How to Calibrate Historical Aerial Photographs: A Change Analysis of Naturally Dynamic Boreal Forest Landscapes

Abstract

Time series of repeat aerial photographs currently span decades in many regions. However, the lack of calibration data limits their use in forest change analysis. We propose an approach where we combine repeat aerial photography, tree-ring reconstructions, and Bayesian inference to study changes in forests. Using stereopairs of aerial photographs from five boreal forest landscapes, we visually interpreted canopy cover in contiguous 0.1-ha cells at three time points during 1959-2011. We used tree-ring measurements to produce calibration data for the interpretation, and to quantify the bias and error associated with the interpretation. Then, we discerned credible canopy cover changes from the interpretation error noise using Bayesian inference. We underestimated canopy cover using the historical low-quality photographs, and overestimated it using the recent high-quality photographs. Further, due to differences in tree species composition and canopy cover in the cells, the interpretation bias varied between the landscapes. In addition, the random interpretation error varied between and within the landscapes. Due to the varying bias and error, the magnitude of credibly detectable canopy cover change in the 0.1-ha cells depended on the studied time interval and landscape, ranging from −10 to −18 percentage points (decrease), and from +10 to +19 percentage points (increase). Hence, changes occurring at stand scales were detectable, but smaller scale changes could not be separated from the error noise. Besides the abrupt changes, also slow continuous canopy cover changes could be detected with the proposed approach. Given the wide availability of historical aerial photographs, the proposed approach can be applied for forest change analysis in biomes where tree-rings form, while accounting for the bias and error in aerial photo interpretation. that capture both slow and abrupt changes in forest ecosystems are required to understand how forest structure changes in time.Remote sensing provides records over large spatial extent even at remote locations. For example, the Landsat satellites provide observations on global coverage from the early 1970s onwards [6]. However, among remotely sensed records, aerial photographs span the longest time period and currently enable the fine-scale analysis of vegetation dynamics over decades in many locations of the world [7]. Accordingly, aerial photographs have been used to study spatial [8-10] and temporal [3,11,12] vegetation dynamics, even at the scale of an individual tree [13].As with any measurements, data derived from aerial photographs contain bias (systematic error) and error (random variation) [14]. For example, georectification errors [15], the aerial photo scale, defined by the focal length of the camera and the altitude from where the photo was taken [9,16], the quality (e.g., contrast, haziness, and shadowiness) [17,18], and texture and grain size of the photograph [18,19], as well as subjectivity in manual feature quantification [20], potentially produce errors. Hence, da...

The structure of boreal old-growth forests changes at multiple spatial scales over decades

“…The Finnish landscapes and the P. mariana-dominated areas in Lac Dionne generally have open canopy structure characterized by low mean canopy cover, while Pistuacanis landscape and the A. balsamea-dominated parts of Lac Dionne have more closed canopy structure. According to Kulha et al (2018), mean canopy cover at the onset of the studied time period was 18% in Hirvaskangas (year 1959), 16% in Pommituskukkulat (1959), 20% in Hongikkovaara (1972), 21% in Lac Dionne (1965), and 37% in Pistuacanis (1965, quantified over all forested 0.1-ha cells in the landscape. Latest canopy cover information showed a mean canopy cover of 26% in Hirvaskangas (2011), 33% in Pommituskukkulat (2011), 32% in Hongikkovaara (2010), 26% in Lac Dionne (2011), and 37% in Pistuacanis (2011).…”

“…We used the corrected canopy cover values from Kulha et al (2018) to analyze changes in canopy cover over three time intervals in each studied landscape. We define canopy cover as the proportion of forest floor covered by the vertical projections of tree crowns.…”

“…We define canopy cover as the proportion of forest floor covered by the vertical projections of tree crowns. In short, the canopy cover values were visually interpreted for the forested parts of the landscapes (excluding, e.g., waterbodies), using stereopairs of aerial photographs taken at three time points and a grid of 0.1-ha cells (4096 0.1-ha cells per landscape; see Kulha et al 2018 for full details on the used photographs, and their processing and interpretation). Henceforth, we call the photographs from the three time points the oldest, middlemost, and newest photographs (Table 1), and the time interval between the oldest and middlemost photographs the first time interval, the time interval between the middlemost and newest photographs the second time interval, and the time interval between the oldest and newest photographs the whole time interval.…”

“…Briefly, in Kulha et al (2018), tree-ring based reconstructions of forest stand development in randomly selected interpretation grid cells (n = 66) were used to correct visual interpretation bias and to quantify random interpretation error. For this, the growth histories of all living and dead trees with a diameter of C 10 cm at a height of 1.3 within the selected cells were determined from tree-ring samples using standard dendrochronological methods ).…”

“…While remote sensing provides records that cover large spatial extent and currently span decades, the lack of spatially explicit ground-truth values limits their usability in long-term change analysis (Lechner et al 2012). Here, we overcame this limitation by using data from Kulha et al (2018), where tree-ring based reconstructions of forest stand development were used to calibrate remotely sensed data from the past several decades and, importantly, to quantify the measurement error in these data. The calibrated time series and the scenespecific error estimates allowed us to examine changes in forest structure at multiple spatial scales, and to assess the credibility of changes at these scales over different periods of time, thus negating this major shortcoming in the analysis of archived remote sensing data.…”