Akademik Veri Yönetim Sistemi
THEORETICAL AND APPLIED CLIMATOLOGY, cilt.135, ss.157-168, 2019 (SCI-Expanded)
Understanding the causes of rainfall anomalies in the West African Sahel to effectively predict drought events remains a challenge. The physical mechanisms that influence precipitation in this region are complex, uncertain, and imprecise in nature. Fuzzy logic techniques are renowned to be highly efficient in modeling such dynamics. This paper attempts to forecast meteorological drought in Western Niger using fuzzy rule-based modeling techniques. The 3-month scale standardized precipitation index (SPI-3) of four rainfall stations was used as predictand. Monthly data of southern oscillation index (SOI), South Atlantic sea surface temperature (SST), relative humidity (RH), and Atlantic sea level pressure (SLP), sourced from the National Oceanic and Atmosphere Administration (NOAA), were used as predictors. Fuzzy rules and membership functions were generated using fuzzy c-means clustering approach, expert decision, and literature review. For a minimum lead time of 1 month, the model has a coefficient of determination R-2 between 0.80 and 0.88, mean square error (MSE) below 0.17, and Nash-Sutcliffe efficiency (NSE) ranging between 0.79 and 0.87. The empirical frequency distributions of the predicted and the observed drought classes are equal at the 99% of confidence level based on two-sample t test. Results also revealed the discrepancy in the influence of SOI and SLP on drought occurrence at the four stations while the effect of SST and RH are space independent, being both significantly correlated (at alpha < 0.05 level) to the SPI-3. Moreover, the implemented fuzzy model compared to decision tree-based forecast model shows better forecast skills.