Remote sensing based assessment of vegetation composition, phenology and fires in Eravikulam National Park, Kerala

A forest ecosystem is the basic and most important ecological unit. The present study aimed to
provide the spatial extent of the vegetation types, vegetation burnt area and baseline information
on phenological metrics in Eravikulam national park, Kerala, India. To classify the land cover
types of Eravikulam National Park, LISS 1V satellite imagery is used. Using object based image
analysis, nine land cover classes are mapped. Sentinel 2 based NDVI datasets were downloaded
using Google Earth Engine Code Editor as the input. The spatio-temporal analysis was made
using NDVI values from Sentinel-2 data of 2018. The key phenological matrices are obtained
using CropPhenology package by giving the multi-temporal NDVI images. In this study 15
phenological matrices are obtained which are used for further phenology analysis. It is found
that except for the forest and the plantation, all other classes are distinguished on the basis of
NDVI thresholds. But forest and plantation could not be classified separately since both the
classes show high NDVI values due to the high canopy density throughout. IRS LISS III and
Landsat 8 OLI data were used for burnt area assessment. Vegetation burnt areas are extracted
using digital classification method for the years 2017, 2018 and 2019. Overall, grasslands area
affected by fires quantified as 649 ha, 636 ha and 881 ha in 2017, 2018 and 2019 respectively.
Of the three years, 1723 ha of grassland area is under occasionally fire affected area, 381 ha is
under frequently fire affected and 62 ha is under fire recurrence. Fire hotspots were studied
using spatial statistics tool (optimised hotspot analysis). Spatial analysis indicates 1847 ha of
dense grasslands and 416 ha of open grasslands are included under fire hotspots. Shola forests
are included under coldspots due to no fires in the study period.

MPhil EI 2018-2019 INT Dr. R Jaishankar