Low wind speed direction extraction using wavelet transform technique

Abstract

Wind in Malaysia is significantly influenced by monsoon season throughout the year and has possessed low wind speed between 3 to 5 m/s. In retrieving wind direction from remotely sensed data, high wind speed (at least above 7 m/s) is needed to produce visible wind streak on the coefficient image. This study aimed to derive low wind speed of wavelet coefficient for wind direction extraction from SAR images. Wavelet coefficient is important to enable detection of wind streak in spatial scale of domain spectrum by recovering wind-wave orientation. To recover the domain spectrum of low wind speed wavelet coefficient, three types of wavelet transform; Fast Fourier Transform (FFT), Short-time Fourier Transform (STFT) and Mexican Hat Wavelet Transform were performed. The best wavelet transform used to derive low wind speed wind direction was determined by using polynomial regression with QuickScat wind direction data. Results show that the r-squared value produced by FFT technique (0.7104), Mexican-hat (0.1427) and STFT (0.0758). Promising result has been showed by FFT in extracting low wind speed direction using enhanced wavelet coefficient derivation method from SAR data.