Environ Monit Assess (2010) 170:231–244
DOI 10.1007/s10661-009-1228-7
Comparison of different semi-empirical algorithms
to estimate chlorophyll-a concentration in inland
lake water
Hongtao Duan · Ronghua Ma · Jingping Xu ·
Yuanzhi Zhang · Bai Zhang
Received: 4 January 2009 / Accepted: 29 October 2009 / Published online: 12 November 2009
© Springer Science + Business Media B.V. 2009
Abstract Based on in situ water sampling and field
spectral measurement from June to September
2004 in Lake Chagan, a comparison of several
existing semi-empirical algorithms to determine
chlorophyll-a (Chl-a) content was made by apply-
ing them to the field spectra and in situ chloro-
phyll measurements. Results indicated that the
first derivative of reflectance was well correlated
with Chl-a. The highest correlation between the
first derivative and Chl-a was at 680 nm. The two-
H. Duan (B ) · R. Ma
State Key Laboratory of Lake Science
and Environment, Nanjing Institute of Geography
and Limnology, Chinese Academy of Sciences,
Nanjing 210008, China
e-mail: htduan@niglas.ac.cn, htduan@gmail.com
J. Xu
State Key Laboratory of Remote Sensing Science,
Jointly Sponsored by the Institute of Remote Sensing
Applications of Chinese Academy of Sciences
and Beijing Normal University,
Beijing 100101, China
Y. Zhang
Institute of Space and Earth Information Science,
The Chinese University of Hong Kong,
Shatin, NT, Hong Kong
B. Zhang
Northeast Institute of Geography and Agricultural
Ecology, Chinese Academy of Sciences,
Changchun 130012, China
band model, NIR/red ratio of R
710/670
, was also an
effective predictor of Chl-a concentration. Since
the two-band ratios model is a special case of
the three-band model developed recently, three-
band model in Lake Chagan showed a higher
resolution. The new algorithm named reverse con-
tinuum removal relies on the reflectance peak at
700 nm whose shape and position depend strongly
upon chlorophyll concentration: The depth and
area of the peak above a baseline showed a linear
relationship to Chl-a concentration. All of the
algorithms mentioned proved to be of value and
can be used to predict Chl-a concentration. Best
results were obtained by using the algorithms
of the first derivative, which yielded R
2
around
0.74 and RMSE around 6.39 μg/l. The two-band
and three-band algorithms were further applied
to MERIS when filed spectral were resampled
with regard to their center wavelengths. Both
algorithms showed an adequate precision, and
the differences on the outcome were small with
R
2
= 0.70 and 0.71.
Keywords Field spectral · Lake Chagan ·
Continuum removal · Three-band model
Introduction
Lakes are valuable water resources and can be
used for fishing, transport, agriculture, industry,