Rapid assessment of plant diversity using MODIS biophysical proxies

Swapna Mahanand; Mukunda Dev Behera; Partha Sarathi Roy

doi:10.1016/j.jenvman.2022.114778

Rapid assessment of plant diversity using MODIS biophysical proxies

J Environ Manage. 2022 Mar 3:311:114778. doi: 10.1016/j.jenvman.2022.114778. Online ahead of print.

Authors

Swapna Mahanand¹, Mukunda Dev Behera², Partha Sarathi Roy³

Affiliations

¹ School of Water Resources, Indian Institute of Technology Kharagpur, W.B., 721302, India. Electronic address: mahanand.swapna@gmail.com.
² School of Water Resources, Indian Institute of Technology Kharagpur, W.B., 721302, India; Centre for Oceans, Rivers, Atmosphere and Land Sciences, Indian Institute of Technology Kharagpur, W.B., 721302, India. Electronic address: mdehera@coral.iitkgp.ac.in.
³ World Resources Institute, New Delhi, 110016, India.

PMID: 35248931
DOI: 10.1016/j.jenvman.2022.114778

Abstract

The spectral information derived from satellite data provides important inputs for assessing plant diversity. If a suitable satellite-derived biophysical proxy is applicable to assess and monitor plant diversity of different biogeographic regions will be of interest to policy makers and conservationists. We selected four biogeographic regions of India, i.e., semi-arid, Eastern Ghats, Western Ghats, and Northeast as the test sites on the basis of variations in moisture availability. The flora data collected for the study sites are the extract of the national biodiversity project 'Biodiversity Characterization at Landscape Level'. The available Moderate Resolution Imaging Spectroradiometer (MODIS)-derived biophysical proxies at high temporal frequencies was considered to compare the biophysical proxies: surface reflectance-red and near-infrared, normalized difference vegetation index-NDVI, enhanced vegetation index-EVI, leaf area index-LAI, and fraction of absorbed photosynthetically active radiation-FAPAR at different temporal scales (monthly, post-monsoon, seasonal, annual) in each selected biogeographic regions of India. Generalized linear model (GLM) and multivariate adaptive regression spline (MARS) were utilized to evaluate the relationship between plant diversity and MODIS-derived biophysical proxies. MARS summarized the suitable biophysical proxies at monthly scale in descending order for the total forest area in semi-arid was red, NDVI, and FAPAR; for Eastern Ghats was EVI, FAPAR, and LAI; for Western Ghats was EVI, LAI, and FAPAR; and for Northeast was NDVI, near-infrared, and red. Furthermore, monthly FAPAR commonly found to be the suitable proxy to large scale monitoring of plant diversity in the moisture-varied biogeographic regions of India, except Northeast. Using artificial neural network, the relationship of plant diversity and monthly FAPAR/NDVI were modeled. The correlation between the predicted and reference plant diversity was found to be r = 0.56 for semi-arid, r = 0.52 for Eastern Ghats, r = 0.52 for Western Ghats and r = 0.61 for Northeast at p-value < 0.001. The study affirms that FAPAR is potentially an essential biodiversity variable (EBV) for carrying out rapid/indicative assessment of plant diversity in different biogeographic regions, and thereby, meeting various international commitments dealing with conservation and management measures for biodiversity.

Keywords: Artificial neural network; Biogeographic regions; Fraction of absorbed photosynthetically active radiation; Generalized linear model; Leaf area index; Multivariate adaptive regression spline.