• Bootsma, H. A., Bootsma M. J. , and Hecky R. E. , 1996: The chemical composition of precipitation and its significance to the nutrient budget of Lake Malawi. The Limnology, Climatology and Paleoclimatology of the East African Lakes, T. C. Johnson and E. O. Odada, Eds., Gordon and Breach, 251–265.

  • Calder, I. R., Hall R. L. , and Bastable H. G. , 1995: The impact of land use change on water resources in sub-Saharan Africa: A modeling study of Lake Malawi. J. Hydrol., 170, 123135, doi:10.1016/0022-1694(94)02679-6.

    • Search Google Scholar
    • Export Citation
  • Crétaux, J. F., and Coauthors, 2011: SOLS: A lake database to monitor in near real time water level and storage variations from remote sensing data. Adv. Space Res.,47, 1497–1507, doi:10.1016/j.asr.2011.01.004.

  • Dai, A., 2011: Characteristics and trends in various forms of the Palmer drought severity index during 1900–2008. J. Geophys. Res., 116, D12115, doi:10.1029/2010JD015541.

    • Search Google Scholar
    • Export Citation
  • Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, doi:10.1002/qj.828.

    • Search Google Scholar
    • Export Citation
  • Drayton, R. S., 1984: Variations in the level of Lake Malawi. Hydrol. Sci., 29, 110, doi:10.1080/02626668409490918.

  • Glad, P. A., 2010. Meteorological and hydrological conditions leading to severe regional drought in Malawi. M.Sc. thesis, Department of Geoscience, University of Oslo, 129 pp.

  • Grimes, D. I. F., and Diop M. , 2003: Satellite-based rainfall estimation for river flow forecasting in Africa: Rainfall estimates and hydrological forecasts. Hydrol. Sci. J., 48, 567584, doi:10.1623/hysj.48.4.567.51410.

    • Search Google Scholar
    • Export Citation
  • Hargreaves, G. H., and Samani Z. A. , 1982: Estimating potential evaptranspiration. J. Irrig. Drain. Div., 108, 225230.

  • Harris, I., Jones P. D. , Osborn T. J. , and Lister D. H. , 2014: Updated high-resolution grids of monthly climatic observations—The CRU TS3.10 dataset. Int. J. Climatol., 34, 623–642, doi:10.1002/joc.3711.

    • Search Google Scholar
    • Export Citation
  • Huete, A., Didan K. , Miura T. , Rodriguez E. P. , Gao X. , and Ferreira L. G. , 2002: Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sens. Environ., 83, 195213, doi:10.1016/S0034-4257(02)00096-2.

    • Search Google Scholar
    • Export Citation
  • Jury, M. R., and Gwazantini M. E. , 2002: Climate variability in Malawi. Part 2: Sensitivity and prediction of lake levels. Int. J. Climatol., 22, 13031312, doi:10.1002/joc.772.

    • Search Google Scholar
    • Export Citation
  • Jury, M. R., and Mwafulirwa N. D. , 2002: Climate variability in Malawi. Part 1: Dry summers, statistical associations and predictability. Int. J. Climatol., 22, 12891302, doi:10.1002/joc.771.

    • Search Google Scholar
    • Export Citation
  • Kidd, C. H. R., 1983: A water resources evaluation of Lake Malawi and the Shire River. World Meteorological Organization, UNDP-MLW/77/012, Geneva, Switzerland, 132 pp.

  • Kumambala, P. G., and Ervine A. , 2010: Water balance model of Lake Malawi and its sensitivity to climate change. Hydrol. J., 4, 152162.

    • Search Google Scholar
    • Export Citation
  • Mandeville, A. N., and Batchelor C. H. , 1990: Estimation of evapotranspiration in Malawi. Institute of Hydrology, 110 pp.

  • Meng, J., Yang R. , Wei H. , Ek M. , Gayno G. , Xie P. , and Mitchell K. , 2012: The land surface analysis in the NCEP Climate Forecast System reanalysis. J. Hydrometeor., 13, 16211630, doi:10.1175/JHM-D-11-090.1.

    • Search Google Scholar
    • Export Citation
  • Neuland, H., 1984: Abnormal high water levels of Lake Malawi—An attempt to assess the future behaviour of the lake water levels. Geogr. J., 9, 323334.

    • Search Google Scholar
    • Export Citation
  • Novella, N. S., and Thiaw W. M. , 2013: African rainfall climatology version 2 for famine early warning systems. J. Appl. Meteor. Climatol., 52, 588606, doi:10.1175/JAMC-D-11-0238.1.

    • Search Google Scholar
    • Export Citation
  • Palamuleni, L. G., and Annegarn H. J. , 2011: Hydrological response to predicted land cover change in the upper Shire River catchment, Malawi. J. Hum. Ecol., 36, 4352.

    • Search Google Scholar
    • Export Citation
  • Potter, C., Zhang P. , Klooster S. , Genovese V. , Shekhar S. , and Kumar V. , 2004: Understanding controls on historical river discharge in the world’s largest drainage basins. Earth Interact., 8, 121, doi:10.1175/1087-3562(2004)008<0001:UCOHRD>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Rodell, M., and Coauthors, 2004: The Global Land Data Assimilation System. Bull. Amer. Meteor. Soc., 85, 381394, doi:10.1175/BAMS-85-3-381.

    • Search Google Scholar
    • Export Citation
  • Saha, S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 10151057, doi:10.1175/2010BAMS3001.1.

