK) test (Mann, 1945; Kendall, 1970) to statistically assess the significance of any downward or upward monotonic trends of the selected indices during the study period. These are non-parametric methods that are not affected by the type of dataset distribution and missing values and are less sensitive to outliers. We employed the Mann–Kendall test from the NCAR Command Language (NCL) built-in function (trend_manken), which returns both the Mann–Kendall trend significance and Theil–Sen estimate of the linear trend. The trend analysis was performed at both annual and seasonal timescales for the PEIs. Spatial distributions of statistically significant trends at the 5% significance level are marked by stippled grid points.

      To assess the interannual variation of each index, we selected the northern subregion (NSR: 15°N–0°, 20°–41°E) and southern subregion (SSR: 0°–15°S, 20°–41°E). The subregional analysis for temporal evolution was adopted rather than whole regional area averaging because the study domain is much characterized by the north–south seasonal migration of the rainfall cycle. Thus, the northern and southern subregions have been considered to represent the Northern and Southern Hemispheric climatic dominancy during boreal summer and winter, respectively. Meanwhile, the climatic tendency during boreal spring and autumn is also embedded in both subregions along their transboundary region, providing the linkage between the two subregions. Temporal series was obtained by area weighting average of the index and then computing the anomaly series from its temporal mean. The trend of the anomaly series was also evaluated to assess the overall change during the study period. The significance of these trends was also assessed by using the Mann–Kendall test.

    4.   Results
    • According to the WMO Severe Weather Forecasting Demonstration Project-East Africa (SWFDP-EA), rainfalls with amounts exceeding 50 mm day−1 are deemed as extreme events of high risk across the region, which occurred over the vast domain and along the Tanzanian and Kenyan coastal belt reaching up to 120 mm day−1 (Fig. 4a). For arid and semiarid areas over Kenya, Somalia, eastern Ethiopia, and Sudan, rx1day features up to over 60 mm day−1 while annual rainfall total is on average less than 500 mm, implying that multiple occurrences of a few such rainfall events may account for their annual precipitation total. These intense rainfall events over arid and semiarid areas play a significant role in recharging ground water levels and episodically serving for water security recessions (Taylor et al., 2013; Cuthbert et al., 2019). Moreover, very heavy rainfalls are usually less than six days over much of arid and semiarid areas (Fig. 4b). However, the most frequent very heavy rainfalls are widely seen over the vicinity of Congo and Lake Victo-ria. Such rainfall events tend to affect people settling along the river streams and low-lying areas that are often hit with floods and exacerbated by their higher vulnerability and exposure conditions, including higher population density on unplanned settlements and poor drainage systems (Kiunsi, 2013; Bushesha and Mbura, 2015). Over the vast area of East Africa, cwd ranges within 5–10 days; however, mountainous areas suffer the longest cwd, especially the Ethiopian highlands with up to 20 day yr−1 (Fig. 4c). These regions also feature a higher average annual number of precipitating days (Cattani et al., 2018). This may be due to the terrain effects that enhance rainfall-producing systems over mountain ranges (e.g., Hession and Moore, 2011; Alex-Ogwang et al., 2014; Enyew and Steeneveld, 2014).

      Figure 4.  (a–c) Annual mean, (d–f) trend, and (g–i) anomaly time series of the extreme rainfall intensity (rx1day; left panels), frequency (r30mm; middle panels), and duration of consecutive wet days (cwd; right panels). The anomaly time series is for the area weighted average over the Northern Subregion [NSR: 15°N–0°, 20°–41°E; red box in (d)] in red line and Southern Subregion [SSR: 0°–15°S, 20°–41°E; blue box in (d)] in blue line, and the overall trend of each line is shown by a dashed line of its corresponding color. The slopes/magnitudes of the trend lines are shown in Table 2.

