Remote Sensing Applications: Society and Environment 33 (2024) 101094

Contents lists available at ScienceDirect

Remote Sensing Applications: Society and

Environment
journal homepage: www.elsevier.com/locate/rsase

Monitoring of reservoir CDOM degradation process by remote

sensing of its surface waters signal
Xavier Sòria-Perpinyà a, *, Eduardo Vicente b, Juan Miguel Soria b
a
Image Processing Laboratory, Universitat de València, Paterna, Spain
b
Cavanilles Institute of Biodiversity and Evolutionary Biology (ICBiBE), Universitat de València, Paterna, Spain

A R T I C L E I N F O A B S T R A C T

Keywords: Many remote sensing studies have been conducted to estimate the chromophoric dissolved
Remote sensing organic matter (CDOM) concentrations of inland waters, but never have yet been conducted the
Landsat-8 CDOM degradation rates analyse its yearly variations. This, limits our understanding of the
Reservoirs organic matter process that occurs in inland waters and thus the carbon cycle. With the aim of
CDOM improving the knowledge of CDOM degradation rates and times, a total of 45 field samples were
Degradation
used to conduct this study. Twenty-three samples were used for the calibration process and other
Photobleaching
twenty-two for validation, obtaining a R2 of 0.87 and an RRMSE of 13%. Finally, all samples were
Annual variation
used to obtain a more robust algorithm. The CDOM degradation rates calculated for reservoirs in
the summer period, with a hydraulic residence time of 30–40 days, were in agreement with the
rates obtained in laboratory experiments, whereas with higher residence times other autoch­
thonous or allochthonous factors began to have influence. Annual variations in the CDOM
degradation rate exhibited the maximum value in spring, simply because more degradable CDOM
was present in springtime water, and the minimum values in autumn and winter, with less light
degradation and the increased entry of more recalcitrant allochthonous CDOM, a result coin­
ciding with those of other studies using laboratory experiments.

1. Introduction
Chromophoric dissolved organic matter (CDOM) is an optically active component of dissolved organic matter and it mainly
originates from degradation of plant materials in terrestrial and aquatic ecosystems (Kirk, 2010), with humic substances being its most
important component (Vähätalo and Wetzel, 2004). Chromophoric is an optical characteristic, defined as that portion of DOM (Dis­
solved organic matter) that is able to absorb solar radiation. When the molecules of many humic compounds of DOM absorb solar
radiation >290 nm, they induce photochemical reactions that can modify the compounds to render them more bioavailable to bacteria
and, in some cases, degrade them entirely to CO2 (Wetzel, 2001).
For this reason, CDOM substances play an essential role in a variety of important processes in surface waters (Vodacek et al., 1997),
such as absorbing light strongly in the UV range and limiting the penetration of biologically damaging UVB radiation into surface
waters and thus acting to protect phytoplankton and other biota (Blough and Green 1995; Blough and Zepp 1990; Ayoub et al., 2012).
Furthermore, CDOM absorbs light according to the energy of the electronic states of the molecule, as well as to its intermediate
vibrational or rotational levels (Ferrari et al., 1996), and at higher levels, this can extend well into the visible wavelength regime and

* Corresponding author.
E-mail address: [email protected] (X. Sòria-Perpinyà).

https://doi.org/10.1016/j.rsase.2023.101094
Received 30 January 2023; Received in revised form 27 October 2023; Accepted 7 November 2023
Available online 11 November 2023
2352-9385/© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

