COS 21-6 - Fifty years of forest cover dynamics in Latvia: New findings from declassified CORONA spy satellite imagery

Monday, August 7, 2017: 3:20 PM
B114, Oregon Convention Center
Zigmars Rendenieks, Mihai Daniel Nita and Volker C. Radeloff, Forest and Wildlife Ecology, University of Wisconsin-Madison, Madison, WI

During the 20thcentury Latvia experienced major changes in land cover, and forest cover has been particularly dynamic. Two World Wars, land nationalization and collectivization, agricultural abandonment and the increase of forest cover under Soviet rule, and changes in ownership after the collapse of socialism, have all affected Latvia’s forests. Our goal was to study the changes in forest cover from 1965 to 2016, and to quantify the properties of new forests growing on former agricultural land. We analyzed declassified CORONA spy satellite images from 1965-1972, and optical satellite images from Landsat 8 for 2016. Detailed stand-level forest inventory data provided ancillary information on forest stand characteristics. We applied Structure-from-Motion photogrammetry to rectify CORONA images and object-based image segmentation and supervised classification to generate land cover maps. We employed analyses of landscape patterns and statistical models to quantify and interpret spatial changes detected in thematic maps.


We found that net increase in forest cover from 1965 to 2016 was 14.2%, but with strong regional variation. In the hilly upland regions with a mosaic of forest patches and pastures (Vidzeme upland and Latgale upland), there was much more abandonment and reforestation than in the flat lowland regions with higher population density and large, contiguous patches of intensively managed agricultural land (Zemgale and Ziemelvidzeme). Boundaries of forests that were state-owned prior to 1940 changed little after the 1960s, and the majority of new forests grew on previously privately-owned agricultural lands, and were dominated by grey alder (34.0% of all new forest area) and birch (24.7%) with low-quality timber. Our results provide a better understanding about spatial and temporal patterns of land abandonment and forest growth. These findings have applications in land management and conservation planning – for example, in the context of continuous land abandonment, our findings help to assess the contribution of young forests (younger than 50 years old) to forest biodiversity in the Eastern Baltic region.