Application of an improved vegetation index from the visible spectrum in
the diagnosis of degraded pastures: Implications for development
Abstract
Inadequate pasture management causes land degradation and negative
impacts on the socio-economic development of agricultural regions. Given
the importance for Brazil and the World of pasture-based livestock
production, the recognition of pasture degradation is essential. The use
of remote sensing satellite systems to detect degraded pastures
increased in the recent past, because of their capability to survey
large portions of Earth’s surface. A struggle nowadays is to improve
detection accuracy and to implement high-resolution surveys at farmland
scale using unmanned aerial vehicles (UAVs). The satellite sensors
capture reflectance from the visible spectrum and near infrared bands,
which allows estimating plant’s vigor vegetation indices. The NDVI is a
widely accepted index, but to generate an NDVI map using a UAV a
relatively high-cost multispectral sensor is required, while most UAVs
are equipped with low-cost RGB cameras. In the present study, a script
developed on the Google Earth Engine image-processing platform
manipulated images from the Landsat 8 satellite, and compared the
performances of NDVI and an improved color index that we coined “Total
Brightness Quotient” of red (TBQR), green (TBQG) and blue (TBQB) bands.
An efficient detection of pasture degradation using the TBQs would be a
good prognosis for the surveys at farm scale where environmental
authorities are progressively using UAVs and forcing landowners towards
pasture restoration. When compared to NDVI, the TBQG showed a
correlation of 0.965 and an accuracy of 88.63%. Thus, the TBQG proved
as efficient as the NDVI in the diagnosis of degraded pastures.