    • Search Google Scholar
    • Export Citation
  • Schneider, U., Fuchs T. , Meyer-Christoffer A. , and Rudolf B. , 2008: Global precipitation analysis products of the GPCC. Global Precipitation Climatology Centre Rep., 12 pp.

  • Schneider, U., Becker A. , Finger P. , Meyer-Christoffer A. , Ziese M. , and Rudolf B. , 2014: GPCC’s new land surface precipitation climatology based on quality-controlled in situ data and its role in quantifying the global water cycle. Theor. Appl. Climatol., 115, 15–40, doi:10.1007/s00704-013-0860-x.

    • Search Google Scholar
    • Export Citation
  • Sene, K. J., 1998: Effect of variations in net basin supply on lake levels and outflows. Hydrol. Processes, 12, 559373, doi:10.1002/(SICI)1099-1085(19980330)12:4<559::AID-HYP594>3.0.CO;2-R.

    • Search Google Scholar
    • Export Citation
  • Servat, E., Hughes D. , Fritsch J. M. , and Hulme M. , 1998: Water resources variability in Africa during the 20th century. IAHS Publ.,252, 97–106.

  • Shela, O. N., 2010: Naturalisation of Lake Malawi levels and Shire River flows: Challenges of water resources research and sustainable utilisation of the Lake Malawi-Shire River system. First WARFSA/WaterNet Symp. on Sustainable Use of Water Resources, Maputo, Mozambique, WARFSA/WaterNet, 1–12. [Available online at http://www.waternetonline.ihe.nl/downloads/uploads/symposium/mozambique-2000/SHELA.PDF.]

  • Sutcliffe, J. V., and Knott D. G. , 1987: Historical variations in African water resources. IAHS Publ.,168, 463–476. [Available online at http://iahs.info/uploads/dms/iahs_168_0463.pdf.]

  • Tapley, B. D., Bettadpur S. , Watkins M. , and Reigber C. , 2004: The gravity recovery and climate experiment: Mission overview and early results. Geophys. Res. Lett., 31, L09607, doi:10.1029/2004GL019920.

    • Search Google Scholar
    • Export Citation
  • Torrance, J. D., 1972: Malawi, Rhodesia and Zambia. Climates of Africa, J. F. Griffiths, Ed., World Survey of Climatology, Vol. 10, Elsevier, 409–460.

  • Velpuri, N. M., Senay G. B. , and Asante K. O. , 2012: A multi-source satellite data approach for modelling Lake Turkana water level: Calibration and validation using satellite altimetry data. Hydrol. Earth Syst. Sci., 16, 118, doi:10.5194/hess-16-1-2012.

    • Search Google Scholar
    • Export Citation
  • Wells, N., Goddard S. , and Hayes M. J. , 2004: A self-calibrating Palmer drought severity index. J. Climate, 17, 23352351, doi:10.1175/1520-0442(2004)017<2335:ASPDSI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Characteristics of the Shire Valley: (a) mean ARCv2 rainfall at 10-km resolution. (b) Topography with place names, white outline is Malawi border. Data is from mean 2002–12 MODIS satellite. (c) Vegetation fraction and (d) daytime surface temperature at 1-km resolution. The dashed box in (a) is the Shire Valley “local” area.

  • View in gallery

    Mean annual cycle of local area (a) rainfall, runoff, latent heat flux, and vegetation; (b) potential evaporation and sensible heat flux; and (c) river flows and lake level. Upper and lower 2.5% averages are given for rainfall and river flow. (d) Correlation of annual cycle between climate indices at 2-month lead and lake level and its outflow (Mangochi); r > 0.6 significant.

  • View in gallery

    (a) Time series of station flows and satellite data. (b) Composite flow anomalies and 12-month sum of PDSI with linear trends. Respective wavelet spectral energy in (c) annual cycle and (d) multiannual fluctuations (period is years, cone of validity indicated by the stepped line). Spectral power is shaded from 90% confidence upward to max value.

  • View in gallery

    Correlation of March–May Shire River flow with December–February: (a) GPCC rainfall and (b) CFS latent heat flux; 500-m contour defines river channel. (c) Correlation of climate indices at successive downstream river gauges. (d) Lag correlation of monthly river flow and PDSI; r > 0.25 significant.

  • View in gallery

    Correlation of basin-averaged detrended PDSI in December–March season with global ECMWF: (a) surface temperature, (b) sea level pressure, and (c) 850-mb streamfunction with arrows highlighting rotational winds. Patterns correspond with wet seasons; r > 0.25 significant.

  • View in gallery

    (a) Daily discharge at Mangochi and Chikwawa: arrow is transition to manipulated base flow. Composite flood maps of day 2: (b) satellite OLR anomaly (W m−2); (c) 850-mb wind (largest vector is 3 m s−1, with vertical motion exaggerated); (d) vertical slice on 35°E of humidity (shaded), meridional–vertical motion (largest vector is 3 m s−1, with vertical motion exaggerated); (e) 850-mb streamfunction anomaly (105 m2 s−1); and (f) 850-mb geopotential height anomaly (m).

  • View in gallery

    (a) Daily delta flow at Mangochi compared with area rainfall, (b) satellite rainfall (mm h−1) on 29 Jan 2008, and (c) 850-mb winds. (d) Flooded homesteads along the Shire River.

