Optimal selection of wavelet transform parameters for spatio-temporal analysis based on non-stationary NDVI MODIS time series in Mediterranean region

Rhif, Manel; Abbes, Ali Ben; Martínez, Beatriz; Farah, Imed Riadh; Gilabert, María Amparo

doi:10.1016/j.isprsjprs.2022.09.007

Cited by 5 publications

(1 citation statement)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…According to the domestic and foreign research on vegetation change, the role of wavelet transform's multi-scale analysis theory in monitoring the dynamic changes of vegetation is demonstrated. It can be found that the wavelet analysis of a set of vegetation index time series can mine the dynamic change characteristics of vegetation within a year, as well as the dynamic change trend of vegetation between years [3,[13][14][15]. With the decomposition and reconstruction ability of vegetation time series data, discrete wavelet with great potential in vegetation prediction can provide a basis for multiscale predicting of vegetation changes and analyzing vegetation change prediction capabilities of different scales of vegetation change features.…”

Section: Introductionmentioning

confidence: 99%

Prediction of Vegetation Change by Discrete Wavelet Decomposition Based on Remote Sensing Time Series Images

Long¹,

Qin²,

Wang³

et al. 2023

View full text Add to dashboard Cite

The development of remote sensing technology has accumulated a large number of remote sensing image time series data for human monitoring of surface vegetation change, which provides a basis for vegetation change prediction. In order to improve the prediction accuracy of vegetation change, this paper uses discrete wavelet to decompose remote sensing image sequences at multiple scales, to explore the difference of influence of different temporal scale change characteristics on vegetation spatio-temporal change prediction, and find the best decomposition scale for vegetation change prediction. In this paper, the research object is the MODIS 13Q1 EVI image data of Hunan Province from 2001 to 2021. The discrete wavelet is adopted to obtain multi-scale vegetation trend components and detailed component sequences, and then complete the LSTM modeling prediction and comparison. The following are the experimental findings: the predictive ability of the discrete wavelet decomposition sequence group is better than that of the original EVI time series to varying degrees. The order of prediction accuracy is: monthly scale > seasonal scale > annual scale > original EVI time series. Thus, it is of reference significance to the research of application scenarios of change prediction of other regionalized variables with multi-scale characteristics.

show abstract

Section: Introductionmentioning

confidence: 99%