@Article{AraújoGalvDala:2023:AnUsPR,
author = "Ara{\'u}jo, Julibana de Abreu and Galv{\~a}o, L{\^e}nio Soares
and Dalagnol, Ricardo",
affiliation = "{Instituto Nacional de Pesquisas Espaciais (INPE)} and {Instituto
Nacional de Pesquisas Espaciais (INPE)} and {Jet Propulsion
Laboratory}",
title = "Sensitivity of hyperspectral vegetation indices to rainfall
seasonality in the Brazilian savannahs: an analysis using PRISMA
data",
journal = "Remote Sensing Letters",
year = "2023",
volume = "14",
number = "3",
pages = "277--287",
month = "mar.",
abstract = "We evaluated the sensitivity of 14 narrowband vegetation indices
(VIs) to rainfall seasonality over the Brazilian savannahs. Five
images obtained in 2020 by the PRecursore IperSpettrale della
Missione Applicativa (PRISMA) tracked the transition from the
rainy (11 May) to the dry season (8 July, 17 August, and 4
September), and towards the beginning of a new seasonal cycle on 3
October. We considered two scenarios in the data analysis. First,
we kept the PRISMA image from 11 May as a reference to evaluate
the VI sensitivity with increasing water deficit from 8 July (49
days without any precipitation) to 4 September (102 days). Second,
we changed the reference image to 4 September to evaluate the
largest VI responses on 3 October after the first rainfall. The
first three VIs (ranked by F-values) having significant changes
with increasing water deficit over grasslands were the Normalized
Difference Vegetation Index (NDVI), Enhanced Vegetation Index
(EVI), and Moisture Stress Index (MSI). The Vogelmann red edge
index (VOG) and Red-Edge Normalized Difference Vegetation Index
(RENDVI) presented the largest F-values over woodlands. In the
second scenario, Red-edge Vegetation Stress Index (RVSI) and
RENDVI were among the first five ranked VIs by t-values over both
areas. The largest changes in VIs were generally observed over
savannah grassland, which is the most sensitive physiognomy to
water deficit. The lowest modifications were noted over riparian
forests, which have access to waters from rivers. The
vegetation-type dependence of the VI changes was also observed
after the occurrence of rainfall. Results suggest the potential
use of different VIs to obtain phenological metrics for more
accurate savannah mapping in Brazil.",
doi = "10.1080/2150704X.2023.2189031",
url = "http://dx.doi.org/10.1080/2150704X.2023.2189031",
issn = "2150-704X",
language = "en",
targetfile = "Sensitivity of hyperspectral vegetation indices to rainfall
seasonality in the Brazilian savannahs an analysis using PRISMA
data.pdf",
urlaccessdate = "20 set. 2024"
}