  • View in gallery

    (a) Daytime land surface temperatures during the period 24–31 Oct 2011 from MODIS satellite, and (b) corresponding 700-mb winds from NCEP-CFS reanalysis.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 587 373 16
PDF Downloads 426 252 16

Malawi’s Shire River Fluctuations and Climate

View More View Less
  • 1 Physics Department, University of Puerto Rico, Mayagüez, Puerto Rico, and University of Zululand, KwaDlangezwa, South Africa
Full access

Abstract

Hydrological fluctuations of Malawi’s Shire River and climatic drivers are studied for a range of time and space scales. The annual cycles of basin rainfall and river flow peak in summer and autumn, respectively. Satellite and model products at <50-km resolution resolve the water deficit in this narrow valley. The leading climate index fitting Shire River flow anomalies is the Climatic Research Unit (CRU) Palmer drought severity index, based on interpolated gauge rainfall minus Penman–Monteith potential evapotranspiration. Climate variables anticipate lake level changes by 2 months, while weather variables anticipate river flow surges by 2 days. Global climate patterns related to wet years include a Pacific La Niña cool phase and low pressure over northeastern Africa. Shire River floods coincide with a cyclonic looping wind pattern that amplifies the equatorial trough and draws monsoon flow from Tanzania. Hot spells are common in spring: daytime surface temperatures can reach 60°C causing rapid desiccation. An anticyclonic high pressure cell promotes evaporation losses of ~20 mm day−1 over brief periods. Flood and drought in Malawi are shown to be induced by the large-scale atmospheric circulation and rainfall in the surrounding highlands. Hence, early warning systems should consider satellite and radar coverage of the entire basin.

Corresponding author address: Mark R. Jury, Physics Department, University of Puerto Rico, Rt. 108, Mayagüez, 00681 Puerto Rico. E-mail: mark.jury@upr.edu

Abstract

Hydrological fluctuations of Malawi’s Shire River and climatic drivers are studied for a range of time and space scales. The annual cycles of basin rainfall and river flow peak in summer and autumn, respectively. Satellite and model products at <50-km resolution resolve the water deficit in this narrow valley. The leading climate index fitting Shire River flow anomalies is the Climatic Research Unit (CRU) Palmer drought severity index, based on interpolated gauge rainfall minus Penman–Monteith potential evapotranspiration. Climate variables anticipate lake level changes by 2 months, while weather variables anticipate river flow surges by 2 days. Global climate patterns related to wet years include a Pacific La Niña cool phase and low pressure over northeastern Africa. Shire River floods coincide with a cyclonic looping wind pattern that amplifies the equatorial trough and draws monsoon flow from Tanzania. Hot spells are common in spring: daytime surface temperatures can reach 60°C causing rapid desiccation. An anticyclonic high pressure cell promotes evaporation losses of ~20 mm day−1 over brief periods. Flood and drought in Malawi are shown to be induced by the large-scale atmospheric circulation and rainfall in the surrounding highlands. Hence, early warning systems should consider satellite and radar coverage of the entire basin.

Corresponding author address: Mark R. Jury, Physics Department, University of Puerto Rico, Rt. 108, Mayagüez, 00681 Puerto Rico. E-mail: mark.jury@upr.edu

1. Introduction

Lake Malawi is located in the African Rift Valley (center: 13°S, 34.5°E) with a basin area of 150 000 km2. The lake is deep (>300 m) with volume of ~8.4 × 109 m3, an area of 29 000 km2, and is 600 km north–south by ~40 km east–west at an elevation of 474 m. There are four major rivers providing inflow to the lake, while the Shire River is the southern outlet extending 400 km from Mangochi (14.5°S) to the Zambezi River. Over the first 130 km, the river meanders across a plateau dropping only 7 m (by 15.4°S). Over the next 80 km (to 16.0°S), there is a fall of 370 m offering hydropower yield. Thereafter the Shire River runs through lowlands, joined by the Ruo River (50 m at 16.6°S) and onto the confluence with the Zambezi.

Lake Malawi level is maintained by annual gains from rainfall (1.3 m) and inflow (0.9 m) against annual losses by evaporation (1.8 m) and outflow to the Shire River (0.4 m) (Torrance 1972; Kidd 1983; Neuland 1984; Drayton 1984; Bootsma et al. 1996). Lake levels are sensitive to the cumulative effect of seasonal wet and dry spells (Servat et al. 1998; Jury and Gwazantini 2002; Kumambala and Ervine 2010). Rains are heaviest from December to February and the lake level crests at ~1 m from March to May (Shela 2010). Evaporation losses are widespread and high from September to November (Mandeville and Batchelor 1990). The Shire River outflow depends primarily on lake levels and secondarily on basin inflow and evaporation (Calder et al. 1995; Kumambala and Ervine 2010).

Lake Malawi level has been monitored for a century (Sutcliffe and Knott 1987). Changes occur at low frequency due to cumulative imbalances of basin rainfall and evaporation, and at high frequency due to local air pressure, meridional winds, and storm runoff. The interdecadal fluctuation of lake level is ~3 m, which is an issue for boat harbors designed for use above 474 m. Runoff is affected by declining forest cover and consequent sediment accumulation in the floodplain (Palamuleni and Annegarn 2011). A population growth rate of 3% (in 2010) puts pressure on the land; agricultural crops have replaced woodlands, and grasslands are overgrazed and sparse. Recent studies have quantified the reduction in forest cover via satellite vegetation fraction. The Shire River receives outflow from Lake Malawi via Mangochi due to topographic slope and water balance. Lake Malombe intersects the river in the upper catchment, while farther downstream there is inflow from the Ruo River.

