Transition from manual (visual) interpretation to fully automated gully detection is an important task for quantitative assessment of modern gully erosion, especially when it comes to large mapping areas. Existing approaches to semi-automated gully detection are based on either object-oriented selection based on multispectral images or gully selection based on a probabilistic model obtained using digital elevation models (DEMs). These approaches cannot be used for the assessment of gully erosion on the territory of the European part of Russia most affected by gully erosion due to the lack of national large-scale DEM and limited resolution of open source multispectral satellite images. An approach based on the use of convolutional neural networks for automated gully detection on the RGB-synthesis of ultra-high resolution satellite images publicly available for the test region of the east of the Russian Plain with intensive basin erosion has been proposed and developed. The Keras library and U-Net architecture of convolutional neural networks were used for training. Preliminary results of application of the trained gully erosion convolutional neural network (GECNN) allow asserting that the algorithm performs well in detecting active gullies, well differentiates gullies from other linear forms of slope erosion — rills and balkas, but so far has errors in detecting complex gully systems. Also, GECNN does not identify a gully in 10% of cases and in another 10% of cases it identifies not a gully. To solve these problems, it is necessary to additionally train the neural network on the enlarged training data set.

This study focuses on the Kuibyshev reservoir (Volga River basin, Russia)—the largest in Eurasia and the third in the world by area (6150 km²). The objective of this paper is to quantitatively assess the dynamics of reservoir bank landslides and shoreline abrasion at active zones based on the integrated use of modern instrumental methods (i.e., terrestrial laser scanning—TLS, unmanned aerial vehicle—UAV, and a global navigation satellite system—GNSS) and GIS analysis of historical imagery. A methodology for the application of different methods of instrumental assessment of abrasion and landslide processes is developed. Different approaches are used to assess the intensity of landslide and abrasion processes: the specific volume and material loss index, the planar displacement of the bank scarp, and the planar-altitude analysis of displaced soil material based on the analysis of slope profiles. Historical shoreline position (1958, 1985, and 1987) was obtained from archival aerial photo data, whereas data for 1975, 1993, 2010, 2011, and 2012 were obtained from high-resolution satellite image interpretation. Field surveys of the geomorphic processes from 2002, 2003, 2005, 2006, 2014 were carried out using Trimble M3 and Trimble VX total stations; in 2012–2014 and 2019 TLS and UAV surveys were made, respectively. The monitoring of landslide processes showed that the rate of volumetric changes at Site 1 remained rather stable during the measurement period with net material losses of 0.03–0.04 m⁻³ m⁻² yr⁻¹. The most significant contribution to the average annual value of the material loss was snowmelt runoff. The landslide scarp retreat rate at Site 2 showed a steady decreasing trend, due to partial overgrowth of the landslide accumulation zone resulting in its relative stabilization. The average long-term landslide scarp retreat rate is—2.3 m yr⁻¹. In 2019 earthworks for landscaping at this site have reduced the landslide intensity by more than 2.5 times to—0.84 m yr⁻¹.

Evaluation of the vegetation and agricultural-management factor (C-factor) is an important task, the solution of which affects the correct assessment of the intensity of soil erosion. For the vast area of the European part of Russia (EPR), this task is particularly relevant since no products allow taking into account the C-factor. An approach based on automated interpretation of the main crop groups based on MODIS satellite imaging data from Terra and Aqua satellites with the LSTM machine-learning method was used to achieve this goal. The accuracy of crop group recognition compared to the open data of the Federal State Statistics Service of Russia was 94%. The resulting crop maps were used to calculate the C-factor for each month of a particular year from 2014 to 2019. After that, summaries were made at the regional and landscape levels. The average C-factor value for the EPR was 0.401, for the forest landscape zone 0.262, for the forest-steppe zone 0.362, and for the steppe zone 0.454. The obtained results are in good correlation with the results of previous field studies and provide up-to-date (based on 2014–2019 data) estimates of C-factor for rainfall erosion (monthly, annual) with high spatial detail (250 m).

This study addresses the challenge of accurate crop detection using satellite data, focusing on the application of Long Short-Term Memory (LSTM) networks. The research employs a “spatial generalization” approach, where a model trained on one geographic area is applied to another area with similar vegetation conditions during the growing season. LSTM networks, which are capable of learning long-term temporal dependencies, are used to overcome the limitations of traditional machine learning techniques. The results indicate that LSTM networks, although more computationally expensive, provide a more accurate solution for crop recognition compared with other methods such as Multilayer Perceptron (MLP) and Random Forest algorithms. The accuracy of LSTM networks was found to be 93.7%, which is significantly higher than the other methods. Furthermore, the study showed a high correlation between the real and model areas of arable land occupied by different crops in the municipalities of the study area. The main conclusion of this research is that LSTM networks, combined with a spatial generalization approach, hold great promise for future agricultural applications, providing a more efficient and accurate tool for crop recognition, even in the face of limited training data and complex environmental variables.

Unmanned aerial vehicles (UAV) have long been well established as a reliable way to construct highly accurate, up-to-date digital elevation models (DEM). However, the territories which were modeled by the results of UAV surveys can be characterized as very local. This paper presents the results of surveying the Sarycum area of the Dagestan Nature Reserve of Russia with an area of 15 sq. km using a DJI Phantom 4 UAV, as well as the methodological recommendations for conducting work on such a large territory. As a result of this work, a DEM with 0.5 m resolution as well as an ultrahigh resolution orthophotoplane were obtained for the first time for this territory, which make it possible to assess the dynamics of aeolian processes at a qualitatively different level.

A new method of instrumental measurement of the intensity of rill and linear erosion on slopes by the method of terrestrial laser scanning is proposed. This method was tested on four plots in...