@phdthesis{Shikhani2024, author = {Shikhani, Muhammed}, title = {Forecasting water temperature of lakes and reservoirs over different time-scales and associated uncertainty and skill}, doi = {10.26127/BTUOpen-6905}, url = {http://nbn-resolving.de/urn:nbn:de:kobv:co1-opus4-69052}, school = {BTU Cottbus - Senftenberg}, year = {2024}, abstract = {Freshwater lakes are facing increasing stress due to climate change and the more frequent and intense extreme events. Therefore, we urgently need to develop workflows to forecast the long-term impacts of climate change under various emission scenarios and near-term forecasts to predict the response of the lake variables' extreme events and inform decision-makers about mid and shortterm variabilities, enabling proactive and informed management. However, developing a forecasting workflow requires a proper understanding of the theoretical concepts and definitions of forecasts and projections, which are often not recognized correctly in the forecasting literature. Forecasts should not only make predictions but also quantify uncertainties or skills to provide reliable information. This thesis aims at developing a general workflow focusing on lake temperature, which can be applied at different forecasting horizons. The workflow employs ensemble techniques, skill scoring, and uncertainty analysis, using state-of-the-art open-source climate data and lake models to provide robust forecasts. The workflows were applied at two sites: Lake Sevan, a large natural alpine lake, and the Wupper Reservoir, a highly flushed temperate reservoir. The long-term end-of-the-century projections were conducted for Lake Sevan by employing a large multi-dimensional ensemble generated by forcing an ensemble of five hydrodynamic lake models with a multi-domain ensemble of CORDEX GCMs (General Circulation Models) and RCMs (Regional Climate Models) projections. The approach also included quantification of the contributions from the sources of uncertainty. The mid-term seasonal forecasts were conducted on the Wupper Reservoir by utilizing an integrated workflow of coupled Catchment-Lake and auxiliary inflow-temperature and outflow models, using the ECMWF SEAS-5 seasonal forecasts as forcing data. This project marked the first application of seasonal forecasts for lakes within a study that included multiple case studies and provided the corresponding forecast skills to address the probabilistic nature of the seasonal forecasts. The short-term forecasts employed a simplified version of the workflow used for seasonal forecasts but used the DWD ICON numerical weather predictions as first application of ICON for lakes. The probabilistic nature of the climate projections poses challenges to the application, as predictability in some regions can be limited. However, the findings of this thesis confirm that the general concept is valid to address the probabilistic nature of climate and weather predictions, deliver forecasts for the targeted horizons, and include skills and uncertainty quantification to ensure reliability. This thesis identify when, where, and how these predictions can be relied on.}, subject = {Forecasting; Lakes temperature; Forecasting skills; Lake modelling; Forecasts uncertainty; Vorhersagen; Seentemperatur; Seenmodellierung; Vorhersageng{\"u}te; Vorhersagen Unsicherheiten; Sewansee; Wupper-Talsperre; Hydrologische Vorhersage; Klima{\"a}nderung; See; Wassertemperatur}, language = {en} }