Lake Malawi outflows to the Shire River are measured at Mangochi and downstream at intervals of about 100 km; flows vary from 50 to 1000 m3 s−1, averaging 480 m3 s−1 in the upper valley. The hydrology has been gradually regulated by low barrages and small reservoirs to support hydropower, particularly since 1993. Although they maintain base flow during spring and amplify lake levels (Sene 1998), the engineering structures are still overtopped by floods, as seen in the results below.

The objective of this paper is to study the climatic drivers of Malawi’s Shire River using direct measurements and model estimates, to analyze the annual to decadal variability, to understand the meteorological forcing of wet and dry spells, and to determine which atmospheric indices best correspond with the hydrology.

2. Data and methods

The southern half of Lake Malawi and Shire River Valley encompass an area 13°–17°S, 33.5°–36.5°E (Fig. 1). Discharge data from the state hydrology service were obtained and, following quality checks and naturalization according to Shela (2010), a single monthly time series 1953–2012 was reconstructed from the Mangochi record (1976+; + refers to the starting year up to 2012) extended by other gauges (Mbewe 1953+ and Liwonde 1965+). Daily values were also obtained at Mangochi and Chikwawa gauges (1976+) to describe flood events. Supplementing the in situ hydrological records are National Aeronautics and Space Administration (NASA) satellite altimetry of lake level (1992+) and gravity measurements of soil water fraction (Tapley et al. 2004; Crétaux et al. 2011; Velpuri et al. 2012) since 2002. Satellite–gauge interpolated rainfall observations in the study area (Fig. 1) at 50-km resolution derive from the Global Precipitation Climatology Centre (GPCC; Schneider et al. 2008, 2014) and the Climatic Research Unit (CRU; Harris et al. 2014). The Palmer drought severity index (PDSI) is used to characterize water budget anomalies, based on observed precipitation minus Penman–Monteith potential evapotranspiration (Hargreaves and Samani 1982; Wells et al. 2004; Dai 2011). Area-averaged latent and sensible heat fluxes and runoff estimated by the Global Land Data Assimilation System (GLDAS) Noah model are employed (Rodell et al. 2004). Additional datasets include the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite infrared surface temperatures and vegetation color fraction at 1-km resolution (Huete et al. 2002); African Rainfall Climatology, version 2 (ARCv2) satellite–gauge merged rainfall at 10-km resolution (Novella and Thiaw 2013); the National Centers for Environmental Prediction (NCEP) Climate Forecast System (CFS); and the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis climate fields at 30–70-km resolution (Saha et al. 2010; Dee et al. 2011; Meng et al. 2012). The data assimilations (GLDAS-Noah, NCEP-CFS, and ECMWF) have issues associated with inhomogeneous forcing. For example the precipitation used in Noah changes over time, while the CFS exhibits discontinuities due to changes in satellite inputs. Grimes and Diop (2003) show that basin-averaged satellite cloud temperature and vegetation fraction are proxies for the water balance and can be fitted to daily African streamflow records with good success.

Fig. 1.
Fig. 1.

Characteristics of the Shire Valley: (a) mean ARCv2 rainfall at 10-km resolution. (b) Topography with place names, white outline is Malawi border. Data is from mean 2002–12 MODIS satellite. (c) Vegetation fraction and (d) daytime surface temperature at 1-km resolution. The dashed box in (a) is the Shire Valley “local” area.

Citation: Journal of Hydrometeorology 15, 5; 10.1175/JHM-D-13-0195.1

To understand regional climate influences on Malawi’s Shire River flow, cross correlations were computed at various lags as maps and time series, similar to the methods of Glad (2010). For the annual cycle and consecutive months, the degrees of freedom is 10 and 60, respectively, so 95% significance is achieved at r > 0.60 and 0.25. Of many climate indices tested, the PDSI was closest to Shire River flow anomalies. Seasonal values of Shire basin-averaged detrended PDSI were cross correlated with ECMWF reanalysis fields (1957–2001) to understand the global climate forcing at interannual to decadal time scales.

Inspecting the daily flow discharge at Chikwawa, a threshold of 1200 m3 s−1 was exceeded on 13 occasions. For this group of floods, composite weather anomalies were calculated from reanalysis and satellite fields, and their structure was examined 2 days before peak flow. Hot spells were analyzed using MODIS satellite land surface temperatures. The warmest case was studied for local structure and regional weather pattern.

3. Results

a. Valley climate characteristics

Mean rainfall in the Shire Valley during the satellite era is 2–3 mm day−1 corresponding to elevations of 500–1000 m, respectively (Figs. 1a,b). A tongue of lower rainfall extends up the Shire Valley from the Zambezi River and its wetlands, while the surrounding Mozambique highlands receive heavier rainfall. Vegetation (Fig. 1c) is relatively homogeneous, having a mean annual color fraction of 0.3 in the Shire Valley, rising to 0.45 in the mountains and in the southeastern wetlands. Surface daytime temperature (Fig. 1d) exhibits a pattern consistent with rainfall and elevation with a mean annual value of 38°C in the Shire Valley indicative of high potential evaporation.

The Shire Valley (13°–17°S, 33.5°–36.5°E) climate exhibits a pronounced annual cycle (Fig. 2). Rainfall rises above 4 mm day−1 from December to March, generating peak runoff in February. In the May–October dry season, a high pressure prevails over southern Africa and rainfall is absent. The rainfall difference between the lower and upper 2.5 percentiles is greatest in February: 2–12 mm day−1. CFS latent heat flux and satellite vegetation index exhibit a lagged response to rainfall with maxima in March and minima in October (Fig. 2a).

