In this study, the monthly maximum and minimum surface temperature (Tmax and Tmin) of all India (AI), and 7 temperature homogeneous regions from India are decomposed into several orthogonal components namely Intrinsic Mode Functions (IMFs) using the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) method. Further, the intrinsic modes obtained are transformed analytically using the Normalized Hilbert Transform coupled with Direct Quadrature (NHT-DQ) to understand the time–frequency characteristics of 16 temperature time series pertaining to eight different regions in India. The non-linear and non-stationary nature of all the time series is shown and the dynamic behavior of dominant time scale in different regions of India is highlighted. The spectral analysis of IMFs of each time series depicted the evolution of temperature over the data length along with the modulation of frequencies. Then, the trends of instantaneous amplitudes are estimated to understand the dominant IMFs resulting in temperature changes in India. The results indicated that the higher order IMFs with inter-decadal periodicity displayed a clearly increasing trend in amplitudes since 1970, supporting the signatures of climate change in India. It is also found that a statistically significant change in amplitudes is observed for all oscillatory modes in North East (NE) region for the minimum temperature time series. It is further noticed that instantaneous amplitudes from oscillatory mode 2 (IMF2) of annual periodicity shows a significant trend for both Tmax and Tmin time series for most of the regions. The trend of instantaneous amplitudes for annual scale oscillatory mode and inter-decadal periodicity of 30 years of West Coast (WC) regions is contrasting character when compared with that in other regions, which depict a distinct response of temperature regime of WC region. Moreover, four climate indices and indicators such as Pacific Decadal Oscillation (PDO), Sunspot Number (SN), Total Solar Irradiance (TSI) and CO2 concentration time series data are decomposed using CEEMDAN. The comparison of these components with the modes of extreme (Tmax, Tmin) and mean (Tmean) annual temperature datasets of AI is performed in the time domain. The correlation analysis established the link between these climatic indicators and different temperature time series from India. It is further noticed that such inter-relationships between different indicators and temperature is mainly deciphered in the low frequency modes.

Application of Hilbert Huang Transform Framework for Characterizing the Multiscale Properties of Air Temperature in India over a Century

期刊论文
发表日期：

2025年9月19日

查看全文

S. Adarsh,
Thomas Plocoste,
Vahid Nourani

摘要

In this study, the monthly maximum and minimum surface temperature (T_max and T_min) of all India (AI), and 7 temperature homogeneous regions from India are decomposed into several orthogonal components namely Intrinsic Mode Functions (IMFs) using the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) method. Further, the intrinsic modes obtained are transformed analytically using the Normalized Hilbert Transform coupled with Direct Quadrature (NHT-DQ) to understand the time–frequency characteristics of 16 temperature time series pertaining to eight different regions in India. The non-linear and non-stationary nature of all the time series is shown and the dynamic behavior of dominant time scale in different regions of India is highlighted. The spectral analysis of IMFs of each time series depicted the evolution of temperature over the data length along with the modulation of frequencies. Then, the trends of instantaneous amplitudes are estimated to understand the dominant IMFs resulting in temperature changes in India. The results indicated that the higher order IMFs with inter-decadal periodicity displayed a clearly increasing trend in amplitudes since 1970, supporting the signatures of climate change in India. It is also found that a statistically significant change in amplitudes is observed for all oscillatory modes in North East (NE) region for the minimum temperature time series. It is further noticed that instantaneous amplitudes from oscillatory mode 2 (IMF2) of annual periodicity shows a significant trend for both T_max and T_min time series for most of the regions. The trend of instantaneous amplitudes for annual scale oscillatory mode and inter-decadal periodicity of 30 years of West Coast (WC) regions is contrasting character when compared with that in other regions, which depict a distinct response of temperature regime of WC region. Moreover, four climate indices and indicators such as Pacific Decadal Oscillation (PDO), Sunspot Number (SN), Total Solar Irradiance (TSI) and CO₂ concentration time series data are decomposed using CEEMDAN. The comparison of these components with the modes of extreme (T_max, T_min) and mean (T_mean) annual temperature datasets of AI is performed in the time domain. The correlation analysis established the link between these climatic indicators and different temperature time series from India. It is further noticed that such inter-relationships between different indicators and temperature is mainly deciphered in the low frequency modes.