A Relative Density Ratio-Based Framework for Detection of Land Cover Changes in MODIS NDVI Time Series

Abstract

To improve statistical approaches for near real-time land cover change detection in nonGaussian time-series data, we propose a supervised land cover change detection framework in which a MODIS NDVI time series is modeled as a triply modulated cosine function using the extended Kalman filter and the trend parameter of the triply modulated cosine function is used to derive repeated sequential probability ratio test (RSPRT) statistics. The statistics are based on relative density ratios estimated directly from the training set by a relative unconstrained least squares importance Fitting (RULSIF) algorithm, unlike traditional likelihood ratio-based test statistics. We test the framework on simulated, synthetic, and real-world beetle infestation datasets, and show that using estimated relative density ratios, instead of assuming the individual density functions to be Gaussian or approximating them with Gaussian Kernels, in the RSPRT statistics achieves better performance in terms of accuracy and detection delay. We verify the efficiency of the proposed approach by comparing its performance with three existing methods on all the three datasets under consideration in this study. We also propose a simple heuristic technique that tunes the threshold efficiently in difficult cases of near real-time change detection, when we need to take three performance indices, namely, false positives, false negatives, and mean detection delay, into account simultaneously. Index Terms-Change detection, extended Kalman filter (EKF), model fitting, MODIS, relative density ratio, time series. I. INTRODUCTIONL AND COVER change detection research has seen significant recent contributions [1]- [19]. However, every proposed framework has its limitations on global scale due to particularity of the task at hand and the circumstances under which it is developed. Hence, no single framework is optimal in a wide range of scenarios simultaneously. Therefore, an efficient change detection framework is always in demand for a particular task and circumstances under consideration. In this Manuscript

Quantifying Short-Term Urban Land Cover Change with Time Series Landsat Data: A Comparison of Four Different Cities

“…Previously, many studies have been carried out with multitemporal satellite images for the monitoring of various land use or land cover types, such as forest [7], water resources [8], wetlands [9,10,11], paddy rice, or other croplands [12,13]. Numerous methods or tools have been proposed for the detection and analysis of these land use or land cover changes, including cellular automata models [14], relative density ratio-based approach [15], Bradley-Terry model [16], unsupervised change detection method [17], principal component analysis [18], as well as the recent cloud computing platform Google Earth Engine [9,12,13].…”

A Probability-Based Spectral Unmixing Analysis for Mapping Percentage Vegetation Cover of Arid and Semi-Arid Areas

“…Based on standard martingale central limit theorem and Gaussian distribution, any non-stationarity in the time series could be detected, indicating beetle infestations. Anees et al [61] further improved this method so that it could be applied to non-Gaussian time series data to detect near real-time land cover changes using a MODIS NDVI time series. The previous studies imply that the integration of the two proposed methods-especially PBSUA-and the detection framework from Anees and Aryal [60] would make it possible to develop a near real-time monitoring approach for PVC dynamics for large arid and semi-arid areas.…”

Automated Detection of Buildings from Heterogeneous VHR Satellite Images for Rapid Response to Natural Disasters

“…In the future, we would extend the proposed method to discover the damage triggered by an earthquake [33], or to detect changes by fusing multiple temporal images [34,35].…”