Fig. 2.
Fig. 2.

Mean annual cycle of local area (a) rainfall, runoff, latent heat flux, and vegetation; (b) potential evaporation and sensible heat flux; and (c) river flows and lake level. Upper and lower 2.5% averages are given for rainfall and river flow. (d) Correlation of annual cycle between climate indices at 2-month lead and lake level and its outflow (Mangochi); r > 0.6 significant.

Citation: Journal of Hydrometeorology 15, 5; 10.1175/JHM-D-13-0195.1

Surface water losses in the Shire Valley are characterized by CRU, version 3 (CRU3), potential evapo-transpiration (Fig. 2b) and exhibit values near 3 mm day−1 from January to July. From September to November evaporation exceeds 5 mm day−1, so the rainy season starts with a surface water deficit. Sensible heat flux from Noah and CFS models follow the potential evapo-transpiration, while the latent heat flux follows vegetation (February peak; Fig. 2a) and stays below 3.5 mm day−1.

River flows averaged over three gauges are compared with lake levels in Fig. 2c. The mean annual cycle of discharge peaks at 650 m3 s−1 in April, 2 months after runoff and 1 month before lake level. Flows gradually decline to a minimum in November. Correlations between the mean annual cycle of climate and 2-month-lagged hydrology indices are illustrated in Fig. 2d. Satellite vegetation and CFS latent heat flux exhibit significant positive values, while ECMWF/CFS sensible heat flux and CRU3 potential evapotranspiration exhibit significant negative values.

b. Hydrology response to climate

Our reconstructed Lake Malawi record displays relatively stationary seasonal pulses during 1953–74 (Fig. 3a). The lake level rose to 477 m and river flow increased in 1978–82. Lake level dropped (473 m) following a prolonged drought in 1992–95 and river flow diminished until 1997. Heavier rainfall induced a rise up to 2003 and river flows have oscillated at ~10% above normal (480 m3 s−1) since then. The wavelet spectrum of river flow annual cycle (Fig. 3b) indicates high amplitude in 1953–65, 1976–78, and 1987–89, and low amplitude in 1966–72 and 1992–96. Annual cycling was again prominent in the recent decade.

Fig. 3.
Fig. 3.

(a) Time series of station flows and satellite data. (b) Composite flow anomalies and 12-month sum of PDSI with linear trends. Respective wavelet spectral energy in (c) annual cycle and (d) multiannual fluctuations (period is years, cone of validity indicated by the stepped line). Spectral power is shaded from 90% confidence upward to max value.

Citation: Journal of Hydrometeorology 15, 5; 10.1175/JHM-D-13-0195.1

Filtering the annual cycle, river flow anomalies (Fig. 3c) display oscillations that are in phase with the 12-month lagged sum of PDSI. There is good correspondence except when the PDSI anticipates flow decline in 1978–84 and rise in 1997–98. A puzzling feature is that the PDSI has a downtrend (more evaporation) while the river flow has an uptrend (possibly sedimentation). The wavelet spectrum of flow anomalies (Fig. 3d) indicates cycling at 6 years in 1953–70 and after 2003, and at 11 years during 1978–98 with a low-frequency harmonic at ~20 years. The 6-yr signal reflects ENSO forcing (Jury and Mwafulirwa 2002), while the 11-yr signal is linked with decadal fluctuations of global climate.

The spatial influence of the local climate is evaluated by correlating the March–May Shire flow with December–February gridded rainfall and CFS latent heat flux in the period 1979–2010. This indicates which parts of the basin participate in multiannual climate forcing. The GPCC rainfall correlation is >+0.4 along 14°S from 34° to 36°E (Fig. 4a) encompassing the southeastern edge of Lake Malawi. However, correlations are zero in the lower Shire Valley indicating that flow fluctuations are primarily driven by the lake water balance. The correlation with CFS latent heat flux exceeds rainfall (>+0.5) especially in the bordering highlands and upper (northern) valley (Fig. 4b).

Fig. 4.
Fig. 4.

Correlation of March–May Shire River flow with December–February: (a) GPCC rainfall and (b) CFS latent heat flux; 500-m contour defines river channel. (c) Correlation of climate indices at successive downstream river gauges. (d) Lag correlation of monthly river flow and PDSI; r > 0.25 significant.

Citation: Journal of Hydrometeorology 15, 5; 10.1175/JHM-D-13-0195.1

Considering lake level and downstream river gauges, the local climatic influences are studied by correlation in Fig. 4c. The cumulative PDSI affects Mangochi flow (r > +0.5) and other locations to a lesser extent. Valley rainfall and local runoff have weak negative influence at Mangochi but rise to positive correlations at Mbewe downstream. Similarly the evaporative losses within the valley have a weak negative influence that grows downstream (<−0.4). Hence, climatic conditions in the Shire Valley affect outflows to the Zambezi lowlands, while weather around the southern edge of Lake Malawi drives inflows. Evaluating the lead–lag influence of the basin-averaged monthly PDSI (Fig. 4d), it is evident that river flow lags the seasonal inputs (peaks at 3, 14, and 25 months) and cumulative outputs, similar to Potter et al. (2004) and Glad (2010).

