Definitions

Lead: A lead designates an unconfirmed, open source report that a heritage site has been damaged, and is therefore utilized as a starting place rather than as a confirmation. Common leads include social media posts, single-source interviews, and unconfirmed news reports.

Potential Impact: A potential impact is a classification used to designate cultural heritage with possible damage. Sites are classified as “potential impacts” when they are determined to have a moderate-to-high probability of having sustained damage due to direct overlap with, or proximity to, remotely-sensed kinetic activity. Both leads and potential impacts are used as tips for queuing further investigation via satellite imagery (i.e., “tip and queue”).

Confirmed Impact: Leads and potential impacts are re-classified as confirmed impacts when additional, independent lines of intelligence confirm that a site has indeed been impacted. Most impacts are confirmed through the identification of damage in satellite imagery. Other confirmations come from multi-sourced news, social media, or Russian news reports, such as reporting on Russia's seizure of artifacts in the Melitopol Museum.

Methodology

This report is the second in a series of confirmed impact analyses. The data presented here represents analytical confirmations between 1 July and 31 August 2022, and not necessarily the dates of impact. To be clear, the July-August window does not imply that the impacts occurred between these dates, merely that CHML analysts confirmed an impact to a site during this time. All impacts occurred from 24 February 2022 onwards. The figure of 207 confirmed impacts adds the more recent 99 confirmed impacts to the last report's 108 confirmed impacts.

The basis of this research is a 28,000-site GIS database maintained at CHML. CHML is housed at the Virginia Museum of Natural History and monitors cultural heritage destruction around the world in partnership with the Smithsonian Cultural Rescue Initiative. The development of the Ukrainian dataset was initiated in April 2021 in order to identify cultural heritage destruction in eastern Ukraine and Crimea after the Russian annexation of the peninsula in 2014. Initially, the inventory consisted of 26,000 sites, but additional data scrapes and contributions by researchers in the field have contributed to a current inventory of over 28,000 sites. CHML’s geographical dataset recognizes Ukraine’s pre-2014 borders, and thus includes heritage sites in eastern Ukraine and Crimea. The research methodology did not include an analysis of damage attribution, and thus the results may include damage caused by either Russian or Ukrainian Armed Forces. This data does not include monuments that were removed by the Ukrainian government, for example, portions of the Arch of Freedom of the Ukrainian People, which were dismantled on 26 April 2022.

CHML's cultural heritage inventory was developed in collaboration with the Smithsonian Cultural Rescue Initiative. The categories of historical and cultural spaces that qualify as cultural heritage sites were taken from the Cultural Heritage Site List (CHSL) data standards, developed by the Penn Cultural Heritage Center (PennCHC) at the University of Pennsylvania Museum (National Science Foundation Grant #1439549). The geographic dataset was derived from open sources, academic publications, gray literature, and the contributions of professional colleagues. New sites are regularly added to the dataset. Nevertheless, there are some notable omissions. CHML's inventory of 28,000 sites does not fully account for all of the cultural heritage sites in Ukraine—in particular, the great number of governmental and educational buildings that are also historic structures. There are also hundreds (if not thousands) of new burial and mass burial sites (from 24 February 2022 onwards) in Ukraine and these have not been added to the inventory.

From February to August 2022, CHML analysts monitored over 28,000 cultural heritage sites in Ukraine using open source research, remote sensing, and high-resolution commercial satellite imagery. The inventory is maintained and visualized in the GIS, with monitoring activities providing inputs to a “site status” field. Monitoring occurred every 48 hours, with the status for each cultural site remaining or shifting between “no impact detected,” “potential impact,” or “confirmed impact.”

Remote fires sensing data released by NASA FIRMS is pulled and overlaid on the cultural inventory GIS layer. Sites in the geographic areas overlapping FIRMS locations are marked as potential impacts. From there, CHML analysts review each potential impact with commercial satellite imagery to identify whether there was any visible damage to the site. Simultaneous research on the open source provides additional reporting of potential impacts via social media and news media. These leads are geolocated so that they can also be queued for confirmation via commercial satellite imagery. Open source leads in particular may document damage to a site that is not visible in commercial satellite imagery, so many of these impacts remain unconfirmed. The conservative nature of CHML’s approach means that confirmations should be taken as a minimum number of actual impacts.

Operational Security Note

Highly detailed and precisely geolocated heritage site data is not included in this study for operational security purposes. However, abstracted and aggregated analyses have value to show broad trends and categories.

New and Updated Total Findings with Charts and Heat Map Visualization

Trends

This new tranche of confirmations generally follows the same patterns observed in the first report of confirmed impacts, with the notable addition of impacts to the Conservatory/Concert Hall category. Thanks to open source reporting concerning impacts to these sites, it was recognized that musical venues were not in CHML’s inventory of heritage sites.

Like the conclusion made in CHML’s first report, analysis of Table 1 suggests that there are inherent resolution problems in the satellite imagery confirmation process. The impacts likely overrepresent larger sites and structures and underrepresent smaller sites, such as memorials and monuments. Damage to sites with a smaller or less visible footprint is less likely to be recognized as such in satellite imagery. The sites will require the use of additional confirmation methods to determine if damage has occurred.