can reduce the amount and quality of the photosynthetically active radiation available to phytoplankton, thus decreasing primary
productivity and affecting ecosystem structure (Bidigare et al., 1993). Therefore, CDOM absorption (degradation) is related to its
sources.
Regarding its origins, CDOM can enter into freshwater from two types of sources—allochthonous sources, from the decomposition
of woody plants in terrestrial environments, and autochthonous sources, from the decomposition of algae and aquatic vegetation
within a water body (Brezonik et al., 2015). Allochthonous material usually exhibits a high lignin content and may be more recalcitrant
than autochthonous components (Devesa-Rey and Barral, 2012). The predominance of one or other is related with both the basin
surface, which determines the higher or lower input of allochthonous material, and the presence of phytoplanktonic algae and
macrophytes.
CDOM degradation is related to microbial metabolism, solar radiation (e.g., in the experiments of Vähätalo and Wetzel (2004),
70-day exposure to solar radiation decomposed 96% of freshwater CDOM) and hydraulic residence time (HRT). However, sunlight,
and particularly the exposure of CDOM to UV light, is the main factor in the degradation and removal of chromophoric compounds
(Cory et al., 2014).
Due to its light absorption characteristics, CDOM influences the water’s reflectance spectrum and can therefore be determined
remotely. Furthermore, the acquisition of surface water information through remote sensing techniques is possible by means of two
methods: empirical and analytical methods. The empirical approach generally produces robust results for the areas and datasets from
which they are derived, and has a demonstrable capability to provide timely and accurate information for a variety of parameters in
lakes and estuaries that can be used for diverse applications (Matthews, 2011). Several previous studies carried out to estimate CDOM
in freshwater have used empirical (regression) methods to analyse reflectance data from ground-, aircraft-, or satellite-based spec­
trometers, and most published relationships involve nonlinear (power) relationships and wavelengths >500 nm (Brezonik et al.,
2015). In the present work, following this method, images from the Operational Land Imager (OLI) sensor on board the Landsat-8 (L8)
satellite were used.
Regarding the bands used in previous studies using the sensors on Landsat satellites, Matthews (2011) and Chen et al. (2022) found
that the band ratio most commonly used in various studies was the green/red ratio. In more recent studies conducted using L8 images,
Olmanson et al. (2020) developed a model using the red/green and infrared/green ratios using the bootstrap forest technique and Chen
et al. (2022) used the blue/red ratio. Although the absorbance of CDOM is only significant below 500 nm, as for the estimation of
chlorophyll a, longer wavelengths are used as the concentration increases. In cases of low CDOM concentration their absorption
contributes greater proportional value to the total absorption coefficient in lower wavelengths (Chen et al., 2022), however in
CDOM-rich complex water bodies, the absorption of CDOM in green bands have the possibility to be strong and hence take consid­
erable proportions of the total absorption of the water (Chen et al., 2017). The red band used is for normalizing purposes as the in­
fluence of CDOM in red band is usually negligible (Kutser et al., 2015).
This demonstrates the possibility of using remote sensing to estimate CDOM, but although the use of the green/red ratio prevails,
unfortunately there is no generalized band ratio. Since the algorithms retrievals are affected by both the influence of each optically
actives constituents (phytoplankton, non-algal particles, and CDOM) on the reflectance spectrum, and the sensors characteristics as
radiometric resolution (Matthews, 2011).
In the present study, we researched CDOM more deeply through the use of remote sensing, establishing an algorithm to estimate
CDOM in the Ebro river basin reservoirs. Furthermore, we focused on two main aspects—on the one hand, we examined the CDOM
degradation process on the reservoirs’ water surfaces and checked whether our observations were in agreement with the values re­
ported in the literature, and on the other hand, we investigated how this degradation varied throughout the year, depending on the

Fig. 1. Reservoir locations within the Ebro basin; names are abbreviated as indicated in Table A1.

2
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

factors affecting the presence and degradation of CDOM.

2. Material and methods

2.1. Study area
The present study was carried out in the Ebro river basin in the northeast quadrant of the Iberian Peninsula, which flows in a
northwest–southeast direction through the depression formed between the Iberian mountain range and the Pyrenees. During the
summer periods from 2013 to 2018, a total of 23 different reservoirs were sampled and a total of 45 samples were analysed. The
reservoirs’ locations within the Ebro river basin are shown in Fig. 1. The reservoirs had total volumes ranging from 6 × 106 m3 (Albina)
to 1534 × 106 m3 (Mequinensa) and a maximum depth between 10 m in Terradets and 107 m in Itoiz. The climate variability greatly
influences the water quality and the reservoirs correspond to five different climate types, according to the Köppen–Geiger classifi­
cation, for the 1981–2010 period: Bsk (cold, semi-arid), Csa (Mediterranean hot-summer), Cfa (humid subtropical), Cfb (temperate
oceanic), and Dfb (humid continental). Reservoir water levels vary broadly throughout the year and are therefore highly dependent on
upstream rainfall and water temperature, meaning that HRT oscillates from 0.02 yr (La Peña) to 0.82 yr (Casp). In addition, other non-
climate factors, such as altitude, vary from 70 m.a.s.l. in Riba-roja to 1434 m.a.s.l. in Baserca. An overview of sampling sites and their
characteristics is presented in Appendix A (Table A1).