Global climate forcing of basin hydrology is analyzed by correlation of the ECMWF reanalysis fields with detrended local PDSI in the December–March season (1957–2001). The surface temperature map (Fig. 5a) indicates a significant negative correlation in the Malawi Shire basin, confirming that lower temperatures coincide with precipitation–evaporation surplus. In the Pacific, there is a La Niña signal: cool east (r < −0.3) warm west (r > +0.3) dipole. The Atlantic and Indian Oceans have mixed temperature signals. Yet there is a region of lower pressure (r < −0.3) over northeastern Africa that extends to the North Atlantic and eastern Indian Ocean (Fig. 5b). Over the eastern Pacific the ECMWF pressure correlation with Shire PDSI is positive, indicating high pressure associated with La Niña. The 850-mb streamfunction reflects the rotational circulation (Fig. 5c) and is locally cyclonic, as expected. Thus, equatorial westerlies and subtropical easterlies boost rainfall over Malawi. A key global feature at the decadal time scale is opposing correlations in the northeastern and southeastern Pacific where anticyclonic winds prevail. The large-scale thermodynamic structure and circulation reflecting the low-frequency component of El Niño–Southern Oscillation (Jury and Mwafulirwa 2002) influences Shire River flow, as represented by the PDSI.

Fig. 5.
Fig. 5.

Correlation of basin-averaged detrended PDSI in December–March season with global ECMWF: (a) surface temperature, (b) sea level pressure, and (c) 850-mb streamfunction with arrows highlighting rotational winds. Patterns correspond with wet seasons; r > 0.25 significant.

Citation: Journal of Hydrometeorology 15, 5; 10.1175/JHM-D-13-0195.1

c. Short-term events

Floods are studied using the daily discharge at Mangochi and Chikwawa at ~200 km downstream (Fig. 6a). The two have similar annual cycles up to 1993, whereafter the base flow is manipulated for hydroelectric yield. Flood events (>1200 m3 s−1) punctuate much of the 32-yr record at Chikwawa due to valley runoff on top of lake outflow. Storm features are analyzed as a composite group (13 cases) two days before peak flow (day 2). Satellite convection [outgoing longwave radiation (OLR)] appears as a northwest–southeast axis over the Shire Valley (Fig. 6b), having developed locally since day 4. Low-level winds make a clockwise loop over Mozambique (Fig. 6c), joined by northerlies from Tanzania and easterlies from Madagascar. There is a composite high pressure over South Africa causing easterly winds there. The meridional overturning Hadley circulation is moist and convergent over Malawi (Fig. 6d), while dry conditions prevail at 0° and 25°S. The streamfunction anomaly is cyclonic in a narrow axis following the Zambezi Valley, and a low pressure is situated over Malawi on day 2 (Figs. 6e,f). Station rainfall in some cases reached 89 mm day−1, and tropical cyclones were prevalent near Madagascar as seen in the following case.

Fig. 6.
Fig. 6.

(a) Daily discharge at Mangochi and Chikwawa: arrow is transition to manipulated base flow. Composite flood maps of day 2: (b) satellite OLR anomaly (W m−2); (c) 850-mb wind (largest vector is 3 m s−1, with vertical motion exaggerated); (d) vertical slice on 35°E of humidity (shaded), meridional–vertical motion (largest vector is 3 m s−1, with vertical motion exaggerated); (e) 850-mb streamfunction anomaly (105 m2 s−1); and (f) 850-mb geopotential height anomaly (m).

Citation: Journal of Hydrometeorology 15, 5; 10.1175/JHM-D-13-0195.1

Although discharge at Mangochi is less variable due to the buffering effects of Lake Malawi, daily surges can occur: 141 m3 s−1 on 1 February 2008 (Fig. 7a). The 850-mb wind map of 29 January exhibited a pair of cyclones to the east and west of Malawi, which caused a line of thunderstorms to move over the lake (Figs. 7b,c). The ECMWF reanalysis was consistent with composite satellite rainfall, and CFS hourly rainfall exhibited strong diurnal cycling. There was a 2-day response lag, affording ample time for early warning. Yet rural settlers often encroach onto floodplains in the dry season, and then become victims when waters rise (Fig. 7d).

Fig. 7.
Fig. 7.

(a) Daily delta flow at Mangochi compared with area rainfall, (b) satellite rainfall (mm h−1) on 29 Jan 2008, and (c) 850-mb winds. (d) Flooded homesteads along the Shire River.

Citation: Journal of Hydrometeorology 15, 5; 10.1175/JHM-D-13-0195.1

Springtime dry spells are characterized by heat waves and dessication. In Fig. 8a, the Shire Valley daytime surface temperatures are mapped from MODIS satellite infrared data during the period 24–31 October 2011. Temperatures of 60°C were observed in the central valley and CFS potential evaporation rates reached ~20 mm day−1. The heat wave was induced by subsiding air from an anticyclonic high pressure cell located south of Malawi (Fig. 8b). Satellite soil water fraction declined sharply in November 2011, and satellite altimetry recorded a 0.12 m month−1 drop in Lake Malawi (cf. Fig. 3a). Such dessication is common in the Shire Valley just before the rainy season.

Fig. 8.
Fig. 8.

(a) Daytime land surface temperatures during the period 24–31 Oct 2011 from MODIS satellite, and (b) corresponding 700-mb winds from NCEP-CFS reanalysis.

Citation: Journal of Hydrometeorology 15, 5; 10.1175/JHM-D-13-0195.1

4. Discussion and summary

This study has analyzed the drivers of Malawi’s Shire River flow. Past work on basin water budgets typically use an equation for delta storage comprising gains from precipitation and inflow, losses from evaporation and outflow, and time integration for volumetric response. Here the basin rainfall, lake level, and outflow are known. Actual evaporation as represented by latent heat flux has a value of ~0.7 m yr−1 and correlates positively with river flow (cf. Figs. 2d and 4b). Latent heat flux moistens the atmospheric boundary layer, enhancing diurnal convection (cf. Fig. 7a). Surface water losses are characterized by potential evapotranspiration (~1.4 m yr−1) and sensible heat flux, which correlate negatively with river flow (cf. Figs. 2b,d).

