RT Dissertation/Thesis
T1 Climate dynamics : the performance of seasonal ensemble forecast for improving food security in Ethiopia
A1 Ware,Markos Budusa
WP 2023/10/11
AB Part one of this thesis aims to define homogenous climatic regions using objective clustering methods and characterize seasonal cycles, trends, and anomalies in precipitation and temperature. Climate-based on amplifies inherent spatiotemporal climate variability in the Horn of Africa due to global, regional, coastal, and local processes. The homogeneous climatic regions and synoptic circulation types were defined using Principal Component Analysis (PCA) PCA–K-means and PCA–Ward’s. Using the decision criteria of respective algorithms, four homogenous climatic regions were determined for Ethiopia. These climatic regions were distinctive in their seasonal cycles, trends, and anomalies in annual and seasonal precipitation and temperature. These results highlight that the trends in precipitation and temperature vary not only between climatic regions but also by rainy seasons. The short rains (received between November and December) increased by 50 mm/decade in the southwestern region where the evergreen forest meets with the long rainy season. The mean annual and seasonal temperature increased between 0.3 and 0.6 °C/decade virtually in all climatic regions. Regionalization methods were sensitive to spatial domain size but not to the length of the time series. 
Climatology of sea-level air pressure showed decreasing northward trend over the study domain, as did the temperature, wind velocity, and relative humidity at 500 hPa. However, geopotential height at 500 hPa and temperature at 850 hPa decreased toward the south over the domain. Circulation types were defined by applying PCA on a composite matrix of the six variables. From the first five Principal Components (PCs), ten circulation types (CTs) were defined over East Africa and then associated with environmental events. CTs clearly distinguished rainy seasons comprising different atmospheric states responsible for varying weathers. The summer season was described by a combination of strong positive anomalies in temperature at 850 hPa, northeasterly winds, and Somali jet at 500 hPa, and weak negative anomalies in temperature at 500 hPa. Trends in the number of days categorized in different CTs showed a significant variation among the groups. The drought events, defined using the consecutive dry days (CDD), correspond with positive anomalies in temperature at 850 hPa, northwesterly and Somali Jet, and negative anomalies in relative humidity at 500 hPa. Flooding, defined using a proxy of 80 mm/day per grid cell, was associated with strong westerly winds at 500 hPa, strong positive anomalies in temperature at the lower troposphere, strong easterlies and southwesterly, and positive anomalies in relative humidity at 500 hPa.
Part two of the thesis aims to assess the performance of the seasonal ensemble forecast over the Horn of Africa for improving food security. A seasonal forecast with a horizon of up to seven months offers a great opportunity for agricultural optimization, which results in an improved economy and food security. For this purpose, the Weather Research and Forecasting (WRF) model was applied for dynamical downscaling of the latest seasonal forecasting system version 5 (SEAS5) for summer 2018 with different microphysics parameterizations, and initial and boundary conditions. Downscaling was performed by a horizontal resolution of 3 km over the topographically complex domain of East Africa. The seasonal ensemble forecast was evaluated using probabilistic metrics like the Brier skill score, probability ranking score, continuous probability ranking score, discrimination score, and ignorance score. The results of the WRF showed that the model has a strong warm bias in the 2m temperature and a wet bias in precipitation. The relative operating characteristics (ROC) curve showed a higher predicting probability of 2m temperature in below-normal and above-normal terciles over northern Ethiopia and the Indian Ocean, where the model performed better, highlighting the advantage of high-resolution simulations compared to ERA5. The median and distribution of WRF, SEAS5, and ERA5 showed remarkable variation between the homogenous climatic regions. Especially the summer of 2018 was wetter relative to climatology, and WRF overestimated this condition in the region.
K1 Klimazone
K1 Somalihalbinsel
K1 Hauptkomponentenanalyse
K1 Horn von Afrika
K1 PCA-K-Mittelwert
K1 PCA-Ward
PP Hohenheim
PB Kommunikations-, Informations- und Medienzentrum der Universität Hohenheim
UL http://opus.uni-hohenheim.de/volltexte/2023/2213