Abstract:
The pattern of occurrence of rainfall is equally important as the amount of rainfall received in a
country like Sri Lanka which has agriculture based economy. Once the pattern of occurrence of
rainfall is adequately modelled, it can be used to provide prior knowledge of properties of varying
rainfall system which is very much essential for agricultural planning. Therefore, this study
investigates the pattern of occurrence of rainfall in Anuradhapura district for both Yala (March to
September) and Maha (October to February) Seasons. The daily rainfall data of 10 years (January,
2008 to September, 2018) were used for the study. According to the amount of rainfall received,
each day was classified as dry or wet. By considering two states as wet and dry, daily rainfall was
modelled as Two-State Markov Chain and then the stationary probability of each state and the
expected length of wet and dry spells for both Yala and Maha Seasons were calculated. It is
revealed that the one-step transition probabilities of a dry day followed by a dry day were high in
both seasons compared to other three types of state transitions. Also, the highest probabilities of
day being dry and wet were detected in Yala and Maha Seasons respectively. During the period
from March to September, the expected length of dry spell was larger than the rest of the year. The
details provided in here enable proper planning of agricultural activities to mitigate the negative
effect due to varying patterns of rainfall.