      The spatial patterns of trend show that rx1day exhi-bits a declining tendency over a larger area than frequency and duration indices. Both the northern and southern subregions are characterized by a reduction in extreme rainfall intensity (Fig. 4g; Table 2) because of the declines of up to −2 mm day−1 yr−1 over much of the western areas from Eritrea to Congo (Fig. 4d). For r30mm, the average decrease in frequency is revealed over NSR (Fig. 4h) as marked by significant declining trends of up to −0.28 day yr−1 over western Ethiopia to Uganda–Congo transect region (Fig. 4e). Besides, both NSR and SSR feature an equivalent weak increase in wetness persistence (Fig. 4i) with noticeable patches of both positive and negative cwd trends (Fig. 4f). Some periods of prolonged decline in frequency, intensity, and duration of precipitation extremes are also identified, like in 2006–2010 (Figs. 4g-i), which is in line with the reported post-2005 prolonged drought episodes (Nicholson, 2016a). However, r30mm and cwd show recovery to positive trends towards the end of the study period. Spatially, significant increasing trends in frequency, duration, and intensity of precipitation extremes are mainly observed over the southeastern region emanating from Madagascar towards the Tanzanian coastal zone.

      Index JF MAM JJAS OND Annual
      NSRSSRNSRSSRNSRSSRNSRSSRNSRSSR
      rx1day−0.192−0.138−0.2230.3880.5440.300−0.327−0.0220.6370.357
      r30mm−0.001 0.001−0.002−0.0040.010−0.002−0.004 0.0050.064 0.002
      cwd−0.010−0.019 0.023 0.018 0.027−0.014−0.005 0.024 0.025 0.022
      Note: Trends significant at the 5% level are shown in boldface.

      Table 2.  Seasonal- and annual-mean trend slopes/magnitudes of the PEIs averaged over the Northern Subregion (NSR) and Southern Subregion (SSR) during 1998–2018. Refer to the bottom panels of Figs. 4-7 to see the trend lines and additional information

    • The seasonal variation of rx1day progresses from south to north following the regional annual rainfall cycle. The wettest conditions during JF are concentrated over the southern areas where intensive rainfall events exceeding 60 mm day−1 are observed (Fig. 5a). The southeastern areas with further northward coastal extension also depict the highest rx1day of up to 100 mm day−1 during MAM (Fig. 5b). Over Kenya, Somalia, and eastern Ethiopia, where MAM is climatologically referred to as long rainy season, rx1day reaches as much as 60 mm day−1. During JJAS, intensive rainfalls of about 50–80 mm day−1 are observed throughout south-westward areas from western Ethiopia (Fig. 5c). Additionally, there is a notable narrow coastal belt that is more pronounced along the Kenya–Somalia coastline with up to 60 mm day−1 heavy rainfall, leaving much of the remaining areas experiencing light or dry rainfall conditions. Bimodal rainfall regime areas across Somalia, Kenya, and Tanzania, where the second rainfall season during OND is climatologically referred to as short rains, experience rainfall of up to 60 mm day−1 mainly confined along the eastern areas (Fig. 5d).

      Figure 5.  The extreme rainfall intensity (rx1day) (a–d) seasonal mean and (e–h) trend during the study period (1998–2018) for JF (first column), MAM (second column), JJAS (third column), and OND (fourth column). (i–l) show the anomaly time series of rx1day (mm day−1) over the Northern Subregion [NSR: 15°N–0°, 20°–41°E; red box in (e)] in red line and Southern Subregion [SSR: 0–15°S, 20°–41°E; blue box in (e)] in blue line for (i) JF, (j) MAM, (k) JJAS, and (l) OND, and the overall trend of each line is shown by a dashed line of its corresponding color. The slopes/magnitudes of the trend lines are shown in Table 2.

      MAM appears to have the highest rx1day compared to other seasons, suggesting more likely greater extreme rainfall impacts, especially along the coast. In Tanzania, Mafuru and Guirong (2018) found that the highest MAM rainfall amounts and heavy rainfall events are concentrated along the coastal belt and western Lake Victoria basin. For instance, on 11 April 2014 (6 May 2015), extreme rainfall events of about 135 mm (110 mm) in 24 h were reported in Dar es Salaam (a Tanzanian coastal city), which led to heavy floods that left several people dead, injured, and displaced from their homes and infrastructures destructed (Mafuru and Guirong, 2018). The 2018 MAM rainfalls were also reported as one of the wettest on record, with prolonged heavy rainfall episodes culminating in severe floods across the region (Kilavi et al., 2018). Actually, during MAM from 2017 to 2020, East Africa suffered several intense record-breaking extreme rainfall events with significantly adverse impacts (Chang’a et al., 2020).

