Komaristaya V., Bezrodnova O., Rudas A. A remote sensing approach to assess algal beta-carotene content in solar salt evaporation ponds / Proceedings of the conference "GIS-Forum-2018" (Kharkiv, 14-16 March 2018). - Issue 2. - V.N. Karazin KhNU, 2018. - P. 42-45.

Abstract

The present research attempts to calibrate satellite images against field data on D. salina
ß-carotene concentration expressed per 1 L of brine. We found the empirical relation between ß-carotene quantity and the index based on two infrared reflectance bands (NIR and SWIR).

Full text

Збірник матеріалів конференції


Victoria Komaristaya1, Olga Bezrodnova1, Alexander Rudas2
1
2



ß

Dunaliella salina, ß-carotene, Landsat 5TM, Landsat 7ETM+, reectance,
near infra-red (NIR), short wave infra-red (SWIR).
Red “bloom” induced by planktonic microalga Dunaliella salina Teod.
is typical of solar salt evaporation ponds. The alga is able to accumulate
ß-carotene, coloring its cells orange-red and, due to its massive development,
shading the pond brine. The algal biomass enriched with ß-carotene could be
the additional high value by-product of solar salt manufacture. The annual
world market of algal ß-carotene as a food color and a potent antioxidant in
food supplements totals about US$200 million, the demand constantly
growing. The market is dominated by Australia based division of Henkel/
Cognis (currently owned by BASF), where the alga is grown in specialized open
ponds [1]. Despite the facts that Ukraine possesses quite large suitable solar
salt works area, and the original idea is authored by Ukrainian researchers
[2], algal ß-carotene still is not manufactured in Ukraine industrially. As
ß-carotene accumulation in pond brine can be visually detected by brightness
of orange-red hue, we suppose that the satellite imagery could be used to
assess the natural resources of algal ß-carotene in Ukraine and compare to
other production sites of the world.
Satellite remote sensing is a standard instrument to monitor and assess
algal blooms, mostly applied to harmful blooms in the open ocean [3].
Little work was published on satellite remote sensing of D. salina “bloom”
[4–6], revealing a specic problem: algal “bloom” is often masked by pink
halobacterial “bloom”, though, it is possible to differentiate them by spectral
signatures [5]. Techniques to quantitatively assess D. salina ß-carotene
concentrations in open ponds by satellite imagery were not proposed in the
literature.
In 2006–2008 we monitored D. salina “bloom” in the ponds of Heroyske

ГІС-ФОРУМ-2018

salt works (46°29’19.84»N, 31°54’15.96»E), Kherson region [7]. The present
research attempts to calibrate satellite images against eld data on D. salina
ß-carotene concentration expressed per 1 L of brine.
Field sampling and sample analysis were described in [7]. Spectral images,
dated ± 6 days around sampling, taken by satellite instruments of
moderate resolution (15 and 30 m), were retrieved from USGS web site via
EarthExplorer interface (data available from the U. S. Geological Survey).
Scenes with cloud cover over the area of interest were discarded, resulting in
8 Landsat 5TM and 13 Landsat 7ETM+ images (courtesy of the U. S. Geological
Survey). Quantum GIS 2.18.15 [8] and Semi-Automatic Classication Plugin
(SCP) 5.3.11 [9] were used. DN values were automatically converted into
reectance, Landsat 7ETM+ images pansharpened, DOS1 atmospheric
correction applied to all the images. Google Satellite image (© 2018,
DigitalGlobe) was used to delineate salt work ponds by vector polygons.
Several types of spectral indices (single band reectance, band ratios, band
differences, and normalized band differences) were calculated pixel-wise and
their mean values for each pond of interest extracted with QGIS Zonal Statistic
Tool. Pearson correlations between mean indices and eld measured variables
were analyzed using R [10] and RStudio [11].
The strongest statistically signicant positive linear correlation was found
between algal ß-carotene concentration in the brine and the normalized
difference of NIR and SWIR (1.55–1.75 nm) reectance, especially for ponds
8 and 11 (g. 1), which were excluded from salt work operation during the
study and managed to stimulate the algal “bloom” [7].
Notably, the correlation coefcient increased as sampling and snapshot
Fig. 1. Pearson
correlation between
brine concentration
of ß-carotene (mg/L)
and normalized
difference NIR-SWIR
index in ponds 8
and 11, depended
on time passed
between sampling
and snapshot (linear
t drawn for 3 days
interval)

