Background of the Environmental Time Series Analysis app

Powered by Sensing Clues and CorrelAid

The Environmental Time Series Analysis app is a monitoring tool that can be used for

  • Monitoring vegetation health

  • Deforestation studies

  • Drought monitoring

  • Land cover classification

  • Built-up expansion monitoring (NDBI)

  • Flood monitoring (NDWI)

  • Forest fire-risk analysis (NDMI)

To this end the app includes 3 visualisations:

  1. Charts, visualising the average NDVI-values per month throughout a selected year.

  2. Maps, visualising the same information geospatially per month throughout a selected year.

  3. Land-cover explorer, to analyse the above per land cover class.

Where NDVI is the abbreviation for Normalised Difference Vegetation Index, an often used metric for monitoring change in  (protected) areas. In future versions of this app, we will also support other indices, such as the NDMI (moisture), NDWI (water), NDBI (Built-up), the EVI (which improves on NDVI in areas with dense vegetation), and more. 

Time series - chart version

The time series line chart shows the temporal dynamics of vegetation for the selected region over a 12-month period, highlighting trends and seasonal variations in vegetation health. Higher NDVI values typically indicate denser and healthier vegetation, whereas lower values may signal sparse vegetation, stress, or land cover changes due to environmental factors such as drought, deforestation, or agricultural activity.

To generate the figure, please select a country, month, and year. The system will process the NDVI values for the selected region and display the corresponding time-series chart, allowing for easy interpretation and comparison across different time periods.

Land-cover explorer

The land-cover explorer provides insight in NDVI values, averaged over an area of interest for a specific land cover type.

Users can analyse how NDVI fluctuates throughout the year, observing peaks during growing seasons and declines during dry or non-growing periods. This visualization is particularly useful for agricultural monitoring, ecosystem assessments, and climate impact studies.

To generate the figure, please select a country, month, year and land cover class. The system will process the NDVI values for the selected region and display the corresponding time-series chart for the land cover type.

Time series - map-version

The map series provides a spatial representation of NDVI values across the selected region, with each pixel displaying the NDVI value at a specific geographic coordinate. This visualization allows users to identify spatial patterns in vegetation health, detect localized anomalies, and compare NDVI values across different areas within the Area of Interest (AoI).

In addition to absolute NDVI values, users can also compute the Delta NDVI, which represents the difference between the NDVI of the current month and the NDVI of the same month in previous years. The Delta NDVI heatmap highlights areas where vegetation health has improved or deteriorated compared to historical averages.

To generate the series, select a country, month, and year. The system will process the NDVI values and, if requested, compute the Delta NDVI to show how vegetation has changed relative to historical data.

Data Collection and Preprocessing

Data Sources

The NDVI data used in this project is obtained from the Sentinel-2 satellite imagery, specifically the "COPERNICUS/S2_SR_HARMONIZED" dataset, through Google Earth Engine (GEE). This dataset provides atmospherically corrected surface reflectance imagery, making it suitable for vegetation analysis and monitoring.

Area of Interest (AoI) and Time Period

The AoI is defined based on asset files available within GEE. The data collection is conducted over a user-specified time period, with the minimum granularity set at a single month. This allows for consistent temporal analysis and trend identification.

Data Preprocessing Steps

  • Sentinel-2 surface reflectance images were filtered based on the AoI and specified date range.

  • Cloud and other unwanted pixels were masked using the Scene Classification Layer (SCL) band.

  • NDVI was computed for each image using the formula: NDVI = (B8 - B4) / (B8 + B4); where B8 corresponds to the near-infrared (NIR) band, which captures vegetation reflectance, and B4 corresponds to the red band, which captures vegetation absorption.

  • A mosaic operation was performed to create a composite NDVI image for each month. This composite represents the best available NDVI values within the given month, reducing the impact of missing or low-quality data.

Data Export

The processed NDVI composite images were exported to Google Drive, clipped to the AOI, and stored using the defined spatial resolution and coordinate reference system (CRS). The exported filenames include the processing month and country name for easy identification.

Credits

This app has been developed by CorrelAid and Sensing Clues and piloted in Zambia at the Mponda project of By Life Connected.