By switching to dark mode you can reduce the energy consumption of our digital service.

Changes in Growing Season over Southern Africa

Multiple Authors
The following is a summarised version of the paper by Tadross, M. et al. Changes in growing-season rainfall characteristics and downscaled scenarios of change over southern Africa: implications for growing maize. The full paper can be downloaded here


Study Area: Stations marked by pins. Background image courtesy of NASA earth observatory

Africa is considered highly vulnerable to climate change (FAO, 2004), largely because many socio-economic activities in Africa, particularly agriculture, depend on climate and especially rainfall. Consequently, climatic variations and change have an impact on the productivity of many socio-economic activities (Obasi, 2005). Within the agriculture sector drought is arguably the most important climatic challenge and has major impacts on rural livelihoods (Buckland et al., 2000). In the 1991/92 drought alone, cereal output in the Southern African Development Community (SADC) region (excluding South Africa) fell from an average of 11.3 million to 6.2 million tonnes, necessitating extensive food imports (FAO, 2004). Furthermore, projections of future change place southern Africa’s agriculture sector at the forefront of climate change vulnerability with potential negative impacts on revenue from dryland farming (Kurukulasuriya et al., 2006). There is therefore a pressing need to assess the capacity of southern African countries to cope with, and adapt to, the impact of climate change.

A major problem for any adaptation initiative, particularly at the local level is insufficient information about what to adapt to. This often results from a lack of awareness of how the climate has changed in the past (and how social systems have responded), how it is expected to change in the future and if observed changes are consistent with projected changes based on models of the climate system (e.g. Hewitson and Crane, 2006; Jones et al., 2004; Tadross et al., 2005). This is further complicated by the need to interpret changes from meteorological analyses (typically expressed as monthly or seasonal averages) in terms of an impact on a particular sector. For the agricultural sector, which is the main source of livelihood for the majority of the region’s population, it is crucial to understand how rainfall characteristics are changing, particularly those affecting planting dates and the crop growth cycle e.g. the start of the rains or frequency and intensity of dry spells and daily rainfall, that can destroy crops if they occur at critical stages of plant growth (Dennet, 1987; Ati et al., 2002; Usman et al., 2005).Besides long-term change, year-to-year variability and its effect on farm management practices is also crucial. The supposition of this shorter-term climate variability on longer term change, within a region that operates close to critical thresholds, is key to formulating adequate adaptation options. Whilst aspects of year-to-year climate variability, such as the El-Nino Southern Oscillation (ENSO) and Antarctic Oscillation (AAO), have a well documented relationship with seasonal climate anomalies in the region (e.g. Reason and Rouault, 2002; Richard et al., 2000; Lindesay, 1988; Mason, 1995; Hulme et al., 2001; Reason and Rouault, 2005), relatively fewer studies have investigated relationships with rainfall characteristics that are closely tied to management decisions and the crop growth cycle (e.g. Usman and Reason, 2004; Tadross et al., 2005; Reason et al., 2005). Certainly, within southern Africa little is currently known regarding how climate change and trends may be affecting daily rainfall characteristics at the local level and for particular times in the seasonal cycle (see New et al., 2006 for a comprehensive assessment of changes in extremes). Such knowledge can help raise awareness about the climate; support sectoral and integrated impact assessments of climate change, lead to improved seasonal forecasting capabilities, and improve the effectiveness of adaptation options in the agriculture sector.

This paper explores observed changes in daily rainfall records and evaluates the agricultural implications of these changes. Indices were developed to represent planting dates, rainfall cessation, dry spell length, frequency of dry days, rainfall intensity and total rainfall during critical stages of the crop growth cycle. As maize is the main SADC food/nutrition source (Smale and Jayne, 2003) we focus on indices relevant to maize cropping. To understand how variations in seasonal rainfall patterns are linked to large-scale climate modes, we relate variability in these indices to ENSO and AAO and calculate trends in these indices using daily rainfall data from 104 stations across Malawi, Mozambique, Zambia and Zimbabwe. Downscaled climate scenarios for the late 21st century from both Regional Climate Models (RCMs) and empirical approaches are used to suggest where and when the scenarios may be consistent with each other and with the observed trends.

