Enhancing climate model simulations: A deep learning approach to bias correction of relative humidity in NEX-GDDP
Enhancing climate model simulations: A deep learning approach to bias correction of relative humidity in NEX-GDDP
Sheida Mafakheri Bashmagh1 Saman Maroufpoor2
1) MSc graduated
2) Researcher
IOI Code
:
ACZE-ACAAZD
Publication :
10th.International Congress of Developing Agriculture, Natural Resources, Environment and Tourism of Iran(10icsda.ir)
Abstract :
Access to accurate climate data is one of the main factors for reliable climate and water resources research. Recently, the National Aeronautics and Space Administration (NASA) Earth Exchange Global Daily Downscaled Projections (NEX-GDDP) database has been recognized as one of the most reliable data sources for climate research. This research aimed to improve the accuracy of daily NEX-GDDP relative humidity data using a Long Short-Term Memory (LSTM) model. Four cities in Khuzestan province, Iran, were chosen as case studies for the period 2000-2014. The results demonstrate that integrating observational maximum and minimum temperature data with NEX-GDDP significantly improved the accuracy of relative humidity predictions. The proposed approach led to a 15.82% increase in the Wilmot index (WI) and a 58.88% reduction in the root mean square error (RMSE).
Keywords : bias correction deep learning nex gddp relative humidity