Spectral Analysis of Temperature and Rainfall Trends using Hybrid Nonparametric-Wavelet Transform Method
Spectral Analysis of Temperature and Rainfall Trends using Hybrid Nonparametric-Wavelet Transform Method
Khurram Sheraz* and Taj Ali Khan
ABSTRACT
The analysis of the meteorological indices provides valuable information about the evolution of climate during past and may predict its reflection in the future. Analysis of trends in meteorological variables is challenging as they are of non-stationary nature and may include both stochastic and noise components. The aim of this research is to integrate the signal processing technique Discrete Wavelet Transform (DWT) with different versions of Mann-Kendall (MK) trend tests for investigating the dominant climate systems characterizing the long-term time-series of the two most important meteorological parameters: mean air temperature (oC) and total rainfall (mm) for the different stations lying in the upper Indus River basin of Pakistan. The observed time series were categorized into monthly, seasonal, seasonally-based, and annual time series for investigating trends. The analysis showed that winters have experienced significant warming over the study area except for stations Nowshera and Parachinar while summers are experiencing decreasing temperature trends. Rainfall trends were dominated by positive trends. The results from the DWT and MK tests (at 5% significance level) on the different data types revealed that for higher-resolution data (monthly); the high-frequency intra-annual periodicities affecting air-temperature trends were dominated by 8-monthly fluctuations, and for the lower-resolution seasonal and annual data the interannual periodicities ranged from 2–4 years. For total rainfall data: 2-8 monthly; 2-4 yearly; and 2-8 yearly periodic modes were found dominant and characterizing the monthly, seasonally-based, and annual trends, respectively. It was found that DWT effectively extracted the two-dimensional time-amplitude-frequency information from the time series that was hidden in the observed raw data i.e. the time-frequency information manifested in the shape of time periodicities of intra-annual to inter-annual events.
To share on other social networks, click on any share button. What are these?
