Disturbance is a major driver of forests ecosystem dynamics. The legacy of historical disturbance to ecosystems is complex and difficult to assess mostly due to the lack of information. For shifts to be quantified, accurate historical forest disturbance estimates are required as a baseline for examining current trends.
Before Landsat Missions, in areas covered with aerial images, mapping forest disturbances on is either expensive or time consuming. For large areas across globe, the first generation of U.S. photo intelligence satellites collected more than 860,000 images of the Earth’s surface between 1960 and 1972, filling the gap in history covering 10 more years back in time with high resolution images.
But since declassification, few projects based on DISI (Declassified Intelligence Satellite Imagery) were developed on large areas. One important set back on using DISI was that processing these images without fiducials or RPC (Rational polynomial coefficients) is time consuming. Our goal was to develop a method for using the US reconnaissance satellite – DISI, in the mapping process.
We developed a novel fast mapping method for applying geocorrection to the original DISI files. The method is based on Structure from Motion (SfM) which is an innovative algorithm from computer vision which combines feature matching across the photos and a bundle-adjustment technology for differential rectification of relief displacement.
Testing the method showed an average of 3 hours processing time for a DISI stereographic pair covering 4000 km2, on a 2 meter GSD (Ground Sampling Distance), with 24 ground control points. The 3D accuracy of final products was between 2-30 meters, mostly influenced by the terrain and quality of the raw images.
By employing novel methods for applying geocorrections we were able to reconstruct forest cover and forest disturbance a large area as far back as the mid 1960 using DISI. Furthermore, the methodology that we developed can be easily applied to many other regions of the world, where historical satellite imagery is available as well as to other historic imagery products.
Mihai is Associate Professor at Department of Forest Engineering, Transilvania University of Brasov. His subject is Forest Hydrology and Hydrotechnical Terrestrial Measurements. As researcher he was involved projects regarding forest management and monitoring, mainly as GIS expert on topics like LULUCF (land use, land-use change and forestry) using remote sensing and drone mapping, modelling habitat distributions for species, evaluating Forest Ecosystem Services (FES), developing and managing GIS databases and WebGIS apps.
Starting September 2016 as Fulbright Scholar at University of Wisconsin-Madison, under the supervision of prof. Volker Radeloff, Mihai is exploring the effect of wars and political conflict on forest ecosystems in Romania by assessing the forest disturbances during Soviet occupation (1945 - 1956) using a heuristic method based on SfM (Structure from Motion) and DISI (Declassified Intelligence Satellite Imagery) and comparing with the recent rates of disturbance in forests after 1990.