Correction of Temperature Forecast Using Spectral Analysis Method on Mountain Area

Since 2014, KMA (Korea Meteorological Administration) has been operating the mountain weather forecast guidance based on the NWP model about 120 sites to give the sufficient information to people and protect human lives by uncertain disaster on mountain. It supports 11 forecast elements such as temperature, wind, humidity, probability of rainfall and so on. Because there are few observational data and unknown atmospheric phenomenon in that area, it is difficult to have a good performance to predict. So we have been tried to develop a better post-processing method. The suggested way in this study is to use the characteristics that the atmosphere has the periodicity. It is expected to know the information for the sites without the observation data using the analysis of the sites with it.

Hans Volland (1995) showed that atmospheric tides are generated mainly thermally by the regular solar heat and have period and A. Wagner (1931) revealed yearly oscillation of temperature in the middle Europe. As a analysis tool for periodicity, Mohit et. al. (2013) suggested the weather temperature computation method with DFT (Discrete Fourier Transform) tool and represented a good agreement for daily temperature in forecast.

In this study, it is carried out for temperature correction on mountain area using FFT (Fast Fourier transform), known as one of spectral analysis method. First, it is investigated for the periodic characteristics of temperature over 3 years and estimated for Power Spectrum Density (PSD) to obtain main components in common for observation. In the same way, it is analyzed for mountain forecast. Finally, on the result of periodic characteristics, the effect of periodic correction for the mountain forecast with observation is evaluated

Key words: Mountain forecast, Spectral Analysis, FFT (Fast Fourier transform), Temperature correction