Last update: August 16, 2011 11:06:05 AM E-mail Print


Agroclimatological characterisation of the Karoo: a preliminary investigation


J.C. Venter

AgroMet (ISCW), Grootfontein College of Agriculture, Middelburg CP, 5900



A method for agroclimatological characterisation of rangeland is discussed with the dwarf-shrub grassland of the Karoo as example. This characterisation is equivalent to the determination of probability or risk of drought and can be used as a basis by insurance schemes against drought. It may also point the way to a method of identification of high-risk areas for desertification.



The risk of desertification is greater in certain agroclimatic regions, particularly in arid and semi-arid ecosystems inherently vulnerable to drought stress. "Only by monitoring both short-and long-term trends in environmental change will society be able to better understand the interrelationships that exist" (Anon., 1987). This necessitates either the systematic monitoring of areas prone to desertification in terms of biomass productivity and soil water (Bowonder, 1987), or keeping track of changes by using models for this purpose.

The ZA 1 -model for rangeland drought indexing was discussed elsewhere (Venter, 1992). As far as its proneness to drought is concerned, an area is characterised by the frequency of occurrence of drought indices ranging from very low to normal. The frequency distribution may differ from one weather station to the next within the same area. These frequencies can be used to determine the probabilities of droughts of varying intensity (Wilhite & Glantz, 1987), and therefore to describe the aridity of a region (Troyo-Dieguez, de Lachica-Bonilla & Fernandez-Zayas, 1990).

The ZA-index comprises the plant-available water expressed as a percentage of the mean available water for a particular month (Venter, 1992). These monthly indices are calculated and mapped as a matter of routine for an extensive network of rainfall stations. However, a complete picture of a drought situation in a region can be obtained only if the probability that a given index may occur at a given place has been established. Such probabilities can be used to define the risks associated with drought situations with different intensities.



Long-term records of monthly rainfall were required in order to determine probabilities. An existing rainfall data base (Adamson, 1987) was used. Maximum air temperature values, being required by the model (Venter, 1992), were obtained from a South African Weather Bureau report (S.A. Weather Bureau, 1986). These temperatures were interpolated to obtain estimations where necessary.

Rainfall records from 50 stations throughout the Karoo Region of the Department of Agricultural Development and neighbouring areas were analysed using the drought model to obtain monthly drought indices for these stations. The cumulative frequency distribution of drought indices was then calculated for each station, the frequency being a percentage of the record length in months. Finally these frequencies (probabilities) were mapped, producing separate maps for the probability of occurrence of indices in the categories < 10, < 20 and up to indices < 110. The mapping was accomplished using multiquadric analysis (Adamson, 1987) of the probabilities associated with the 50 stations.



Only two of the maps (Fig. 1 and 2) are discussed. Different increments were used in each case for adequate spacing of the isolines. The frequencies mapped are regarded as the probabilities that an index will be lower than a particular value at all points in the region.



The line marked 0,2 in Fig. 1 connects all locations where a ZA-index < 10 occurs 0,2 % of the time in months, i.e. one month out of 500 or once in about 42 years. Unless adequate long-term rainfall records are available, no definitive statements can be made about the probability that such a rare event will occur or about the position of the 0,2 line. However, the exact positions of lines are less important than the fact that the map indicates that the Karoo can be divided into two distinct areas, namely the very drought-prone north-west and I the less drought-prone south-eastern and eastern part where the index is > 10.

The area where a drought index lower than 10 occurs most often is located north of Carnarvon (1,4 isoline in Fig. 1) and such a catastrophic drought can be expected more frequently than 7 months out of 500, i.e. about once every 6 years.

Although a large part of the Karoo never experiences droughts characterised by indices < 10 (Fig. 1), it must be borne in mind that the results are more appropriate when dealing with shrubland drought situations (Venter, 1992). A similar investigation of droughts in Bushman grass veld, for instance, may reveal a different picture, although the north-western part should still prove the most drought prone.

Fig. 2 is of particular importance for "normal" situations. If a 50 % frequency of occurrence defines a normal situation, it is clear that a drought index in the region of 80 or even slightly less should be regarded as normal in the far north-western and northern part. In the eastern parts of the Karoo however, small pockets around Middelburg and towards the south-west of Aliwal-North were identified (not shown) where an index of 100 may be considered as the upper limit of a normal situation. As a general rule for the Karoo as a whole, a normal situation can therefore be defined as one with indices ranging between 80 and 100.

In order to decide on the situation at a specific location in a specific month as far as the risk attached to it is concerned, one merely has to compare its index for that month with the relevant map (e.g. Fig. 1 or 2).



Indices generated on a monthly basis can be used to calculate remuneration of policy holders participating in a drought insurance scheme (Parry & Carter, 1987). Premiums could be based on probabilities (risks) stored in a data base created for that purpose (Changnon & Changnon, 1990). The feasibility of such schemes for different kinds of drought (e.g. shrubland, savanna and others) is subject to further research.

The results must be regarded as preliminary due to the limited number of stations (50) that could be used, as well as the fact that mean rainfall totals rather than medians were used. Both aspects will be re-addressed once enough historical rainfall records of sufficient length become available.

1. Zucchini-Adams



Thanks are due to staff members of AgroMet for preparation of the maps.



ADAMSON. P.T. 1987. South African Rainfall Data Base. Department of Water Affairs. Technical Report TR133. Pretoria.

ANON.. 1987. Drought research (priorities. In Planning for Drought, edited by D.A. Wilhite & W.E. Easterling. West view Press. Boulder & London.

BOWONDER. B.. 1987. Environmental problems in developing countries. Prog. phys. Geogr. 11,246-259.

CHANGNON. S.A. & CHANGNON.J.M..1990. Use of Climatological Data in Weather Insurance. J. Climat. 3. 568-576.

PARRY. M.L. & CARTER. T.R.. 1987. Climate impact assessment: a Review of some approaches. In Planning for Drought. edited by D.A. Wilhite & W.E. Easterling. Westview Press. Boulder & London.

TROYO-DIEGUEZ. E.. DE LA CHICA-BONILLA. F. & FERNANDEZ-ZAYAS,J.L.. 1990 . A simple aridity equation for agricultural purposes in marginal zones. J. arid Environ. 19. 353-362.

VENTER. J.C.. 1992. Drought characterisation based on Karoo shrubland productivity. S. Afr. J. Sci. 88.154-157.

S A WEATHER BUREAU. 1986. Climate of South Africa - climate statistics up to 1984. Department of Environment Affairs. Report WB40. Pretoria.

WILHlTE. D.A. & GLANTZ, M.H.. 1987. Understanding the drought phenomenon: the role of definitions. In; Planning for Drought. Edited by D.A. Wilhite & W.E. Easterling. Westview Press. Boulder & London.




Karoo Agric, Vol. 4, No 4, 1992 (1-3)