Introducing the Time Series Change Visualization and Interpretation (TSCVI) method for the interpretation of global NDVI changes

Abstract

This paper presents a novel method for the visualization of changes in vegetation related variables. This method, termed Time Series Change Visualization and Interpretation (TSCVI), allows to summarize changes associated to both vegetation productivity and phenology in a single map. To that end, three metrics are retrieved on an annual basis from plotting NDVI (Normalized Difference Vegetation Index) values on a polar plot. Changes in these metrics are then analyzed and mapped in an IHS (Intensity Hue Saturation) image, where colors indicate changes regarding the growing-season (earlier or later occurrence, stronger or weaker seasonality), while changes associated to productivity are evidenced by the image intensity. TSCVI metrics are presented for 445 sites corresponding to different land covers, and their sensitivity to dataset characteristics (temporal compositing, presence of clouds) is analyzed. We also applied the TSCVI method to GIMMS NDVI3g data, by analyzing the changes in the TSCVI annual metrics for the periods 1982–1986 and 2009–2013: the resulting map captures well-known changes evidenced in previous studies. Finally, we applied the proposed approach to Sentinel 2 data to visualize charges in the province of Valencia (Spain) between 2018 and 2019. The TSCVI approach can be used for a wide range of remotely-sensed variables in addition to NDVI. All the TSCVI routines used in this work are available freely for download at https://www.uv.es/juy/resources.htm.

Keywords

• Change
• Vegetation
• NDVI
• Time series
• Multitemporal
• Visualization