A statistical analysis of relationships between climate and hydrology at various time and space scales yielded a number of findings:

  • Direct measurements of lake level and river flow help quantify climatic terms in the water budget at seasonal to multiannual time scales.

  • Modern satellite and model reanalysis products at 50-km resolution have the ability to resolve hydrological fluctuations in the African Rift Valley.

  • The leading climate index fitting Shire River flow is the CRU self-calibrating Palmer drought severity index, based on interpolated gauge rainfall minus Penman–Monteith-calculated potential evapotranspiration.

  • Global climate variables anticipate lake level changes by 2 months. Regional weather variables anticipate river flow surges by 2 days.

  • Global climate patterns related to local precipitation–evaporation surplus include an eastern Pacific cool phase La Niña, low pressure over northeastern Africa, and rotational winds.

  • Shire River floods coincide with a cyclonic looping wind pattern that amplifies the equatorial trough and draws monsoon flow from Tanzania. In some cases there is a tropical cyclone near Madagascar.

  • Hot dry spells are a recurring feature of spring: daytime surface temperatures reach 60°C causing rapid desiccation. The anticyclonic high pressure cell and its subsidence lead to evaporation losses >10 mm day−1 over brief periods.

If flood and drought in Malawi are induced by the large-scale atmospheric circulation and rainfall in the surrounding highlands, then a narrow focus on Shire Valley impacts will not support early warning. Monitoring by satellite and radar is needed. Weather forecast accuracy could improve with transmission of Aircraft Meteorological Data Relay (AMDAR) profiles from regional airports. Further progress will require a resumption of daily Shire River flow gauging and timely reporting to international centers for operational data assimilation. Further research could include an analysis of climate change using phase 5 of the Coupled Model Intercomparison Project (CMIP5) projections. It is believed that the Shire River has reached its limit of manipulation and abstraction. Soil conservation is needed to control sedimentation and keep the growing rural population on a sustainable socioeconomic path.

Acknowledgments

The author thanks L. Palamuleni of Northwest University, South Africa, and M. Kleynhans of Aurecon, South Africa, for provision of daily Shire River flow and station rainfall. Satellite and model interpolated data were analyzed from the Climate Explorer, IRI Climate Library, NASA Giovanni, JPL-UC Grace, and USDA altimetry websites.

REFERENCES

  • Bootsma, H. A., Bootsma M. J. , and Hecky R. E. , 1996: The chemical composition of precipitation and its significance to the nutrient budget of Lake Malawi. The Limnology, Climatology and Paleoclimatology of the East African Lakes, T. C. Johnson and E. O. Odada, Eds., Gordon and Breach, 251–265.

  • Calder, I. R., Hall R. L. , and Bastable H. G. , 1995: The impact of land use change on water resources in sub-Saharan Africa: A modeling study of Lake Malawi. J. Hydrol., 170, 123135, doi:10.1016/0022-1694(94)02679-6.

    • Search Google Scholar
    • Export Citation
  • Crétaux, J. F., and Coauthors, 2011: SOLS: A lake database to monitor in near real time water level and storage variations from remote sensing data. Adv. Space Res.,47, 1497–1507, doi:10.1016/j.asr.2011.01.004.

  • Dai, A., 2011: Characteristics and trends in various forms of the Palmer drought severity index during 1900–2008. J. Geophys. Res., 116, D12115, doi:10.1029/2010JD015541.

    • Search Google Scholar
    • Export Citation
  • Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, doi:10.1002/qj.828.

    • Search Google Scholar
    • Export Citation
  • Drayton, R. S., 1984: Variations in the level of Lake Malawi. Hydrol. Sci., 29, 110, doi:10.1080/02626668409490918.

  • Glad, P. A., 2010. Meteorological and hydrological conditions leading to severe regional drought in Malawi. M.Sc. thesis, Department of Geoscience, University of Oslo, 129 pp.

  • Grimes, D. I. F., and Diop M. , 2003: Satellite-based rainfall estimation for river flow forecasting in Africa: Rainfall estimates and hydrological forecasts. Hydrol. Sci. J., 48, 567584, doi:10.1623/hysj.48.4.567.51410.

    • Search Google Scholar
    • Export Citation
  • Hargreaves, G. H., and Samani Z. A. , 1982: Estimating potential evaptranspiration. J. Irrig. Drain. Div., 108, 225230.

  • Harris, I., Jones P. D. , Osborn T. J. , and Lister D. H. , 2014: Updated high-resolution grids of monthly climatic observations—The CRU TS3.10 dataset. Int. J. Climatol., 34, 623–642, doi:10.1002/joc.3711.

    • Search Google Scholar
    • Export Citation
  • Huete, A., Didan K. , Miura T. , Rodriguez E. P. , Gao X. , and Ferreira L. G. , 2002: Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sens. Environ., 83, 195213, doi:10.1016/S0034-4257(02)00096-2.

    • Search Google Scholar
    • Export Citation
  • Jury, M. R., and Gwazantini M. E. , 2002: Climate variability in Malawi. Part 2: Sensitivity and prediction of lake levels. Int. J. Climatol., 22, 13031312, doi:10.1002/joc.772.

