GECM: geospatial environmental change monitoring, a software package to detect trends and abrupt change in earth observation time series.

Keywords: Time Series, Raster, Environmental Change, Remote Sensing
Session Type: Virtual Paper Abstract
Day: Tuesday
Session Start / End Time: 3/1/2022 05:20 PM (Eastern Time (US & Canada)) - 3/1/2022 06:40 PM (Eastern Time (US & Canada))
Room: Virtual 11

Abstract

With the proliferation of geospatial raster spatial time series, there is a growing demand for methods of to analyze these archives, especially for change techniques . In this paper, we present four time series techniques to detect and model long and short term trends and changes in geospatial time series. These include: linear trends, Short Term Seasonal Trend Analysis (STA), anomalies based on PROPHET and ARIMA. We applied these techniques to three time series case studies: impact of hurricane Rita, rapid urban growth in Ouagadougou and long term environmental changes in the Hindukush region. Our goal is to provide easy to use functions and methods for non Remote Sensing experts within the R ecosystem. In several methods such as PROPHET, we use an anomaly based approach that examines outliers and co-occurring changes in time and space to filter noisy isolated pixels. Results from a MODIS NDVI time series for hurricane RITA show the detection of abrupt change using PROPHET. The same data also shows rapid urban periphery growth in Ouagadougou over the 2001-2016 time period using STA. The Hindukush case study illustrates the detection of long term greening trends using NDVI GIMMS data. These encouraging results confirm the usefulness of the of methods. Future plans include improving functionality, efficiency of algorithms and adding methods to keep pace with the field of change detection. We call the package Geospatial Environmental Change Monitoring (GECM). GECM is was the result of the REACT project funded by BELSPO.