environmental-time-series

You can detect trends and breakpoints in environmental time series by applying Mann-Kendall, Sen's slope, and BFAST structural change algorithms to NDVI, EVI, or rainfall data. The skill outputs trend_results.csv and decomposition_plot.png for per-pixel or per-site analysis.

To calculate recovery trajectories from NDVI time series, the skill identifies disturbance breakpoints and models post-disturbance recovery patterns. It generates recovery_metrics.csv and recovery_trajectory.png to visualize and quantify the environmental recovery process.

Yes, you can use pandas and numpy for anomaly detection in remote sensing time series. The skill leverages these dependencies alongside scipy to perform seasonal decomposition and highlight unusual behavior, outputting results to anomaly_series.csv and anomaly_plot.png.

The best way to run per-pixel trend analysis on raster time series is using Mann-Kendall and Sen's slope calculations per time unit. This skill processes raster or site-based environmental indicators to identify directional changes and regime shifts across the dataset.

You need to provide a time-series CSV to run the trend analysis. The script processes the input data to generate trend_results.csv, anomaly_series.csv, and decomposition_plot.png, supporting environmental indicators like LST, EVI, and rainfall.