Mapping conifer understory within boreal mixedwoods from Landsat TM satellite imagery and forest inventory information

Abstract

Information about conifer understory within deciduous-dominated mixed-wood stands would increase the possible range of management options for these complex communities, yet cost-effective and accurate inventory methods remain elusive. Maps of conifer understory produced from field-checked photo interpretation were compared with classified images created from Landsat Thematic Mapper data and two image classifiers. The highest accuracy achieved was 71 % using an evidential reasoning classifier that integrated satellite remote sensing observations with stand inventory information. The image map did provide an advantage by capturing some of the spatial variability of conifer understory that is not captured by photo interpretation methods. Predicting the presence and spatial distribution of conifer understory is difficult because its establishment is influenced by many factors such as ecosite, available substrate, distance to seed source and mechanisms of recruitment that are not typically available in spatial formats. The image maps are considered estimates that may be suitable for broad strategic planning, and may serve as validation information for national satellite land cover mapping initiatives.Keywords: boreal mixedwood, conifer understory, image classification, Landsat TM, reflectance, forest inventory, vegetation indexDe l'information concernant le sous-étage de conifères dans les peuplements forestiers mixtes dominés par les espèces à feuilles caduques offrirait une plage plus étendue d'options de gestion de ces communautés complexes, mais des méthodes d'inventaire rentables et exactes restent difficiles à mettre au point. Des cartes du sous-étage de conifères produites d'après l'interprétation de photos vérifiée sur le terrain ont été comparées à des images classées créées à partir de données de l'appareil de cartographie thématique du Landsat sur deux classificateurs d'images. La meilleure exactitude obtenue, qui s'établissait à 70 %, l'a été sur un classificateur par raisonnement de véridiction intégrant les observations par satellite de télédétection et l'information d'inventaire de peuplements. La carte-image offrait un avantage en saisissant une certaine part de la variabilité spatiale du sous-étage de conifères qui n'est pas saisie par les méthodes de photo-interprétation. La prédiction de la présence et de la répartition spatiale du sous-étage de conifères est difficile parce qu'un grand nombre de facteurs influencent son établissement comme l'écosite, le substrat disponible, la distance à la source de semence et les mécanismes de recrutement, facteurs pour lesquels l'information n'est pas toujours disponible sous forme spatiale. Les cartes-images sont considérées comme des estimations pouvant convenir à la planification stratégique générale et peuvent servir d'information de validation dans le cadre d'initiatives nationales de cartographie par satellite de la couverture terrestre.

Deciduous Forest Structure Estimated with LIDAR-Optimized Spectral Remote Sensing

“…Hall et al [49] used Landsat TM to map coniferous understory in a Canadian forest. Xain [63] also attempted an assessment of forest structure using forest inventory data and Landsat ETM+ imagery; however, their best results had an associated error of 51%-75%, which does not meet the accuracy requirements of 80% established for forest inventory by the USGS National Park Service Vegetation Mapping Program [64].…”

“…Modified normalized difference ( ) and modified simple ratio ( ) are two indices that have been found to function better with hyperspectral imagery than the standard normalized difference or the simple ratio [49]. In the model utilized in this study, the and were calculated using Equations (4) and (5).…”

Boreal mixedwood species composition in relationship to topography and white spruce seed dispersal constraint

“…We used two overlapping Landsat TM quarter scenes from the same year to maximise the phenological differences in the hardwoods (Ghitter and Hall, 1995;Hall and Klita, 1997;Hall et al, 2000), and obtained data for bands 3, 4 and 5 (red, near infrared and middle infrared) of a quarter scene. The first image was taken on 14 March and the second on 2 June 1994.…”

Integrating Modelling and Expert Knowledge for Evaluating Current and Future Scenario of Large Cardamom Crop in Eastern Nepal

“…Several studies have been conducted to distinguish understory vegetation using remote sensing approaches. In this regard, evergreen understory vegetation were identified from Landsat images of leaf off-season in deciduous forest [11,12]. Phenological difference between overstory and understory vegetation was also used to detect understory vegetation using Landsat images [13,14].…”