Akademik Veri Yönetim Sistemi
11th INTERNATIONAL CUKUROVA AGRICULTURE AND VETERINARY CONGRESS, Adana, Türkiye, 27 - 28 Aralık 2025, ss.863-877, (Tam Metin Bildiri)
Precipitation data is critical for water resource planning and management, especially under the
current impacts of climate change, and will become even more important in the future. One of
the most preferred sources of information for estimating future precipitation amounts under the
influence of climate change is the CMIP6 Global Circulation Models (GCMs). However, due
to their relatively low spatial resolution, systematic errors in precipitation forecasting, and the
need for complex downscaling procedures to ensure reliable use in regional studies, these model
outputs have not been widely adopted yet. In this study, using three different global circulation
model datasets, monthly precipitation amount estimations for the period 1979–2014 were
obtained for the Amasra, Bartın, and Kozcağız stations, where measured observation data are
available. The estimation of precipitation amounts at the station points was carried out using
the Multivariate Adaptive Regression Splines (MARS) algorithm, which is widely applied in
the literature. For the calculation and validation of model parameters, the dataset was divided
into training and testing subsets with a ratio of 80%–20%. Although the MARS model showed
a relatively positive improvement in precipitation estimation compared to the raw global
circulation model precipitation data, it was observed that the precipitation estimates for the
selected stations should not be used without bias correction. Consequently, the downscaled
MARS precipitation model results represented the observed precipitation better than the lowresolution
raw GCM model outputs according to the Kling-Gupta Efficiency (KGE) and Nash-
Sutcliffe Efficiency (NSE) indicators. Accurate precipitation estimations are critically
important for the sustainability of water resources and the reliability of environmental and
agricultural practices. This research discusses the importance of bias correction and evaluates
the advantages and disadvantages of a widely used methodological approach in future climate
projection studies related to precipitation.