      In JF, the highest increase in rx1day by up to 1.6 mm day−1 yr−1 was confined over the southeastern areas from Madagascar towards the Tanzanian coast (Fig. 5e). However, the northern and western areas were more characterized by declining tendency, leading to an average decline over both NSR and SSR (Fig. 5i). Precipitation extremes during MAM have been becoming less intense over large areas except along the northern Tanza-nian coast, Kenya, Ethiopia, and eastern areas, with SSR exhibiting a higher declining rate than NSR (Figs. 5f, j). Several studies also pointed out the decline of long (MAM) rains from the 1980s and even more abruptly post 1998 (Williams and Funk, 2011; Lyon and Dewitt, 2012; Liebmann et al., 2014, 2017; Lyon, 2014). None-theless, Wainwright et al. (2019) found that the decline was due to the shortening of rainy season (caused by later onset and earlier cessation) but not a decrease in rainy intensity. Besides, noticeable partial recovery of rx1day during the 2010s (Fig. 5j) was in line with the reported partial recovery of MAM rains from earlier cessation (Wainwright et al., 2019).

      Our analysis shows that wet areas during JJAS and OND were characterized by a reduction in rx1day with significance pronounced along the western Ethiopia–Congo transect region, giving a higher average NSR declining rate than SSR (Figs. 5g, 4h). On the other hand, the eastern areas with intensifying tendency of rx1day during MAM show opposite trends during OND except off the coast of Tanzania. An earlier study also detected a substantial reduction in summer (JJAS) rainfalls during the 1948–2009 period over many areas of East Africa (Williams et al., 2012). Their observed drying trend post the 1980s was attributed to increasing warming trends over the southern tropical Indian Ocean, which was linked with an increase in low-level subsidence over East Africa. Besides, OND rainfalls (short rains) revealed a wetting tendency during the 1979–2012 period, though with higher interannual variability over the eastern horn areas (Liebmann et al., 2014). The increase in rainfall was attributed to the warming of the western In-dian Ocean, whereas variability was associated with the Indian Ocean Dipole and El Niño–Southern Oscillation interannual fluctuations. Over these eastern horn areas, particularly along the northeastern Kenya–Somalia border and eastern Ethiopia, we also observed patches of increase in rx1day during OND, suggesting its wetting persistence over the recent decades.

    • The spatial distribution of mean seasonal frequency of precipitation extremes reveals north–south and west–east gradients. During JF, most frequent precipitation extre-mes appeared over the most southeastern areas of East Africa, reaching up to 3 day month−1 (Fig. 6a). On average, however, the vast area north of 10°S showed less than 1.5 day month−1. Further analysis indicates that frequent very heavy rainfalls during MAM prevailed along the Tanzania coastal zone and over Lake Victoria with an average of more than 2 day month−1 (Fig. 6b). Additionally, we observed an average of at least one r30mm event per month over southern areas of Tanzania and Ethiopia, central Kenya, and northern Congo.

      Figure 6.  As in Fig. 5, but for the frequency of extreme rainfall events (r30mm).

      A range of 1–2.5 day month−1 of extreme rainfalls demarcated well the climatological wet areas in boreal summer (JJAS), with the highest frequency seen over western Ethiopia and western South Sudan (Fig. 6c). This is not surprising as much of the northwestern regions of the study domain were reported experiencing bulk of their annual rainfall during the boreal summer, which was even considered as the eastern reach of the West African monsoons (Yang et al., 2015; Funk et al., 2016; Nicholson, 2017; Wenhaji Ndomeni et al., 2018). On the other hand, frequent r30mm events during OND are noticeable over Congo and Lake Victoria (Fig. 6d). Moreover, along the eastern horn areas, about 0.75–1.5 day month−1 of r30mm is noted during this remarkable short rainy season, particularly over southern areas of Somalia and Kenya.