Збірник матеріалів конференції

References:
1. Beta-Carotene Market Analysis By Source (Algae, Fruits & Vegetables, &
Synthetic), By Application (Food & Beverages, Dietary Supplements, Cosmetics, &
Animal Feed) And Segment Forecasts To 2024. GV15198380. — Grand View Research,
2016. — 78 p.
2. Дрокова І. Г. Водорість Dunaliella salina Teod. як джерело одержання ß-ка-
ротину/І. Г. Дрокова//Укр. ботан. журн. — 1961. — Т. 18, № 4. — С. 110–112.
3. Stumpf R. P. Remote sensing of harmful algal blooms/R. P. Stumpf,
M. C. Tomlinson//Remote sensing of coastal aquatic environments. — Springer,
Dordrecht, 2007. — P. 277–296.
4. Oren A. Development and spatial distribution of an algal bloom in the Dead Sea:
a remote sensing study/A. Oren, N. Ben-Yosef//Aquatic microbial ecology. — 1997. —
V. 13, № 2. — P. 219–223.
5. Richardson L. L. Remote sensing of algal bloom
dynamics/ L. L. Richardson// BioScience. — 1996. — V. 46, № 7. — P. 492–501.
dates converge up to 3 days (Table 1). Further decrease of correlation coefcient
and p-value increase at 0–2 days interval were probably due to insufcient
amount of data remained.
Pearson correlation between brine concentration of ß-carotene
mg/L and normalized difference NIR-SWIR index depended
on time passed between sampling and snapshot

We found the empirical relation between ß-carotene quantity and the index
based on two infrared reectance bands (NIR and SWIR). Theoretical grounds
of that remain unclear, although near infra-red reectance spectroscopy is
already in use to evaluate in vivo carotenoid content in plant fruits [12]. Some
more algorithms (e.g. three-band indices, spectral angle etc.) could be tested
to nd better t between eld data and satellite imagery of D. salina ponds.
Time, days Pearson correlation
coefcient
p-value
0 
1 
2 
3 
  
  
  

ГІС-ФОРУМ-2018

6. Vijay R. A multi-temporal analysis for change assessment and estimation of
algal bloom in Sambhar Lake, Rajasthan, India/Vijay R., Pinto S. M., Kushwaha V. K.
[et al.]//Environmental monitoring and assessment. — 2016. — V. 188, № 9. — P. 510.
7. Komaristaya V. P. Ecological peculiarities of natural populations of hyperhalobe
microalga Dunaliella salina Teod. in solar salt work ponds of the South of Ukraine
and Russia/V. P. Komaristaya, A. N. Rudas, N. M. Tatischeva [et al.]//The Journal
of V. N. Karazin Kharkiv National University. Series: Biology — 2014. — V. 20,
№ 1100. — P. 315–323.
8. QGIS Development Team (2018). QGIS Geographic Information System. Open
Source Geospatial Foundation Project. http://qgis.org.
9. Congedo L. Semi-Automatic Classication Plugin Documentation. — 2016. — 266 p.
10. R Core Team (2017). R: A language and environment for statistical computing. R
Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
11. RStudio Team (2016). RStudio: Integrated Development for R. RStudio, Inc.,
Boston, MA URL http://www.rstudio.com/.
12. Brenna O. V. Application of near-infrared reectance spectroscopy (NIRS) to
the evaluation of carotenoids content in maize/O. V. Brenna, N. Berardo//Journal of
Agricultural and Food Chemistry. — 2004. — V. 52, № 18. — P. 5577–5582.