Tropical rainfall variability — The agroclimatic impact

In tropical lowlands, agriculture is largely controlled by rainfall conditions. In these areas, most elements of climate, and particularly temperatures, are very uniform both over place and time, but rainfall is much more variable. Lack of sufficient rainfall is the main limiting factor in agriculture in most tropical areas. Long-term predictions, expected from agroclimatologist, should be mainly concerned with rainfall, but the irregularity of this element makes this a difficult task.

Therefore an obvious need exists for methods to quantify rainfall variability in relation to agriculture. These methods should be simple, so that they can be used wherever necessary without the use of a computer. Peninsular Malaysia offers favourable conditions for the study and application of these methods.

Annual rainfall totals present a picture of endless variations, with no clear trend or cycles. The seasonal rainfall distribution is much more regular, as large-scale weather patterns are frequently controlled by the monsoonal wind systems. Agricultural conditions determined by seasonal rainfall can be described most conviniently by probability figures. When these are related to crop water needs, as is done in the Agricultural Rainfall Index, the risk of droughts can be indicated accurately enough for most agroclimatic purposes. The proper selection of crops and correct timing of planting can also be decided by this method.

The irregularity of rainfall from day to day affects the growth and yields of many crops, particularly those with shallow root systems. Even during seasons which are generally considered favourable for crop production, temporary shortages or surpluses of rainfall occur often. To obtain estimates of these effects a daily water-balance model is introduced. Risks are indicated by the number of days with severe moisture stress and by the amount of surplus rain water. A comparison with a more sophisticated computer programme shows the validity and accuracy of the model.