      The spatial distributions of seasonal trend in the frequency of precipitation extremes are localized over a few areas. The results show that the areas from Madagascar to southern Tanzania feature a significant and the highest increase in the frequency of r30mm by up to 0.08 day month−1 yr−1 in JF (Fig. 6e). However, insignificant decreasing trends are noticed further inland, and both NSR and SSR regions exhibit averaged weak overall trends, with higher interannual fluctuations over the SSR (Fig. 6i). Remarkable increasing frequency of r30mm incidences during MAM mainly prevails along the Tanzania coastal zone, whereas declining patterns are spotted over the northeastern parts of Congo basin and southeastern parts of Tanzania (Fig. 6f). However, both NSR and SSR show an averaged declining trend whose recovery has been notable since 2013 (Fig. 6j). During JJAS, r30mm has been significantly decreasing in frequency, particularly along the region from western Ethiopia to northeastern Congo basin, which determines the major decreasing trend in the NSR (Figs. 6g, k). JJAS notably witnesses coherent and the highest decreasing trend patterns compared to other seasons. This declining tendency may reflect a prolonged reduction of the summer (JJAS) rainfall since the 1980s, which may be attributed to the warming of the southern tropical Indian Ocean (Williams et al., 2012). OND features weak r30mm trends over both NSR and SSR, with the locale around the Uganda–Congo border, with decreasing patterns; whereas increasing signatures are spotted along western Congo and the southeastern part of the domain (Figs. 6h, l). About 10%–20% of the rain in MAM and JJAS is also reported to decline across Ethiopia, Sudan, and Uganda from the 1970s to the late 2000s leading to diminishing areas viable for agropastoral activities (Funk et al., 2011, 2012a, b). As both r30mm and rx1day reveal predominantly decreasing trends over these areas in JJAS, it suggests that the summer rainfalls have continued to decline during the recent two decades. This drying tendency was previously predicted to be exacerbated by the warming trend and further leading to food insecurity and social conflicts to a rapidly growing population within the region (Funk et al., 2011, 2012a, b).

    • The duration of extreme rainfalls is remarkably long, with persistent wetness dominating over the highlands. In JF, the longest duration (up to 12 days) is concentrated over southern parts, especially over highlands; however, there is a notable southward increasing gradient across the domain (Fig. 7a). This is also the case for the mean spatial patterns of rx1day and r30mm indices during JF. In MAM, JJAS, and OND, longer durations of cwd (above 7 days) are mainly distributed over highlands, similar to its annual patterns (Figs. 7b-d). The longest cwd is predominantly observed over Ethiopian highlands during JJAS, reaching more than 15 days. Additionally, the Kenya–Tanzania coastal belt also features up to 9 days of cwd during MAM.

      Figure 7.  As in Fig. 5, but for the extreme duration of consecutive wet days (cwd).

      The spatial pattern of cwd trends in JF shows patches of decreasing tendency over most parts except along a stretching region from Madagascar to southwestern Tanzania with positive trends (Fig. 7e). A significant increasing tendency of cwd during MAM is noticeable over some parts of Congo, Lake Victoria, and along the Tanzania coastal zone towards Madagascar, which accounts for NSR and SSR increasing trends (Figs. 7f, j). On the other hand, its shortening tendency is mostly seen over the Kenya–Tanzania border and southwestern highlands. For the JJAS season, a higher wetness tendency over NSR and SSR depicts a negative slope because an increase in cwd is seen across much of Sudan whereas a declining tendency prevails over southern to eastern coastal land areas (Figs. 7g, 6k). During OND, SSR features weak increasing wetness duration due to notable cwd increasing trends across westward areas from western Tanzania, with a longer duration noted in 2006 (Figs. 7h, l). On the other hand, eastern land areas southward from southern Somalia and Ethiopian highlands are characterized by shortening of cwd. Downward (upward) trends of cwd shed light on the emerging risk of drought (flood) conditions (Zhang et al., 2012); however, this may further depend on the amount of precipitation accumulated during those successive wet days as well as their associated frequency and intensity (Zhang Q. et al., 2011). Moreover, the increasing tendency of cwd may also raise soil moisture storage and ground water recharge (Zolina et al., 2010).