Some points of relevance for adaptation

  • The paper examines in a detailed way the onset, cessation and consistency of the growing season for southern Africa over the period 1960-2005.
  • The paper demonstrates well that an intelligent analysis of historic meteorological data can provide useful information on trends in agriculturally important variables
  • Combining the meteorological analysis with relevant information about the growing requirements of different crops (in this case maize) provides allows the identification of regions which are close to the critical threshold of rainfall required to grow certain cultivars.
  • Downscaled climate data can be used to examine changes in agriculturally relevant climate variables to see where agricultural tresholds may be crossed in the future.
  • Combining information on the trends in climate with information on expected future changes can inform adaptation decision-making.
  • In many cases it is the current trend in climate which is most important for adaptation, as future changes in rainfall may not occur for 40-50 years. This is not to say, however, that the longer-term projections should be ignored.


We have calculated indices designed to measure aspects of intra-seasonal rainfall variability related to maize cropping over southeastern southern Africa. This analysis has proceeded from three definitions of planting dates and a single definition of rainfall cessation. It has been demonstrated that there have been weak trends for later planting and earlier cessation dates in the north, which has generally led to shorter rainfall seasons This is especially significant over southern Zambia as the duration of the rainfall season is close to critical thresholds (for planting 130-day maize cultivars), which is an incentive for farmers to plant as early possible. Trends between 1960 and 2005 for shortened rainfall seasons at Livingstone station show that although trends are insignificant they lead to more frequent failure to cross critical thresholds of seasonal duration required for maize cultivation. Climate variability, through both the AAO and ENSO (SOI), is also demonstrated to be an important additional pressure, altering the character of the cropping season in fundamental ways. Over much of the region a negative SOI (El-Nino) is associated with an early start to the season (using Planting criteria A), though not necessarily with consistent rain – a direct result of an increase in the number of dry days and mean dry spell length. Generally, there is also an early Cessation, which places added strain on farmers who are now faced with a shorter season of less consistent rain, which the temptation to plant early. Under these circumstances it is easy to see how crop failure is a regular occurrence during El-Nino. With the thresholds in seasonal duration and trends for later planting dates noted above making it increasingly more difficult – essentially the window of useful rainfall for maize cropping appears to be shrinking in the north where climate variability may then impact cropping more frequently. An additional rainfall factor that may also be important when planting late, and the cropping season extends into late summer/early autumn, is the mode of the AAO. If it promotes more rainfall during the ripening phase this may have negative consequences for crop development, though the phase of the AAO has been tending towards conditions that reduce rainfall. Whilst this is now beneficial, continued trends for later planting may eventually result in reduced rainfall during the growth phase, which will have a negative impact on crop yield.

Increases in mean dry spell length and reductions in rain day frequency are also demonstrated over Zambia, Malawi and Zimbabwe during the rainfall season (as defined from planting date to rainfall cessation). That this is true particularly when the planting date is taken at its earliest (ignoring false starts) suggests that changes are occurring at the beginning of the season, reinforcing the evidence that the start of consistent rainfall for planting has been getting later over these regions. These observations are also consistent with regional trends (noted within the literature) for later onset, an increase in the frequency of high pressures over the continent and increased length of the dry season. Empirically downscaled climate change scenarios, for six representative stations spanning the region, suggest that increases in late summer rainfall (total and number of rain days) can be expected over widespread areas. However, empirical downscalings for early summer are relatively uncertain, depending on location; some GCMs suggest positive changes in rainfall and others negative changes. There is an indication that further north and east these changes tend to be mostly positive, tending to be more negative further west. The RCM downscalings of a single GCM mostly depend on the choice of RCM and the characteristics of the hydrological cycle each simulates. However, both RCMs tend to agree on increases in late summer rainfall, consistent with the results from the empirical downscaling, though changes during early summer are often less certain.

Allowing for the large uncertainty in changes simulated under CO2 induced anthropogenic climate change, the projected changes are not at odds with those changes noted in the observational record, though statements of attribution are not possible. Indeed rainfall changes under climate change may not be apparent for 70+ years (Christensen et al., 2007). However, observed changes are significant and important, especially to those involved in agriculture. In this regard it should be noted that there have been documented increases in temperature (which are expected to continually increase in the future) over these same regions, which will place additional water-related stresses on agriculture even without any change in rainfall. Changes in factors such as land-use and multidecadal climate factors can all have a bearing on observed trends, and even CO2 induced climate change alone may be nonlinear – reductions in raindays may be counteracted by increases in rainfall intensity and it is the relative strength of competing processes, at different timescales moving into the future, that will determine changes found in the observational record. Understanding which of these changes may occur in tandem with each other, at different locations, is key to the implementation of successful adaptation strategies.

Back to Step 3: Identifying Hazards

Related Pages

Add your project

Exchange your climate change adaptation projects and lessons learned with the global community.