Detecting Invasive Species and Soil Quality Indexes by using Imaging Spectroscopy: two ECOPOTENTIAL related studies from Ben Gurion University presented at the 11th EARSeLSIG imaging spectroscopy workshop

06-08/02/2019, Brno, Czech Republic

The 11th workshop of the EARSeL Special Interest Group on Imaging Spectroscopy has taken place in Brno, Czech republic, on February 06-08, 2019. The special interest group encourages international discussion among specialists working with innovative spectral Earth Observation technologies and organizes Workshop every two years (the last one was held in Zürich) and with more than 20 years long tradition it is recognized as the leading Workshop in this field in Europe, as well as worldwide, bringing together students and professionals from universities, research organizations and private companies to present, exchange and discuss new ideas and research achievements related to all aspects of imaging spectroscopy. Imaging spectroscopy is addressing today’s key environmental and societal challenges. It is also expanding from traditional airborne platforms towards new ground-based, unmanned airborne and satellite systems. The workshop illustrated advanced methods and applications in Earth and Environmental science, dealing with both airborne and satellite technologies. 

Among those, two ECOPOTENTIAL related works were presented by the Remote Sensing Laboratory of Ben Gurion University (Israel), partner in ECOPOTENTIAL, together with the Agricultural Research Organisation – Volcani Centre.

The two studies focussed on mapping invasive plant species by monitoring their phenological characteristics, and on the use of Imaging Spectroscopy for assessing the quality of soils in relationship to land use. Here the relevant information and the abstracts. The programme of the workshop can be read at this URL:

Integrated Hyperspectral and Multispectral Approach for Mapping Invasive Plant Species Based on Phenological Characteristics

Tarin Paz-Kagan (1), Natalya Panov (2), Micha Silver (2), Arnon Karnieli (2)

(1) Agricultural Research Organization Volcani Center, Israel; (2) The Remote Sensing Laboratory, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Sede-Boker Campus

The spread of invasive plant species (IPS) has been recognized as the second most important threat to biodiversity after habitat destruction. Since the spatial extent of IPS is essential for managing invasive species, the current study aims to identify and map the aggressive invasion plant species of Acacia salicina and Acacia saligna in the coastal plain of Israel. This goal was achieved by integrating airborne imaging spectroscopy with spaceborne multispectral remote sensing data.

We developed an integrated approach for mapping the IPS based on the phenological flowering stage, using hyperspectral and multispectral images. The hyperspectral images at higher spatial and spectral resolutions were used to train the multispectral images at the species level. We incorporated a series of statistical models to classify the IPS location and to recognize their distribution and density. A Support Vector Machine (SVM) classification algorithm was applied twice: first for species classification with the hyperspectral data, then with multispectral data, taking advantage of the flowering phenology, using the trained output data from the first step. 

The classification yielded an overall kappa coefficient accuracy of 0.89 in the multispectral image. Additionally, we studied the influence of various environmental and human factors on the IPS’s spreading by using a random forest (RF) model to understand the mechanisms underlying successful invasions and to assess where IPS have a greater likelihood of occurring. This algorithm revealed that high density of Acacia is positively correlated to elevation; temperature pattern; and distances from rivers, settlements, and roads. Our results demonstrate how integration of remote sensing data in different spectral resolutions assists in determining IPS proliferation, and provides detailed geographic information for conservation and management efforts to prevent their future spread.

Using Imaging Spectroscopy for Detecting and Mapping of Land-Use Effects on Soil Quality in Dryland

Nathan Levi (1), Arnon Karnieli (1), Tarin Paz-Kagan (2)

(1) The Remote Sensing Laboratory, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Sede-Boker Campus 8499000 Israel; (2) Agricultural research organization (ARO), Volcani center, Israel

Global population growth in the past few decades increase the need in providing food, shelter, and other services, and has resulted in transformation of many natural ecosystems into human-dominated ones. Land-use (LU) dynamics usually accompanied by large increases in exploiting of resources, along with considerable loss of biodiversity, affect ecosystems structure, and function and may cause deterioration of environmental conditions, which is reflected in soil quality (SQ). 

SQ differences among LU can be observed with airborne hyperspectral imaging spectroscopy (IS). Our aim is to measure SQ performances, based solely on spectral differences, and mapping of soil properties among three LU practices (agro-ecosystems, agro-pastoral grazing, and natural reserves) in an arid dryland environment of the Central Negev Desert, Israel. To achieve this goal, we developed and implemented a spectral soil quality index (SSQI) using IS method, which is generated from both laboratory and field spectrometry, for upscaling from point scale to airborne IS at a local scale. 

To characterize and quantify SQ, an integrative approach of 14 physical, biological, and chemical soil properties were examined and transformed into additive scoreless soil quality indices (SQI), which were compared among LU and geographical units (north, centre, and south flight line). 

Significant differences in SQI values were found in part of all LU and geographical areas. Statistical and mathematical methods for evaluation of soil properties significance and spectral differences were used, including partial least squares – regression (PLS-R) and partial least squares – discriminate analysis (PLS-DA). The PLS-DA classification accuracy results of the laboratory spectral data resulted with an overall kappa coefficient accuracy of 0.95. IS application can be used for SQI assessment, address soil alteration and degradation in areas of land-use changes.