    5.   Conclusions and discussion
    • This study examined the intensity, frequency, and duration of precipitation extremes across the East African region over the recent two decades (1998–2018). On average, the most frequent and intensive precipitation extremes prevailed across three major regions: the Congo basin, Lake Victoria basin, and southeastern coastal region. This predominantly demonstrates the major moisture source regions for the East African domain, which favors precipitation extremes in conjunction with the prevailing atmospheric circulations. The Congo basin primarily benefits from the abundant moisture supply from its tropical rainforests apart from remote sources (Dyer et al., 2017). The westerlies from this region, usually referred to as the Congo air mass, which is characteristically moist, favor precipitation extremes over East Africa (Williams et al., 2012; Mafuru and Guirong, 2018). The Lake Victoria basin also derives its prevalent extreme wet conditions from the diurnal variations across the lake and offshores, which is also promoted by the ample spatial coverage of the lake (68,870 km2; Onyango et al., 2020). Meanwhile, the most prolonged duration of consecutive wet days is prevalent over the highlands, indicating the orographic influence on wet conditions (Hession and Moore, 2011; Alex-Ogwang et al., 2014). The Turkana region exhibits low cwd, r30mm, and rx1day, promoting arid conditions and demarcating relatively high values between Ethiopia and Uganda–Kenya highlands. The aridity conditions are associated with divergent circulations along the entrance and core of the prevailing Turkana low-level jet (Nicholson et al., 2016b).

      The observed precipitation extremes (except cwd) over the Congo and Lake Victoria basins up to western Ethiopia tended to decrease, especially in JJAS with pronounced significant trends. This drying tendency has been associated with ecological changes, including the decline in Congo forest greenness, terrestrial water storage, and an increase in land surface temperature (Chambers and Roberts, 2014; Zhou et al., 2014; Jiang et al., 2019). The drying trends have also been implicated by several drought episodes in spite of abrupt flood events (Nicholson, 2016a). However, the observed noticeable patches of increasing trends of consecutive wet days imply the prevailing light to moderate precipitation, lessening the severity of drying tendency. Besides, projections reveal a likely increase in the intensity of heavy rainfall events across the Congo basin towards the end of the 21st century (Blunden et al., 2011).

      On the other hand, precipitation extremes across the southeastern coastal region are featured with increasing tendency, especially in JF and MAM. The cities along this coastal zone have also been impacted by several flood events exacerbated by the low-lying coastal plains and weak or non-resilient infrastructures (Chang’a et al., 2020). This coastal region and much of the southwestern Indian Ocean are often vulnerable to tropical cyclones and storms that have tremendously resulted in loss of lives, properties, and damage to societal infrastructures and ecosystems, especially if they make landfall (Chikoore et al., 2015). Although a majority of tropical cyclones and storms are tracked along the southwestern Indian Ocean and landfalling over Madagascar and Mozambique (Mavume et al., 2009; Fitchett and Grab, 2014), they also implicate substantial rainfall changes across some parts of East Africa (Wainwright et al., 2021). For instance, in March 2018, the tropical cyclones Eliakim and Dumazile over Madagascar were also linked to increased rainfall in East Africa (Wainwright et al., 2021). During 1994–2007, tropical cyclones were more intense and frequent over the southwestern Indian ocean than 1980–1993, partly ascribed to an increase in sea surface temperature (SST) (Mavume et al., 2009).

      Previous studies have also reported warming of the southern tropical Indian Ocean since the 1960s, accompanied by a significant increase in evaporation, convection, and precipitation (Funk et al., 2008; Williams and Funk, 2011). These may have also implicated the increasing tendency of precipitation extremes over the southeastern coastal areas of East Africa, partly enhanced by the prevailing inflow of the south equatorial ocean current (Schott et al., 2009). Nevertheless, the warming of the Indian Ocean is attributed to an increase in subsidence over much of the inland areas of East Africa, significantly reducing moisture influx from the Congo air mass (Funk et al., 2008; Williams and Funk, 2011; Williams et al., 2012). This largely accounts for the observed decreasing tendency in precipitation extremes across much of the study domain, including the Congo and Lake Victoria basins. Moreover, there was an increased zonal SST gradient over the tropical Pacific Ocean (Liebmann et al., 2014) triggered by the westward extension of the Indo-Pacific warm pool (Williams and Funk, 2011), which enhanced the Indian Ocean Walker circulation subsiding over East Africa associated with the drying tendency.

