Erratic local rainfall patterns make reliable predictions very complicated
By HJ Sartorius von Bach.
Namibia’s climate is no stranger to dry periods, and extreme weather patterns such as droughts and floods continue to place the country in a precarious position. In this contribution I will only focus on droughts, which can be defined as the natural disaster of below-average precipitation with significant impact on the natural environment, human society and the economy.
The open question is whether reliable drought predictions can be made.
First, we need to understand the definition of drought. Meteorologically, a drought can be understood as the period during which the actual moisture supply is consistently below the climatic supply. This definition is different from an agricultural drought. The latter is based on less water supply than needed, i.e. drought begins only when vegetation can no longer obtain water from the soil at a sufficient rate to replenish that lost to the air by transpiration.
Therefore, a drought does not necessarily begin on the day that rain ceases, but rather when the soil moisture is drained. Thus the severity of a drought depends on the condition of the soil at the beginning of a dry period.
This brings us to deficiency in total water supply as compared to total water demand, which is hydrologically referred to as the inadequacy of water systems. It implies, that if a large reservoir is full of water, a silo is full of grain or a pasture land retains some reserve, one dry period will not result in a drought. Drought is only the consequence of the accumulated effect of several years. Drought occurs because of many reasons and mainly in areas with lower rainfall patterns. Dry conditions can be triggered by high levels of reflected sunlight or high pressure systems, restricting the development of rainfall over a certain area.
With fast-developing technologies, scientists have begun to understand drought patterns. They follow the El Niño phenomenon to detect the beginning of a cycle and its severity. For example, already in 2012, meteorologists forecast that southern Africa was entering a severe drought period consisting of 10 years. This development together with global warming triggered activities with substantial direct impact on agriculture and indirectly on the regional economy at large.
Unfortunately, this early warning was not incorporated into proper planning. The purpose of this contribution therefore calls for including advanced planning for drought to be incorporated into the execution of national tasks.
Drought analysis depends on many factors. Most records of any significance for both meteorological factors and streamflow go back at best 75 to 100 years and in most cases only between 30 to 50 years. Predictions are therefore limited and dependent on what has happened in the more recent past.
Internationally, considerable work has been going on in deriving information on climatic changes from tree rings, ice cores and lake sediments. Various cycles have been identified. Medium term ice core series for example showed that some 300 years ago a samll ice age period persisted for several decades and that for the past 100 years, a relatively warm climate pattern was observed.
Longer time span series showed that about 5000 to 6000 years ago, the world experienced its warmest period recorded. The cycle showed that it took about 40,000 to 70,000 years to reach a glacial maximum; the last one was 18,000 years ago. Data showed that it only took between 2000 to 5000 years to completely destroy the formed continental-sized ice sheet. Therefore, analysis shows that it is likely that we are still in a warm trough within a long-term cycle, i.e. the current climatological variability might continue for at least the next few thousand years.
Focusing on droughts within the historic data, the earliest recorded drought dates back to the year 1540, when probably the worst drought took place in Central Europe without any precipitation for 11 months leading to temperature increases of 5 to 7 degrees above the normal. Recorded severe droughts occurred in India in 1900, in Russia in 1921, China 1929, the southern USA in the 1930’s and more recently in Kenya, Brazil, Australia, and other countries. The severity of these droughts varied for each event. Nevertheless, the overall impact of these severe droughts would have been less critical once known to allow protective steps to be put in place.
In Namibia, droughts are regular and over the past years they caused economic decline, food shortages, environmental degradation and hardship among the poor. Only recently, monitoring of droughts is executed by the bi-annual Food and Nutrition Security Monitoring Assessment, as an early warning mechanism to caution the region on economic impact and in particular on affected sectors.
Unfortunately, data is not sufficient for proper analysis. Nevertheless, seasonal precipitation totals could provide useful information to determine the occurrence of drought or non-drought situations.
Available annual precipitation data from central Namibia was used to illustrate historical rainfall patterns. Windhoek data, although not complete, is available since 1891, data from the Waterberg area was made available since 1892 and from the Kamanjab area since 1942.
The following graph shows the variability of the annual rainfall. Moving averages were used to obtain general observations. The information shows that Namibia had drought periods around 1902, 1933, 1961, 1982, 1998 and presently probably the worst of all. It is interesting to observe that the different rainfall stations presented similar patterns, although the western location entered the dry period earlier. It can be observed, that the trough of dry years existed for at least some years although with some deviations, it appeared that the period between 1900 and 2019 included 4 cycles of approximately 28 years each. If that is correct, we almost have reached the end of the trough of the current cycle, which was earlier predicted for 2012.
The next step was to determine whether there is truth to these cycles. Mathematical estimations were applied to find equations which could present rainfall patterns for these stations. It is clear that the limited data and the variability of annual data produced low statistical significance. However, visually the existence of a cyclical pattern with varying aptitude can be observed. It is interesting to note that the first and the last cycle consisted of almost 28 years and that the period between 1934 and 1989 was aggregated as one larger cycle.
To test the appropriateness of the cycles, each sample with stochastic frequency components was normalized to have a zero mean and unit variance. It presented a similar but an inverse cyclical pattern where the error term reached its highest aptitude during dry times. As cyclical patterns remained, it may be suggested that the isolation of cycles were not perfected and that additional influence, such as changing weather patterns, contributed to the significant increase of the last cyclical error term.
The above contribution shows that more research is required for an early warning of droughts. Trends need to be known and causes sufficiently understood to develop reliable outlooks about the likelihood of drought occurrences, their duration and severity. These outlooks would be of significant importance for the agricultural sector and national economic planning. Results could be used to safeguard against future famines, possible food shortages and to counter rising food prices.