In this article, prepared together with Ewa Bubula, we investigate whether modern satellite technology can detect subtle changes in crop condition and answer the question: did the rapeseed fields near Kłodawa feel the bitter cold from earlier this year?
From frost to orbit: do satellites see more than we do?
Spring is already in the air and winter coats are making their way to the back of the wardrobe, but for farmers March is the time of reckoning. The real winter that gripped Poland not so long ago left important questions behind. Walking past fields today, we see green growth stirring back to life - but what was actually happening to the plants under the snow and ice just a few weeks ago?
Rather than guess, we decided to look at the matter from a broader perspective - literally. We present an analysis prepared with Ewa Bubula, our GIS analyst, in which we try to assess how severe frosts affected the actual condition of rapeseed crops.
The core assumption of the analysis is that satellite data, specifically from the Sentinel-2 mission, provides an objective and precise tool for making that assessment.
Which raises the obvious question: can you really "see" the effects of frost stress from several hundred kilometres above the ground?
Where does the data come from? Study area and data reliability
For the analysis to be credible, we could not rely on random observations. The foundation of the study was official spatial data from ARiMR, the Agency for Restructuring and Modernisation of Agriculture, for the Wielkopolska region. From these resources, by selecting field boundaries based on the crop code, Ewa extracted the specific plots declared as growing winter rapeseed.
The area chosen for the study was the vicinity of Kłodawa. This particular area of interest (AOI) was selected because the registry showed as many as 958 rapeseed fields, providing enough data for meaningful testing. Ewa defined the boundaries of the study area manually, by visually analysing the crop structure in the region, to ensure the most representative plots were included. This combination of official records and manual verification gives us confidence that the conclusions rest on solid data.
Boundaries of winter rapeseed fields near Kłodawa (Wielkopolska region), mapped from official ARiMR data.
To understand what was happening at ground level, we turned to the global atmospheric reanalysis from the Copernicus Climate Data Store. The dataset used, ERA5 hourly time-series data on single levels from 1940 to present, was produced by assimilating ground-based observations, satellite data and numerical modelling. This gave us a consistent picture of air temperature at 2 metres above ground level throughout the study period.
Copernicus Climate Change Service (2025): ERA5 hourly time-series data on single levels from 1940 to present.
Winter in the lens: comparing before and after the frost event
To understand the scale of change, we needed to travel back in time and compare two key moments from this winter. The first reference point is an image from 30 December 2025. It was a near-perfect day for the satellite: skies were completely clear and there was no snow on the fields. This let us see the rapeseed crops in their natural, undisturbed state.
The situation had changed significantly by the second image, taken on 3 February 2026. The frost event had already passed, and meteorological data from the ERA5 model (Copernicus Climate Data Store) confirms that air temperature 2 metres above ground near Kłodawa dropped dramatically during that period, reaching around -20°C. Although skies were also clear, the fields were dusted with snow. Importantly for our analysis, snow cover was not complete; the white layer did not cover every plant, leaving us a window to check how the rapeseed had coped with the cold.
Comparing these two images illustrates the core challenge of working with optical data in winter. Snow on the ground is a real problem for analysts, as it can effectively mask the true condition of the plants underneath.
Sentinel-2 satellite image of the Kłodawa area, 30.12.2025. Fields clearly visible in green-brown tones, snow-free. Copernicus Sentinel data 2025.
Sentinel-2 satellite image of the Kłodawa area, 03.02.2026. Snow cover and post-frost landscape visible. Copernicus Sentinel data 2026.
Technical challenges: why winter is "expert difficulty"
Working with satellite data in our part of the world is a lesson in patience during winter. The main challenge is cloud cover. At this time of year the sky over Poland is almost permanently overcast, which drastically limits the number of days when the satellite can capture an image usable for analysis. Finding a "clean" image that is simultaneously cloud-free and as snow-free as possible requires painstaking searches through data archives.
Snow itself is another tough nut to crack. While it makes for a picturesque landscape, for satellite sensors it forms a barrier that can significantly distort vegetation condition results. If a field is covered by a white layer, the satellite sees ice crystals instead of plants, making any reliable measurement impossible.
How do we handle this without producing misleading results? The answer is advanced masking technology. We use the SCL layer (Scene Classification Layer), which acts as a filter. This layer classifies every pixel in the image into a specific category (vegetation, soil, water, clouds, snow, cloud shadows, and so on). Based on this classification, a mask is generated that excludes from the analysis any pixels classified as clouds, snow, or other areas unsuitable for further processing. The result is a mask of so-called "good pixels." We focus exclusively on those parts of the fields that are exposed and allow for a reliable assessment of rapeseed condition.
Methodology: how do we "measure" plant condition?
How do you actually check whether rapeseed is "doing well" from space? The key is the NDVI (Normalized Difference Vegetation Index). It works like a health barometer for plants: healthy, dense, green leaves strongly reflect near-infrared radiation while absorbing red light.
Mathematically, we express this with a straightforward formula:
A high score indicates a plant in full vigour; a drop in the value signals stress or damage to plant tissue.
In the analysis, Ewa did not limit herself to a single measurement. To capture the dynamics of change, she calculated the NDVI for both dates - before the frost event and after - and then computed the difference (known as dNDVI) between the two images. This subtraction is what allowed us to objectively identify where rapeseed condition deteriorated most.
Interpreting the results: what does the satellite see?
What did we actually find after applying all the analytical filters to the fields around Kłodawa? Rather than overwhelming you with tables of statistical values like "frost_mean" or "frost_median," below is a clear map in which rapeseed condition has been colour-coded by category.
To keep the assessment objective, we applied specific thresholds for drops in the vegetation index:
The results indicate that the frost episode may have affected the condition of some winter rapeseed crops, while the majority of fields showed no significant change or only a moderate drop in the vegetation index. The conclusion from the analysis is clear: satellite data objectively confirms that the frosts affected the plants, triggering measurable physiological stress.
Final analysis output in QGIS with the NDVI change legend applied. Contains modified Copernicus Sentinel data 2025/2026.
Conclusions: this is only half the story
Today's technology gives us remarkable capabilities, but working with optical data in winter comes with real limitations. High cloud cover during this season means that clear weather windows, when the satellite can capture a usable image, are rare.
Snow on the ground can also create a barrier that makes real-time assessment of crop condition difficult. At the same time, and this is worth noting in the context of a frost event, snow actually acts as a natural insulating layer for crops, protecting them from the coldest temperatures.
Despite these challenges, we managed to capture the moment when nature tested the resilience of Wielkopolska's fields. The full scale of this winter's impact on yields will only be assessable in spring, when vegetation picks up again and images become cleaner and free from snow interference. We are already planning a return to Kłodawa during the crop regrowth period - stay with us to see how the rapeseed recovers from its winter ordeal.