Mapping alien invasive species, P. juliflora using hyperspectral remote sensing: a case study of Delhi
Student name: Ms Swati Rawat
Guide: Dr Chander Kumar Singh
Year of completion: 2013
Host Organisation: DTRI, Defence Research and Development Organisation, New Delhi
Supervisor (Host Organisation): Dr L.K. Sinha
Abstract: The rapid spread of alien invasive species is posing a serious threat to biodiversity and ecological integrity of
the native habitats. Prosopis juliflora is one of the very aggressive invader that has invaded the
natural habitats of many native species in forests and ridges of Delhi. Many of the draining channels
in Delhi are clogged with P. juliflora which obstruct the water flow and resulting low water levels in
the water table. Also, it does not allow any other plant species to grow under its canopy and results
in monoculture in wastelands. In such scenario remote sensing is found to be a useful tool for
mapping and monitoring the spread of P. juliflora.
In this study, an attempt has been made to map the spatial distribution of Prosopis juliflora across
Delhi and National Capital Region using Hyperion data. The Hyperion image used is carefully
processed by applying various image processing techniques such as radiometric correction,
atmospheric correction and MNF transformation. SMACC endmember extraction technique is used
to find vegetation endmembers in the image where vegetation pixels were masked on the basis of
NDVI. Endmember with the closest match to ASD spectra of Prosopis was used for mapping. Apart
from the ASD spectra, image spectra of Prosopis were also collected and used for mapping. SAM
Classification technique was used for mapping in both the cases. The overall accuracy in first case,
wherein ASD spectra were used, is 71.43% and Kappa coefficient is 0.56. In case of mapping with
image spectra, the overall accuracy is 89.28% and kappa coefficient is 0.70.
Keywords: Invasive species, Prosopis, Hyperspectral data, Spectral smile, ASD spectra.