Detecting new Buffel grass infestations in Australian arid lands: evaluation of methods using high-resolution multispectral imagery and aerial photography

Abstract

We assess the feasibility of using airborne imagery for Buffel grass detection in Australian arid lands and evaluate four commonly used image classification techniques (visual estimate, manual digitisation, unsupervised classification and normalised difference vegetation index (NDVI) thresholding) for their suitability to this purpose. Colour digital aerial photography captured at approximately 5 cm of ground sample distance (GSD) and four-band (visible–near-infrared) multispectral imagery (25 cm GSD) were acquired (14 February 2012) across overlapping subsets of our study site. In the field, Buffel grass projected cover estimates were collected for quadrates (10 m diameter), which were subsequently used to evaluate the four image classification techniques. Buffel grass was found to be widespread throughout our study site; it was particularly prevalent in riparian land systems and alluvial plains. On hill slopes, Buffel grass was often present in depressions, valleys and crevices of rock outcrops, but the spread appeared to be dependent on soil type and vegetation communities. Visual cover estimates performed best (r2 0.39), and pixel-based classifiers (unsupervised classification and NDVI thresholding) performed worst (r2 0.21). Manual digitising consistently underrepresented Buffel grass cover compared with field- and image-based visual cover estimates; we did not find the labours of digitising rewarding. Our recommendation for regional documentation of new infestation of Buffel grass is to acquire ultra-high-resolution aerial photography and have a trained observer score cover against visual standards and use the scored sites to interpolate density across the region.

UAVs and Machine Learning Revolutionising Invasive Grass and Vegetation Surveys in Remote Arid Lands

“…The present study represents a competitive approach for the use of UAVs and machine learning-based classification models compared with alternative solutions. It complements the research outcomes on buffel grass of Marshall et al [ 19 ] by confirming a feasible, accurate, lightweight and relatively cheap solution for invasive grass mapping. With regard to invasive grasses in arid lands, this paper has demonstrated that using only high-resolution RGB images and single pixel-wise classification satisfies the need for accurate and efficient detection and segmentation solutions.…”

“…Research from Olsson et al [ 18 ], for instance, demonstrated the importance of using hyperspectral imagery for invasive grass detection as compared with satellite imagery. A feasibility study of sensing technology by Marshall et al [ 19 ], for example, illustrates the potential for regional mapping of buffel grass infestations in arid landscapes using high-resolution aerial photography in red, green, blue (RGB) colour model at a cm/pixel scale over multi- and hyperspectral technologies for overall detection rates.…”

Toward a High Spatial Resolution Aerial Monitoring Network for Nature Conservation—How Can Remote Sensing Help Protect Natural Areas?

“…Applying the latest remote sensing technologies in an innovative way can expand the range of monitoring tools and enhance nature conservation and scientific research efficiency. Demands on the use of high-resolution remote sensing procedures and spatial data in nature conservation projects [9][10][11][12], in environmental and restoration ecology [13][14][15], mapping and monitoring natural phenomena are growing [16][17][18][19]. However, data collection is confronted with the controversial challenges of increasing spatial resolution, higher frequency of surveys, and ever-faster evaluation providing readily available information on increasingly large areas [20][21][22][23][24][25][26][27].…”

Comparing Pixel and Object-Based Approaches to Map an Understorey Invasive Shrub in Tropical Mixed Forests

“…However, VHR data are increasingly proving to be of much use in the process of invasion mapping specifically because of the precision and detail that these data provide in separating signatures of different land cover types. These data have been used largely in temperate regions as well as grasslands to single out invasion by one or two species occurring in homogeneous stands ( Laliberte et al, 2004 ; van Lier et al, 2009 ; Marshall et al, 2014 ). Although VHR data is now widely being used for invasion mapping, many scientists argue that the amount of detail provided by these data may sometimes prove to be too much for easy separation of intended classes ( Nagendra and Rocchini, 2008 ).…”