      Some earlier studies also reported the observed ann-ual declining tendency in frequency and intensity of precipitation extremes across some parts of Congo to Lake Victoria basins though this was based on sparsely gauged observations (Aguilar et al., 2009; Omondi et al., 2014; Ongoma et al., 2018b). This has implied the prolonging of the drying tendency since the latter half of the 20th century. Besides, over the southeastern region, Vincent et al. (2011) found general decreasing trends of precipitation extremes during 1961–2008, implying that the observed upward trends reversed over the recent decades. Our study also shows that during 1998–2018, the MAM decreasing trends in intensity and frequency of precipitation extremes mainly occurred over the first decade and later turned to upward trends. This may suggest a decadal–multidecadal variability signal linked to the natural variability of SST over the tropical Pacific Ocean (Lyon, 2014). On the other view, the swing from downward to upward trend may unveil the reported East Afri-can climate paradox with an observed MAM drying trend in recent decades and projected wetting tendency in future (Lyon and Vigaud, 2017; Walker et al., 2020). Moreover, weak overall trends in precipitation extremes are observed in OND, indicating its higher interannual variability tendency, which has been mainly linked to SST signatures associated with the Indian Ocean Dipole and El Niño–Southern Oscillation (Liebmann et al., 2014; Nicholson, 2015).

      Generally, the observed seasonal patterns of precipitation extremes largely conform to their respective annual spatial and temporal patterns but further highlight the north–south and west–east seasonal progression of precipitation extremes within the region. A similar south–north gradient (but of precipitation amount) in JF season was revealed by Yang et al. (2015), attributable to the JF strong north–south SST gradient along the coast. Moreover, they found a general strong north–south conditional instability with northern (southern) areas of the domain being extremely stable (less stable), which favors dry (wet) conditions. These observations also reflect the positioning and migration of the ITCZ, which is often attributed to rainfall seasonality across the domain (Yang et al., 2015; Nicholson, 1996, 2018; Wenhaji Ndomeni et al., 2018).

      This study’s peculiarity includes integrating the analysis of precipitation extremes with the high-resolution TRMM satellite product validated against meteorologi-cal station observations. Also, it provides a broader view of precipitation extremes patterns across the East Afri-can region and bordering areas, unlike most previous studies with only local to subregional coverages. More-over, analyses were made across all seasons within the region, which received little attention before, particularly for precipitation extreme indices. Unlike other studies that have used several precipitation extreme indices, our study chose only three indices for detailed focus on duration, intensity, and frequency of precipitation extremes across the region. The widely observed decreasing trends of precipitation extremes indicate higher susceptibility of the East African region to climate-induced changes in various sectors, including agriculture and ecosystems. The drying tendency has induced higher uncertainty on regional food security, which is over-reliant on rainfed subsistence farming, highlighting the need for upscaling irrigation schemes. Meanwhile, the observed increasing tendency of precipitation extremes along the coastal region indicates the significance for good planning of the coastal cities with more resilient and supportive infrastructures to extreme precipitation events.

      Acknowledgments. The TRMM Multisatellite Preci-pitation Analysis (TMPA 3B42) is freely available on-line at https://doi.org/10.5067/TRMM/TMPA/DAY/7. Topographical elevation of the study domain uses the quality-controlled global Digital Elevation Model (DEM) data from the Global Land One-km Base Elevation (GLOBE) Project acquired from NOAA NGDC GLOBE data library at https://iridl.ldeo.columbia.edu/SOURCES/.NOAA/.NGDC/.GLOBE/topo/datafiles.html. The R software package for ClimPACT2 is available on GitHub at https://github.com/ARCCSS-extremes/climpact2. Data analysis and graphics visualization throughout this study are mostly done by using NCAR Command Language (NCL) Version 6.6.2 (http://dx.doi.org/10.5065/D6WD3XH5) software (Meier-fleischer et al., 2017).

Reference (86)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return