    • Search Google Scholar
    • Export Citation
  • Jury, M. R., and Mwafulirwa N. D. , 2002: Climate variability in Malawi. Part 1: Dry summers, statistical associations and predictability. Int. J. Climatol., 22, 12891302, doi:10.1002/joc.771.

    • Search Google Scholar
    • Export Citation
  • Kidd, C. H. R., 1983: A water resources evaluation of Lake Malawi and the Shire River. World Meteorological Organization, UNDP-MLW/77/012, Geneva, Switzerland, 132 pp.

  • Kumambala, P. G., and Ervine A. , 2010: Water balance model of Lake Malawi and its sensitivity to climate change. Hydrol. J., 4, 152162.

    • Search Google Scholar
    • Export Citation
  • Mandeville, A. N., and Batchelor C. H. , 1990: Estimation of evapotranspiration in Malawi. Institute of Hydrology, 110 pp.

  • Meng, J., Yang R. , Wei H. , Ek M. , Gayno G. , Xie P. , and Mitchell K. , 2012: The land surface analysis in the NCEP Climate Forecast System reanalysis. J. Hydrometeor., 13, 16211630, doi:10.1175/JHM-D-11-090.1.

    • Search Google Scholar
    • Export Citation
  • Neuland, H., 1984: Abnormal high water levels of Lake Malawi—An attempt to assess the future behaviour of the lake water levels. Geogr. J., 9, 323334.

    • Search Google Scholar
    • Export Citation
  • Novella, N. S., and Thiaw W. M. , 2013: African rainfall climatology version 2 for famine early warning systems. J. Appl. Meteor. Climatol., 52, 588606, doi:10.1175/JAMC-D-11-0238.1.

    • Search Google Scholar
    • Export Citation
  • Palamuleni, L. G., and Annegarn H. J. , 2011: Hydrological response to predicted land cover change in the upper Shire River catchment, Malawi. J. Hum. Ecol., 36, 4352.

    • Search Google Scholar
    • Export Citation
  • Potter, C., Zhang P. , Klooster S. , Genovese V. , Shekhar S. , and Kumar V. , 2004: Understanding controls on historical river discharge in the world’s largest drainage basins. Earth Interact., 8, 121, doi:10.1175/1087-3562(2004)008<0001:UCOHRD>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Rodell, M., and Coauthors, 2004: The Global Land Data Assimilation System. Bull. Amer. Meteor. Soc., 85, 381394, doi:10.1175/BAMS-85-3-381.

    • Search Google Scholar
    • Export Citation
  • Saha, S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 10151057, doi:10.1175/2010BAMS3001.1.

    • Search Google Scholar
    • Export Citation
  • Schneider, U., Fuchs T. , Meyer-Christoffer A. , and Rudolf B. , 2008: Global precipitation analysis products of the GPCC. Global Precipitation Climatology Centre Rep., 12 pp.

  • Schneider, U., Becker A. , Finger P. , Meyer-Christoffer A. , Ziese M. , and Rudolf B. , 2014: GPCC’s new land surface precipitation climatology based on quality-controlled in situ data and its role in quantifying the global water cycle. Theor. Appl. Climatol., 115, 15–40, doi:10.1007/s00704-013-0860-x.

    • Search Google Scholar
    • Export Citation
  • Sene, K. J., 1998: Effect of variations in net basin supply on lake levels and outflows. Hydrol. Processes, 12, 559373, doi:10.1002/(SICI)1099-1085(19980330)12:4<559::AID-HYP594>3.0.CO;2-R.

    • Search Google Scholar
    • Export Citation
  • Servat, E., Hughes D. , Fritsch J. M. , and Hulme M. , 1998: Water resources variability in Africa during the 20th century. IAHS Publ.,252, 97–106.

  • Shela, O. N., 2010: Naturalisation of Lake Malawi levels and Shire River flows: Challenges of water resources research and sustainable utilisation of the Lake Malawi-Shire River system. First WARFSA/WaterNet Symp. on Sustainable Use of Water Resources, Maputo, Mozambique, WARFSA/WaterNet, 1–12. [Available online at http://www.waternetonline.ihe.nl/downloads/uploads/symposium/mozambique-2000/SHELA.PDF.]

  • Sutcliffe, J. V., and Knott D. G. , 1987: Historical variations in African water resources. IAHS Publ.,168, 463–476. [Available online at http://iahs.info/uploads/dms/iahs_168_0463.pdf.]

  • Tapley, B. D., Bettadpur S. , Watkins M. , and Reigber C. , 2004: The gravity recovery and climate experiment: Mission overview and early results. Geophys. Res. Lett., 31, L09607, doi:10.1029/2004GL019920.

    • Search Google Scholar
    • Export Citation
  • Torrance, J. D., 1972: Malawi, Rhodesia and Zambia. Climates of Africa, J. F. Griffiths, Ed., World Survey of Climatology, Vol. 10, Elsevier, 409–460.

  • Velpuri, N. M., Senay G. B. , and Asante K. O. , 2012: A multi-source satellite data approach for modelling Lake Turkana water level: Calibration and validation using satellite altimetry data. Hydrol. Earth Syst. Sci., 16, 118, doi:10.5194/hess-16-1-2012.

    • Search Google Scholar
    • Export Citation
  • Wells, N., Goddard S. , and Hayes M. J. , 2004: A self-calibrating Palmer drought severity index. J. Climate, 17, 23352351, doi:10.1175/1520-0442(2004)017<2335:ASPDSI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
Save