Review Jurnal Penginderaan Jauh Sinar Tampak "Evaluation of Standar & Regional Satellite Chlorophyll-a Algorithms for MODIS in the Bohai & Yellow Seas, China" oleh Wang et al. (2019)
4. Introduction
Chlorophyll-a (chl-a) concentration
has long been regarded as a
powerful way of representing
phytoplankton biomass in the oce
an.
In recent decades, remotely sense
d
satellite ocean colour products ha
ve been widely used to detect syn
optic
spatial and temporal patterns and
variability of sea surface chl-a
concentrations.
However, when applied to coastal
waters which are shallow in depth
and/or have high turbidity, satellite
-derived chl-a may produce
significant biases due to inaccurac
ies in atmospheric correction mod
els
and improper inversion algorithms
.
Therefore, careful evaluation of th
e
existing satellite chl-a algorithms a
nd developing new regional algorit
hms are essential for satellite oce
an colour applications over opticall
y complex waters.
5. To evaluate the performance of the standar
d OC3M (ocean chl-a three-band algorithm f
or MODIS (moderate-resolution imaging sp
ectro
radiometer)) chl-a algorithm.
The aim
6. Data & Methods
In situ measurements of
chl-a
Figure 1.
(a) Map of the BYS and (b)–(j) the sites of the in situ chl-a measurem
ents during each cruise;
(b) 20 April–4 May 2010; (c) 2 May– 20 May 2012;
(d) 2 November–19 November 2012; (e) 22 June–9 July 2013;
(f) 6 November–24 November 2013; (g) 28 April–18 May 2014;
(h) 7 November–23 November 2014; (i) 17 August-5 September 2015
;
(j) 15 January–30 January 2016.
7. The daily MODIS/Aqua standard local area
coverage (LAC) remote-sensing reflectance (
Rrs)
images for 10 available visible bands were
downloaded from the NASA
(National Aeronautics and Space
Administration)
Satellite chl-a data
spatial resolution of approximately 1 km ×1 km
and over the time span from
4 July 2010 to 31 December 2016.
8. 1 2
3 4
Method for satellite chl-a data validation
The satellite chl-a values were extract
ed and compared against the coincide
nt in
Situ chl-a measurements.
Visual inspections of spatial difference
s
between in situ, OC3M and GAM
(correlation analyses).
The BYS was subdivided into three
subregions Including:
a coastal region (<20 m isobath);
an offshore region (>50 m isobath);
as well as a transitional region (betwe
en 20 and 50 m isobaths).
The chl-a series for the specific regi
ons were calculated and compared
with
the corresponding in situ chl-a serie
s
over each time span during the year
s 2010–2016.
9. Statistical comparisons
Results
2(a). The MRD value of 1.
13 mg m−3 indicates a dist
inct
overestimation in OC3M c
hl-a estimates.
2(b). In contrast, the GAM
chl-a algorithm was
characterized by much low
er uncertainty (MAPD = 33
.56%, MRD = 0.09 mg m−
3).
10. Spatial comparisons for each cruise
Figure 3. Spatial distribution
s
of:
(a) in situ chl-a measuremen
ts;
(b) OC3M satellite chl-a;
(c) GAM satellite chl-a;
composites for each of the 9
cruises (from left to right in
order: January 2016, April–
May 2010, April–May 2014,
May 2012, June–July 2013,
August–September 2015,
November 2012, November
2014 and November 2013);
(d) the series of chl-a averag
es during the period of each
cruise.
11. Time series comparisons
Figure 4. The time series of monthly chl-a
averages generated from satellite products for
:
(a) the entire BYS region,
(b) coastal region,
(c) transition region, and
(d) deep region.
The solid line is the OC3M chl-a, the dotted lin
e is GAM chl-a, and filled circle is in situ chl-a
averages for each cruise.
Table 1. Results of the Spearman rank correlation
analyses between satellite-derived and in situ
measured chl-a over different regions.
12. Conclusio
n
The extent of overestimati
on of chl-a values by the
standard OC3M algorithm
values was quantified, and
the extent of distortion of c
hl-a seasonality was show
n to be significant.
The result of this study
opposes the use of the
standard OC3M algorithm i
n the BYS.
Compared with the
OC3M algorithm, a
previously reported
regional GAM chl-a
algorithm for the BYS
produced more
accurate derived chl-a
values and a
seasonality consistent
with the in situ
measurements, makin
g it a more suitable ch
l-a algorithm for use in
the BYS.