Abstract:
Observational evidence indicates that global climate changes have significantly affected a diverse set of natural
and human systems and activities in many countries and consequently the global community is facing the
impact of such natural disasters. Longer dry spells is one of the recurrent feature of the natural disaster in the
dry zone of Sri Lanka. The unpredictable pattern of dry spells have already caused significant damages to the
livelihood of people and the economy of the country. A review on statistical anlysis on dry spells by
Mathugama and Peiris (2011) showed that no studies were reported to predict the starting date or length of dry
spells. This research was therefore initiated to explore the possibility of forecasting starting period and the
length of the four longest dry spells within a year ('critical dry spells - CDS') in the selected five locations in the
DL| agro-ecological region in Sri Lanka. There is a significant correlations (p<0.05) among starting dates of
successive critical dry spells, but such association was not found for the length of the CDS. Log regression
models and weighted regression models were developed to forcast the starting dates of second, third and fourth
critical dry spells separately for all locations All the models and all parameters were significant (p<0.005) and
the models were tested for an independant set of data. However, a model was not able develop for the starting
date of the first CDS. Critical dry spell length series is very complicated due to structural and behavioral
changes influenced by climate and also not equally spaced. Two new types of non linear models were developed
using existing bilinear models. First one is based on normal non linear model with component aXP with
additive error and then add bilinear terms to the model. The second new approach was to add an exogeneous
input variable to the bilinear model. The results obtained in this study will helpful to minimize unexpected
damage due to droughts and will help effective and efficient planning in disasters management.