Temporal Validation of Four LAI Products over
Grasslands in the Northeastern Tibetan Plateau
Gaofei Yin, Ainong Li, Zhengjian Zhang, and Guangbin Lei
Abstract
Time series of leaf area index (
LAI
) products are now widely
used, and the temporal validation is the prerequisite for
their proper application. However, a systematical compari-
son between different products using both direct and indi-
rect methods is still lacking. The objective of this paper is
to assess and compare the temporal performances of four
LAI
products: Moderate Resolution Imaging Spectroradi-
ometer
(MODIS)
LAI (MOD)15A2
,
MOD15A2h
, Geoland2 Version
1
(GEOV1)
, and Global Land Surface Satellite (
GLASS
). The
study area, which is dominated by grasslands, is located in
the northeastern Tibetan Plateau (
TP
), and temperature is
the main stress factor affecting grass growth. Both a correla-
tion analysis with temperature and a direct comparison with
temporally continuous
LAI
reference maps were implemented
in our temporal validation experiments. The results show
that no single product can capture the rapid change and the
seasonal trend in
LAI
simultaneously, and the compositing
period used in each product determines the quality of the
corresponding
LAI
time series. The
MOD15A2
and
MOD15A2h
products, which have short compositing windows (eight
days), are suitable for detecting rapid change. A grazing-
induced biomass decrease that occurred around day of year
205 in 2014 in our study area was clearly revealed in these
two products. For the
GEOV1
and
GLASS
products, which have
compositing windows of 30 days and 1 year, respectively, the
grazing date was shifted (
GEOV1
) or even invisible (
GLASS
).
However, products with prolonged compositing windows
may be more robust to observation noise, and the resulting
products may be suitable for capturing the seasonal trend.
This study highlights that the concurrent use of data from
various sensors onboard different satelli
tion of new generations of satellites (e.g.,
promising ways to further improve existi
Introduction
The leaf area index (
LAI
), which is a key vegetation structure
and function parameter, describes the surface area available
for energy and mass exchanges between vegetation and the
atmosphere. Currently, several coarse-resolution (~1 km)
LAI
products are routinely produced, including the Moderate
Resolution Imaging Spectroradiometer (
MODIS
) (Myneni
et
al.
2002), Geoland2 Version 1 (
GEOV1
) (Baret
et al.
2013), and
Global Land Surface Satellite (
GLASS
) (Xiao
et al.
2014) prod-
ucts. These
LAI
products are widely used as proxy indicators
of vegetation status to monitor ecosystem variations in re-
sponse to climate change and anthropogenic activities (Jiapaer
et al.
2015). Although the
LAI
products have been extensively
validated, spanning a wide range in space (Camacho
et al.
2013; Fang, Wei, and Liang 2012; Garrigues
et al.
2008), their
performance in time series is still not clear.
To assess their performance in time series, temporal vali-
dations (
TVs
) of
LAI
products are needed (Fang
et al.
2019b;
Xie
et al.
2019). Current
TV
studies perform two categories of
analyses, specifically indirect
TVs
and direct
TVs
. An indirect
TV
compares the relationship between
LAI
trends and key
meteorological variables in areas where these variables limit
plant growth (Yan
et al.
2016b). On the other hand, a direct
TV
uses multiple
LAI
reference maps within a single year, gener-
ally derived from up-scaling the
in situ
LAI
measurements, to
assess the temporal dynamics of uncertainty in
LAI
products
(Fang
et al.
2019a; Yin
et al.
2017a). Generating temporally
continuous
LAI
reference maps is the key issue in performing
direct
TVs
(Yin
et al.
2017a). To the best of our knowledge,
there are very few studies, if they exist, to implement both di-
rect and indirect
TVs
simultaneously to systematically assess
the temporal performance of
LAI
products.
The objective of this paper is to fill the gap existing in
the
TV
of
LAI
products. To fulfill the research objective, both
the direct and indirect
TV
methods were used for four com-
monly used
LAI
products, i.e.,
MOD15A2
(Myneni
et al.
2002),
MOD15A2h
(Yan
et al.
2016a; Yan
et al.
2016b),
GEOV1
(Baret
et
al.
2013; Camacho
et al.
2013), and
GLASS
(Xiao
et al.
2014).
The remainder of the paper is structured as follows. In the
next section, we introduce the data and method used in this
study. The “Results” section presents both the direct and
indirect
TV
results; a discussion is presented in the section
“Discussion”, and conclusions are given in the last section.
Collection
The research was conducted in a 5 km × 5 km region centered
on ~33
°
55
′
N, 102
°
51
′
E that is close to Huahu Lake in the
Zoige National Nature Reserve (Figure 1a). It is located at the
northeastern edge of the Tibetan Plateau (
TP
). The study area
is mainly covered by alpine meadows, which have substantial
ecological vulnerability (Li
et al.
2012). In recent years, the
degradation of grasslands has become more and more serious,
which is directly due to the influence of human activities, such
as excessive grazing (Roy
et al.
2014). The climate is cold and
wet, and the temperatures, which range from
-
15°C to 15°C
(as measured at a nearby meteorological station; see Figure 2),
are the main limited factor for the temporal variations in the
grasslands in our study area (Hansen
et al.
2013). The spatial
variations of grass growth are mainly controlled by altitude,
and it ranges from 3400 m to 3500 m (see Figure 1b) according
to the
ASTER GDEM
V3 dataset (
METI
and
NASA
2019).
Gaofei Yin* is with the Faculty of Geosciences and
Environmental Engineering, Southwest Jiaotong University,
Chengdu 610031, China (
).
Ainong Li*, Zhengjian Zhang, and Guangbin Lei are with the
Research Center for Digital Mountain and Remote Sensing
Application, Institute of Mountain Hazards and Environment,
Chinese Academy Sciences, Chengdu 610010, China.
(
)
*Corresponding authors.
Photogrammetric Engineering & Remote Sensing
Vol. 86, No. 4, April 2020, pp. 225–233.
0099-1112/20/225–233
© 2020 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.86.4.225
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
April 2020
225
