Abstract:
The area encompassed by mountainous terrain, topography is the principal controller of the
spatial pattern of rainfall through orographic effects; moreover, mountainous atmospheric
dynamics have a significant influence on rainfall prediction. In this case, this study aimed to
examine how such circumstances influence rainfall prediction in eastern and southeastern
Ethiopia (ESE). Ethiopia Meteorology Institute’s (EMI’s) rainfall gridded daily for the years
1991–2020; European Center for Medium-range Weather Forecast (ECMWF) and Global
Forecast System (GFS) hourly rainfall for the years 2011–2020; Fifth Generation ECMWF
Atmospheric Reanalysis (ERA5) atmospheric dynamics hourly for 2011–2020; and Global 30
Arc-Second Elevation (GTOPO30) datasets were utilized in this study. Principal component
analysis conducted to evaluate orographic factors with the rainfall, indicated that comparatively
similar mean annual rainfall distribution patterns were fitted with the topographic pattern. The
annual rainfall gradient with a strong relationship of 84% has shown an increasing rate with
elevation below 3,500 m and a decreasing rate above 3,500 m with a weak association (17%).
The ECMWF model outperformed the GFS model in predicting rainfall over midland and
highland areas, while the GFS model indicated better performance in lowland regions of ESE, as
demonstrated by the probabilistic metrics. In the same manner, the deterministic metrics
revealed that both models performed under estimation over highland, above estimation over
western midland, and relatively better over lowland of ESE. Inaccuracies in rainfall prediction
by the models were highlighted by significant influences of mountainous atmospheric dynamics
on orographic rainfall prediction variation through ESE, as emphasized by the ERA5. The
Froude number (Fr), computed at a pressure height range and by reducing the original elevation
of the ESE to 75%, 50%, and 25%, illustrated that the Peak Mountains of the ESE acted as
blocking barriers and altered atmospheric dynamics. Furthermore, multiple stepwise regression
analysis identified rainfall determination values between 27.6% and 86.6% for ECMWF and
31.7% and 79.9% for GFS, and the best top-selected predictors were boundary layer height,
vertically integrated moisture flux, and convective available potential energy. Nevertheless,
these mountainous atmospheric dynamics have shown a significant ability to determine rainfall
prediction; in-depth studies are compulsory at high spatio-temporal resolution
