Monitoring war destruction from space using machine learning

Existing data on building destruction in conflict zones rely on eyewitness reports or manual detection, which makes it generally scarce, incomplete, and potentially biased. This lack of reliable data imposes severe limitations for media reporting, humanitarian relief efforts, human-rights monitoring, reconstruction initiatives, and academic studies of violent conflict. This article introduces an automated method of measuring destruction in high-resolution satellite images using deep-learning techniques combined with label augmentation and spatial and temporal smoothing, which exploit the underlying spatial and temporal structure of destruction. As a proof of concept, we apply this method to the Syrian civil war and reconstruct the evolution of damage in major cities across the country. Our approach allows generating destruction data with unprecedented scope, resolution, and frequency—and makes use of the ever-higher frequency at which satellite imagery becomes available.

A brief history of ball lightning observations by scientists and trained professionals

With thousands of eyewitness reports, but few instrumental records and no consensus about a theory, ball lightning remains an unsolved problem in atmospheric physics. As chances to monitor this transient phenomenon are low, it seems promising to evaluate observation reports by scientists and trained professionals. The following work compiles 20 published case histories and adds 15 from the author's work and 6 from a Russian database. Forty-one cases from eight countries, 1868–2020, are presented in abstract form with a synthesis. The collection of cases does not claim to be complete. Six influential or notable ball lightning cases are added. It is concluded that well-documented cases from trained observers can promote fieldwork and stimulate and evaluate ball lightning theories. Scientists who have not reported their experience are invited to share it with the author.

Tornadoes in the Russian Regions

New data are presented on tornadoes over land in Russia for the period of 1900–2018 based on various sources. In total, information on 1763 tornadoes was collected, including 993 tornadoes from eyewitness reports on tornado passage and/or associated impacts and 770 tornadoes from satellite data on tornado-induced windthrows. Both single tornadoes and tornado outbreaks, the cases of formation of several tornadoes within one meso- or synoptic-scale system, were reported. On average for 2009– 2018, more than 100 tornadoes are observed in Russia per year, including 15 significant tornadoes (with a wind speed >50 m/s) and one intense tornado (with a wind speed >70 m/s). In some years, these rates can be significantly higher and reach 342, 52, and three tornadoes per year, respectively. Tornadoes are observed on about 41 days per year, up to 68 days per annum in some years. The frequency of occurrence of tornadoes of different categories and the probability of their passage over a point on the ground were estimated. These estimates can be further used to assess a risk of tornado-hazardous situations. The general underestimation of the number of tornadic events in routine meteorological observations and existent statistics is discussed.

Seismic observations, numerical modeling, and geomorphic analysis of a glacier lake outburst flood in the Himalayas

Glacial lake outburst floods (GLOFs) are a substantial hazard for downstream communities in vulnerable regions, yet unpredictable triggers and remote source locations make GLOF dynamics difficult to measure and quantify. Here, we revisit a destructive GLOF that occurred in Bhutan in 1994 and apply cross-correlation–based seismic analyses to track the evolution of the GLOF remotely (~100 kilometers from the source region). We use the seismic observations along with eyewitness reports and a downstream gauge station to constrain a numerical flood model and then assess geomorphic change and current state of the unstable lakes via satellite imagery. Coherent seismic energy is evident from 1 to 5 hertz beginning approximately 5 hours before the flood impacted Punakha village, which originated at the source lake and advanced down the valley during the GLOF duration. Our analysis highlights potential benefits of using real-time seismic monitoring to improve early warning systems.

Выставка изобразительного искусства, посвященная 500-летию эпоса «Джангар»: первые профессиональные художники Калмыкии

The article examines the emergence of Kalmykia’s professional arts activities in the 1930s. Goals. It seeks to review formation stages of the ethnic school, determine its essentials and local features as part of Soviet artistic culture. In this regard, a fundamental research problem is that the vast majority of local pre-war visual arts works had been lost during the Great Patriotic War and Kalmyk deportation (1943–1956). Materials and Methods. Considered are works of Kalmykia’s first professional artists ― such as I. Nuskhaev, L. Ochirov,R. Bogoslavsky, etc. ― inspired by epic heritage and realia of Socialist construction. The study employs a historical-and-cultural approach which determines the use of sources (archival papers, eyewitness reports, available pictorial works and studies); characterizes the pre-war life of the republic and its integration into Socialist cultural environment. Results. The paper reveals a central event of the era, examines the preparation and arrangements of the 1940 jubilee exhibition to celebrate 500 years of the Kalmyk heroic epic of Jangar. Conclusions. Cultural heritage proved most essential to the formation of Soviet Kalmykia’s pictorial arts trend, central to the latter being traditions of Russian realistic art represented by ethnos-related works of V. Favorsky and his colleagues.

Ambient conditions prevailing during hail events in central Europe

Around 26 000 severe convective storm tracks between 2005 and 2014 have been estimated from 2D radar reflectivity for parts of Europe, including Germany, France, Belgium, and Luxembourg. This event set was further combined with eyewitness reports, environmental conditions, and synoptic-scale fronts based on the ERA-Interim (ECMWF Reanalysis) reanalysis. Our analyses reveal that on average about a quarter of all severe thunderstorms in the investigation area were associated with a front. Over complex terrains, such as in southern Germany, the proportion of frontal convective storms is around 10 %–15 %, while over flat terrain half of the events require a front to trigger convection. Frontal storm tracks associated with hail on average produce larger hailstones and have a longer track. These events usually develop in a high-shear environment. Using composites of environmental conditions centered around the hailstorm tracks, we found that dynamical proxies such as deep-layer shear or storm-relative helicity become important when separating hail diameters and, in particular, their lengths; 0–3 km helicity as a dynamical proxy performs better compared to wind shear for the separation. In contrast, thermodynamical proxies such as the lifted index or lapse rate show only small differences between the different intensity classes.