Earth observations
Since 2021 we started to deeper investigate the power of Earth observations using Sentinels from the Copernicus program. We are mainly interested in local applications concerning vegetation and water resources. Our recently developed expertise is about city monitoring, vegetation coverage evolution, drought alert or heat island search.
City vegetation monitoring
As part of the Interreg project “Satellite monitoring of the region as support for public administration in climate change adaptation strategies” (NFP403201DJK6), we are engaged in monitoring cities. We use satellite data from the Copernicus program and monitor chlorophyll concentration and its development over time. The goal is to identify sensitive areas in terms of climate change.
We focus on data from the Sentinel 2 satellite and are interested in the condition of vegetation in cities. Interested parties from public administration can join the project, and upon agreement, we can include the city/municipality in the monitored areas.
Time development of vegetation
By combining specific wavelengths in the red and infrared ranges, we can monitor chlorophyll concentration and thus assess the condition of vegetation. Our pilot study monitors two representative sectors of Trnava. One is a park dominated by trees. The second sector includes part of a housing estate where the vegetation is predominantly grass. We measure the value of the so-called NDVI (Normalized Difference Vegetation Index), which expresses the amount of chlorophyll in the monitored section of the city. We average the NDVI value from March 1 to the end of October since 2018.
The following graphs show how the years 2018 to 2021 behaved very similarly, but the very dry year 2022 shows a sharp decline in the NDVI value since the beginning of the year. So far, 2023 is behaving similarly to the years 2018 to 2021.
The following figure shows a direct comparison of individual years. The vegetation growth profile is very similar. From March to June, the NDVI value rises and, after peaking, begins to decline slightly. We assume that the decline is caused by warm summers and gradual drying. This is most noticeable in the case of grass cover, which is not protected by a tree layer. The decline continues into August, and then the data is scattered. Either they decline or return to approximately the peak value from before the summer. The behavior from August onwards probably depends on the amount of precipitation.
The year 2022 behaved very similarly, except that the beginning of March was significantly different. Further developments until July are standard compared to other years. In July, the NDVI reached its lowest values compared to other years. This is when the drought peaked. The onset of rainfall at the end of summer improved the situation and the NDVI rose.
The analysis shows low NDVI values already at the beginning of monitoring in the dry year of 2022, namely in March. We do not know whether this is directly related to the upcoming drought in summer, but it may be an indicator of the overall condition of vegetation in the coming year. If so, the drought in 2022 could have been predicted immediately after winter. This would mean that the origin of the drought is not in the precipitation during the monitoring period, but rather in winter. So far, 2023 is behaving similarly to the years 2018 to 2021.
In the next graph, we compare NDVI data with average precipitation totals from SHMU. We compare the NDVI value with the average precipitation for several selected months. We combine the months from October (10) of the previous year to March (03) of the analyzed year. Precipitation totals show a similar trend to the NDVI profile when the beginning of the year (months 01, 02, and 03) is included. So, if we take the end of the previous year (data 10-12), the precipitation trend does not describe the situation well, as it includes a locally strong minimum in 2019 that is not visible in the NDVI data. When we add the months from the beginning of the year (data 10-01, 10-02, 10-03), the decline in 2022 is more pronounced. The study shows that the drought in 2022, or at least the low NDVI values at the beginning of the year, is strongly influenced by weak precipitation activity in the months of January to March.
Data from January to March and December to March show a decline since 2021, but this is not reflected in the NDVI trend. However, data from October to December and October to March show a significant peak in 2021. It is possible that the lower precipitation total in January to March 2021 was compensated by increased precipitation at the end of 2020 in October to December. The year 2022 apparently did not have such compensation. We therefore conclude that the impact of the autumn months is also important.