Mapping Productivity and Essential Biophysical Parameters of Cultivated Tropical Grasslands from Sentinel-2 Imagery

Amparo Cisneros,Patricia Menezes,Peterson Fiorio,Nieves Pasqualotto,Shari Van Wittenberghe,Gustavo Bayma,Sandra Furlan Nogueira

doi:10.3390/agronomy10050711

Abstract

Nitrogen (N) is the main nutrient element that maintains productivity in forages; it is inextricably linked to dry matter increase and plant support capacity. In recent years, high spectral and spatial resolution remote sensors, e.g., the European Space Agency (ESA)’s Sentinel satellite missions, have become freely available for agricultural science, and have proven to be powerful monitoring tools. The use of vegetation indices has been essential for crop monitoring and biomass estimation models. The objective of this work is to test and demonstrate the applicability of different vegetation indices to estimate the biomass productivity, the foliar nitrogen content (FNC), the plant height and the leaf area index (LAI) of several tropical grasslands species submitted to different nitrogen (N) rates in an experimental area of São Paulo, Brazil. Field reflectance data of Panicum maximum and Urochloa brizantha species’ cultivars were taken and convoluted to the Sentinel-2 satellite bands. Subsequently, different vegetation indices (Normalized Difference Vegetation Index (NDI), Three Band Index (TBI), Difference light Height (DLH), Three Band Dall’Olmo (DO), and Normalized Area Over reflectance Curve (NAOC)) were tested for the experimental grassland areas, and composed of Urochloa decumbens and Urochloa brizantha grass species, which were sampled and destructively analyzed. Our results show the use of different relevant Sentinel-2 bands in the visible (VIS)–near infrared (NIR) regions for the estimation of the different biophysical parameters. The FNC obtained the best correlation for the TBI index combining blue, green and red bands with a determination coefficient (R2) of 0.38 and Root Mean Square Error (RMSE) of 3.4 g kg−1. The estimation of grassland productivity based on red-edge and NIR bands showed a R2 = 0.54 and a RMSE = 1800 kg ha−1. For the LAI, the best index was the NAOC (R2 = 0.57 and RMSE = 1.4 m2 m−2). High values of FNC, productivity and LAI based on different sets of Sentinel-2 bands were consistently obtained for areas under N fertilization.

Highlights

The increase in food demand and land use pressure has driven changes in Brazilian cattle production
The objective of this study is to test different vegetation indices, using hyperspectral field data resampled to the Sentinel-2 satellite spectral configuration, to estimate essential biophysical parameters of tropical cultivated grasslands subject to different doses of N in the São Paulo region
In situ measured parameters combined with field spectroscopy allowed us to establish the best vegetation index applicable to the Sentinel-2 band settings for essential biophysical parameters retrieval for grassland species

Summary

Introduction

The increase in food demand and land use pressure has driven changes in Brazilian cattle production. In the period from 1950 to 2006, beef production increased about six times, making the country one of the world’s largest cattle producers [1]. Spatial mapping analysis of Brazilian pastures between 1985 (118 million ha) and 2017 (178 million ha) indicated that the greatest expansion of livestock occurred between 1985 and 2002. In this period an increase in Brazilian pasture areas of 57 million ha occurred, mostly in the northern region of the country and, to a lesser extent, in the Midwest. From the 2000s, adoption of new technologies by the producers led to an increase in productivity, which reflected in increases in the animal stocking rate of pastures and in the individual weight gain of the cattle [1]

Objectives

Methods

Results

Discussion

Conclusion