Snow leopard habitat modeling using neuro fuzzy technique & a comparative analysis with traditional overlay methods: a case study Chamoli and Pithoragarh district, Uttaranchal
Student name: Ms Senjuti Sen
Guide: Dr Chander Kumar Singh
Year of completion: 2013
Host Organisation: Indira Gandhi Conservation Monitoring Centre, WWF-India, New Delhi
Supervisor (Host Organisation): Dr G. Areendran
Abstract: Snow leopard (uncia uncia) is the elusive, mysterious predator of the Himalayan region,
found in the 5 Himalayan states of Jammu and Kashmir, Himachal Pradesh, Uttarakhand,
Sikkim, and Arunachal Pradesh. This predator is a medium sized cat, with thick fur, stocky
limbs and long tail which helps it to survive in the harsh climatic condition of the mountain
regions. According to IUCN Reports, the coarse global population estimate for the snow
leopard is c. 7,400 individuals, and the species is classified as Endangered in the IUCN’s Red
List. India , has 10% of the snow leopard population in a less than 5% of its habitat area.
Though the habitat has natural barriers, yet the animal faces serious danger. Moreover, the
animal is not restricted to wildlife sanctuaries only, but is found spread across the whole area.
In the following work, habitat suitability modelling for snow leopards has been undertaken to
accommodate this behaviour of the cats in the state of Uttaranchal. Landsat images of ETM+
as well as TM is being used to extract the land use land cover of the study area .ASTER DEM is
being used to find the slope , relief, aspect and ruggedness index of the area. Prey citation data
along with Snow leopard citation data further completes the database that is required for
snow leopard habitat modelling. Finally a neuro-fuzzy technique has been applied using
MATLAB ANFIS editor to determine the suitable habitat zones for the snow leopard in the
Himalayan state. The surface generated has been divided into six suitability zones depending
upon the combination of various input parameters. The model has been finally validated with
ground observation data, and has been compared with traditional weighted overlay method. It
has been seen that while neuro-fuzzy has successfully defined various suitability zones,
weighted overlay technique has rather given a very generalized idea about suitability of the
area for snow leopard habitat. Though this technique is being used for studies including
landslide susceptibility, pollution vulnerability, implementing it in habitat modelling is
comparatively new and since it takes into consideration of the fact that everything is not
Boolean, it has the ability to provide a more realistic approach towards habitat modelling.
Thus geo-spatial techniques used here will give a cost effective solution towards conservation
of these elusive yet endangered cats.
KEYWORDS: Habitat suitability, neuro fuzzy, ANFIS, Ruggedness Index, R, MATLAB, slope,
aspect, membership functions, weighted overlay.
