A method for estimation of leaf area index (LAI) using Sentinel-1 synthetic aperture radar (SAR) C-band dual-polarization image was developed for temperate deciduous and mixed forests. Instead of usual backscattering ratio the regressional relationship was restored between LAI and the relative difference polarization index (RDPI) adjusted by the mutual orientation of radar viewing direction and normal line of land surface local element. The mean absolute error of the LAI estimates was 0.11 for all forest types within study site. The specific mean LAI estimation error was 0.057 for hornbeam (Carpinus betulus), 0.058 for beech (Fagus sylvatica) and 0.066 for pine (Pinus sylvaticus) dominated stands.

Forests of Slovak Eastern Carpathians provide essential services to society ensuring human well-being. Primeval forests that remained there untouched have exceptional value. Unsustainable logging and climate change intensify forest disturbances and may cause substantial forest loss. The engagement of remote sensing allows timely monitoring and prediction of negative consequences. The leaf area index is selected from the other forest’s condition indicators to assess the forests’ quantity and disturbance. The conceptual approach for risk assessment of forest disturbance is proposed. This approach uses multiscale multi-temporal Earth observation data, such as synthetic aperture radar imagery and high-level satellite data products. The output response of the forest’s condition was restored by multivariate regression with radar backscattering coefficients, relative difference polarization index, and local incidence angle. The low-resolution leaf area index pre-developed Copernicus data products make it possible to characterize the forest’s condition without ground truth measurements. According to the elaborated methodology, a time series of the leaf area index was mapped, and risks analysis was performed according to hazard functions. A resulting risk’s gradations map is a good tool for future research and decision-making support in nature management. The high spatial resolution of the output maps allows getting a more detailed assessment of forest disturbance behavior.

Geostatistical and geospatial assessment of soil pollution with heavy metals in Pavlograd city (Ukraine)

The subject of this article is the concept of a geoinformation platform for landmine detection. Modern warfare and its increasing scale have become a relevant topic today. Undetected explosives threaten business (agriculture, logistics, etc.) and human lives. The problem becomes more acute with the rapid extension of minefield areas, which requires significant time and resources and carries high risks. Remote sensing leverages landmine detection possibilities, providing useful information about landmine displacement with no additional risk during data collection over a large area. This study aims to present a combined approach for revealing hidden landmines using UAVs equipped with different sensor types. The tasks to be solved are to define the overall structure and components of the geoinformation platform, choose the technological solutions for each of them, and implement the system prototype that makes it possible to extend its configuration in the future. The methods used are remote sensing, automated object detection, and centralized data processing in a geographic information system (GIS). Multispectral imagery and magnetometric remote measurements create the background information required to detect landmines and other explosive objects. The results of this study provide a general framework, i.e., a geoinformation platform for landmine detection and mapping. The tasks include UAV-based remote data gathering, UAV mission planning and flight control, data processing and mapping via general GIS, and updating new landmine signatures in the corresponding database. The landmine detection process uses information from the landmine signature database to verify suspicious objects. The results are presented in the form of a probabilistic map, which supports the decision-making process of demining. Conclusion. The proposed approach significantly decreases the time required for landmine detection and mitigates demining risks, which is crucial for dealing with the consequences of war. At present, the concept is being developed in the form of a geoinformation platform research prototype involving an open-source Quantum GIS (QGIS) software system and Python programming language, which is used to create plug-ins for QGIS. The entire landmine remote detection process can be fully automated. Future studies will involve extensive experimental testing and may involve convolutional neural networks (CNN) as a detection mechanism.