Satellite data is increasingly used in agriculture to support decisions about land management, productivity, and resilience in the face of climate variability. From tracking crop growth, ground cover and tree shelter belts to identifying farming practices such as tillage and harvest, satellites provide a consistent, objective view of how landscapes are changing.
Traditionally, understanding these patterns has relied on on-ground field observations, farm records, or local experience. While these remain essential, they can be difficult to scale across large properties or to compare consistently through time. Satellite observations complement on-the-ground knowledge by offering a repeatable, landscape-wide perspective that helps reveal trends that may otherwise be hard to see—such as gradual declines in ground cover, differences in paddock performance between seasons and years, or how vegetation responds to the weather extremes and land management.
Modern satellite data, such as that provided by the European Space Agency through their Sentinel satellites, is freely available, regularly updated, and detailed enough to be useful at the scale of individual paddocks. This opens up new opportunities not only for farmers, but also for land managers, advisors, researchers, and policymakers who are interested in understanding how agro-ecosystems respond to weather conditions, management choices, and long-term environmental change.
As farmers increasingly seek to adopt practices that are more resilient in the face of increasing climate uncertainty and extremes, satellite monitoring of vegetation and management can deliver objective data to understand how different management strategies drive ecological and economic outcomes on agroecosystems.
PaddockTS-web is our new, open online platform designed to make these satellite observations and insights accessible to a broad audience. It provides intuitive visualisations and summaries of vegetation dynamics and management units, “paddocks”, anywhere in Australia. Comparing paddock performance over multiple years, whether visually or quantitatively, can help farmers to better understand how their decisions (and the ensuing weather) shape paddock productivity across their property.
Uses Sentinel-2 observational reflectance data to provide earth observations roughly biweekly with ~10m pixel resolution.
Automatic delineation of likely paddock boundaries, based on change over time for a spectral index, achieved using an AI model.
Two modes:
mode1 No user input required, just select a region and time frame of interest.
mode2 User-provided paddock maps with labelled management each year.
Produces intuitive visual and graphical outputs to summarise paddock-level vegetation history, complementing on-ground knowledge and records.
Predicts land use, crop type and key crop developmental stages, output in standard data tables for further analysis.
Scalable (using the code-based version of mode1) to replicate analyses across Australia.
PaddockTS tracks remotely sensed vegetation indices through time for paddocks located using AI.
In this region, "green-up" came after drought-breaking rains in 2020, as shown by the normalised-difference vegetation index (NDVI). But paddocks can have unique vegetation dynamics.
The paddock outlined in red (left, bottom) stands out with a different signal. Two mid-year dips in NDVI occurred when livestock were let into the paddock to graze around a central dam.
mode1 With no prior information about a location, users simply select a region of interest and time frame of interest, and an automated analysis pipeline will trawl through and analyse satellite and weather data archives dating back as early as 2017.
mode2 Users with paddock maps saved as (file types) and an optional paddock annotation data table (.csv) can gain deeper insights the combine "on the ground" knowledge and satellite data. Paddock annotation data typically record, for a given year, management actions (e.g. which crop was sown, and when sowing/fertilzer/harvest occured), observations of natural phenological transitions (e.g. flowering time) and agro-ecosystem outcomes (e.g. crop yields, soil carbon measurements).
For more computationally heavy uses... PaddockTS source is available on github (under development). Anticipated uses include: 1) running mode1 at very large spatial scales, such as across Australia, and 2) running mode2 with case-specific annotation data and for farm-customised analyses.
Contribute to PaddockTS We are currently working to develop PaddockTS-web so that users can optionally contribute their labelled paddock annotation data for research aiming to explore how management practices jointly influence ecosystem resilience and crop productivity under future climate scenarios.