Estimation of Trees Outside Forests using IRS High Resolution data by Object Based Image Analysis

Abstract

ABSTRACT:Assessment of Trees outside forests (TOF) is widely being recognized as a pivotal theme, in sustainable natural resource management, due to their role in offering variety of goods, such as timber, fruits and fodder as well as services like water, carbon, biodiversity. Forest Conservation efforts involving reduction of deforestation and degradation may have to increasingly rely on alternatives provided by TOF in catering to economic demands in forest edges. Spatial information systems involving imaging, analysis and monitoring to achieve objectives under protocols like REDD+, require incorporation of information content from areas under forest as well as trees outside forests, to aid holistic decisions. In this perspective, automation in retrieving information on area under trees, growing outside forests, using high resolution imaging is essential so that measuring and verification of extant carbon pools, are strengthened. Retrieval of this tree cover is demonstrated herewith, using object based image analysis in a forest edge of dry deciduous forests of Eastern Ghats, in Khammam district of Telangana state of India. IRS high resolution panchromatic 2.5 m data (Cartosat-1 Orthorectified) used in tandem with 5.8 m multispectral LISS IV data, discerns tree crowns and clusters at a detailed scale and hence semi-automated approach is attempted to classify TOF from a pair of image from relatively crop and cloud free season. Object based image analysis(OBIA) approach as implemented in commercial suite of e-Cognition (Ver 8.9) consists of segmentation at user defined scale followed by application of wide range of spectral, textural and object geometry based parameters for classification. Software offers innovative blend of raster and vector features that can be juxtaposed flexibly, across scales horizontally or vertically. Segmentation was carried out at multiple scales to discern first the major land covers, such as forest, water, agriculture followed by that at a finer scale, within cultivated landscape. Latter scale aimed to segregate TOF in configurations such as individual or scattered crowns, linear formations and patch TOF. As per the adopted norms in India for defining tree cover, units up to 1 ha area were considered as candidate TOF. Classification of fine scale (at 10) segments was accomplished using size, shape and texture. A customised parameter involving ratio of area of segment to its main skeleton length discerned linear formations consistently. Texture of Cartosat-1 2.5 m data was also used segregate tree cover from smoother crop patches in patch TOF category. In view of the specificity of the landscape character, continuum of cultivated area (b) and pockets of cultivation within forest (c) as well as the entire study area (a) were considered as three envelopes for evaluating the accuracy of the method. Accuracies not less than 75.1 per cent were reported in all the envelopes with a kappa accuracy of not less than 0.58. Overall accuracy of entire study area was 75.9 per cent with Kappa of 0.59 ...

Automated Classification of Trees outside Forest for Supporting Operational Management in Rural Landscapes

“…Furthermore, previous studies stopped at step 1 of classification. Using high-resolution satellite imagery, Pujar et al [16] automatically mapped three TOF classes (Point, Line, and Patch) with an overall accuracy of 75.1% in an agricultural landscape. In their study based on full-waveform laser scanner data, Straub et al [12] classified four classes: Non-tree vegetation, Forest, Non-forest vegetation-group of trees, Non-forest vegetation-single trees.…”

“…An unsupervised ISODATA classification algorithm was used to separate TOF from other elements. With the same data type, Pujar et al [16] used an object-based approach. A first coarse scale of segmentation was used to classify land cover.…”

“…Concerning the TOF geometrical classification, Straub et al [12] classified TOF on the basis of their form and size into two classes: Single tree objects or Connected groups of trees. Singh and Chand [15], Pujar et al [16], and Seidel et al [10] made a more detailed geometrical classification into three classes: Single objects, Linear objects, and Ample objects. Singh and Chand [15] classified TOF in these three classes by visual interpretation.…”

“…Singh and Chand [15] classified TOF in these three classes by visual interpretation. Pujar et al [16] realized it using spectral and geometric segment parameters computed during the segmentation process in an object-based approach. The overall accuracy was 76%, and the kappa coefficient was 0.59.…”

The Use of Remote Sensing Data to Estimate Land Area with Forest Vegetation Cover in the Context of Selected Forest Definitions

“…Aerial and satellite imagery have been successfully used since the early 2000s to estimate the area of forested areas with an accuracy of 80% for forest areas [20][21][22][27][28][29][30][31][32], 95% for secondary forests [18,19,[33][34][35], and 75% for trees in agricultural areas [36]. Other sophisticated image-based methods include object segmentation and supervised classification [30,35], but few of them consider the geometric features of forest vegetationheight [32,34] and canopy [34] mentioned in the international forest definitions with an overall accuracy of 95%.…”