2.2. Field data collection and laboratory measurements

Samplings were carried out as part of the monitoring and control of heavily modified water bodies, accomplished by the Ebro Basin
Authority in compliance with the hydrological planning instructions that were approved in the Decree ARM/2656/2008, as part of the
implementation of the Spanish Water Framework Directive. A total of 34 field campaigns were carried out, 23 reservoirs were visited,
and finally we obtained a total of 45 data points, because some reservoirs were sampled more than once; this information is sum­
marized in Appendix A. Sampling campaigns were accomplished in the summer period, which is the time in which the maximum
stability of the water mass is observed during stratification and is the most unfavourable period for water quality because phyto­
plankton blooms can occur (EC, 2000).
The sampling protocol involved locating a point in the waterbody in the area of maximum depth, confirmed by echo-sounding,
situated 300–500 m from the dam to prevent disturbances by the possible outflow. Near the dam, the water quality is generally
better because of the self-depuration effect of reservoirs (the reduction in nutrient and suspended solid concentrations). A poor
ecological status recorded at this point can be inferred to be generalized throughout the reservoir. Once anchored, the coordinates of
the sampling point were geo-referenced with a GPS. The samples were taken at a depth of 2 m for two reasons: (1) the most sunlight
entering a clear water body is absorbed approximately within the first 2 m of the water column (Lillesand et al., 2014), and (2) the
modelled rates of photochemical decomposition were in good agreement with the measurements in a 2 m deep photolytic stratum
(Vähätalo and Wetzel, 2004). Samples were collected using a Ruttner bottle, then stored under dark and cool conditions for transfer to
the laboratory. Once in the laboratory, total suspended solids (TSS) were analysed using the gravimetric method (ISO-11923-1997),
the chlorophyll-a concentration (Chl_a) was measured using the spectrophotometric method, and CDOM was measured using fluo­
rescence spectroscopy. Chl_a samples were filtered through 0.4–0.6 μm GF/F glass fibber filters, extracted using standard methods
(Shoaf and Lium, 1976), and calculated with the Jeffrey and Humphrey (1975) methods. The filtered water was preserved in 125 mL
Nalgene® bottles in darkness and refrigerated (4 ◦ C) until the moment of analysis to measure CDOM, which was always executed
within 7 days of taking the sample.
A Hitachi F-7000 fluorescence spectrophotometer was used for analysis to obtain excitation emission matrix fluorescence spec­
troscopy (EEMS) data for each sample. EEMS is used to perform fluorescence measurements across a range of excitation and emission
wavelengths, presented as a 3D plot of fluorescence (Coble et al., 1990; Coble, 1996). A 1 cm quartz cuvette was used, which was
successively washed with MQ water between samples. The CDOM fluorescence intensity values were obtained at the excitation/e­
mission (Ex./Em.) pair of 370/460 nm, corresponding to EEMS Peak C, which is related to the presence in water of humic substances of
terrestrial or anthropogenic origin (Coble, 2007). The fluorescence intensity of CDOM concentrations was converted into μg L-1 of
quinine sulphate equivalents (QSE) based on the measurement of the fluorescence intensity of a serial dilution of quinine sulphate
dehydrate (Coble et al., 1998).

Table 1
Radiometric and spatial resolution details of OLI sensor bands. Modified from http://landsat.usgs.gov/landsat8.php.

Band Description Spatial resolution (m) Wavelength (μm)

1 (B1) Coastal/Aerosols 30 0.435 – 0.451

2 (B2) Blue 30 0.452 – 0.512
3 (B3) Green 30 0.533 – 0.590
4 (B4) Red 30 0.636 – 0.673
5 (B5) Near infrared) 30 0.851 – 0.879
6 (B6) SWIR 1 30 1.566 – 1.651
7 (B7) SWIR 2 30 2.107 – 2.294
8 Pancromatic 15 0.503 – 0.676
9 Cirrus 30 1.363 – 1.384
10 Thermal infrared 1 100 10.60 – 11.19
11 Thermal infrared 2 100 11.50 – 12.51

3
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

Regarding the CDOM expression in remote sensing, its concentration is usually expressed using the unit of absorption coefficient
(m-1) at a specific wavelength. The spectral absorbance is a measure of how much light is absorbed at a specific wavelength, whereas
fluorescence is a process where a substance emits longer wavelength light when exited to short wavelength light. Absorbance and
fluorescence CDOM laboratory measurements are often strongly correlated (Ferrari and Dowell, 1998; Singh et al., 2010; Liu et al.,
2014), but differences have been observed (Spencer et al., 2007; Stedmon et al., 2000) and may be indicative of changes in dissolved
organic matter composition that affect the ratio of absorbance to fluorescence (Slonecker et al., 2016). In agreement with Slonecker
et al. (2016), future CDOM research could benefit from the adoption of standard reporting units of CDOM and the establishment of
quantitative equivalencies for alternate measurement units. However, to facilitate the comparison between our data with other remote
sensing studies, we provide a further correlation between absorbance and fluorescence CDOM values. This is possible because we have
the absorbances at 400 nm of the filtered water, since the water colour (Mitchell and McDonald, 1991) of some samples (31) was
determined. And the absorption coefficient of CDOM at a fixed wavelength in the blue region (usually 400 or 440 nm) of the spectrum
is used as an indirect measure of CDOM concentration (Kallio, 2012). Therefore, relating the fluorescence data obtained to the ab­
sorption coefficient at 400 nm. Absorbance was converted to Napierian absorption coefficients (Kirk, 2010) using:
2.303 × A400
a400 = (1)
l

where: a400 is the absorption coefficient at 400 nm, A400 is absorbance at 400 nm, and l is cell path length (m). Absorbances were
blank (Milli-Q) corrected before conversion. CDOM values are reported as a400 (m-1).

2.3. Image processing and algorithm retrieval

An OLI sensor is mounted on board the United States Geological Survey (USGS/NASA) Landsat-8 satellite and has been in operation
since April 2013. The characteristics of the images are given in Table 1 and these were used to obtain the retrieval algorithm.
A total of 24 OLI images of Landsat Collection 1 Level 1, coincident with field campaigns (Appendix A, Table A2) were downloaded
from USGS official web page (http://earthexplorer.usgs.gov/). Atmospheric correction was made using QUAC (Quick Atmospheric
Correction) of the ENVI 5 program, obtaining surface reflectance. The ENVI QUAC method generates reflectance spectra within the
range of about 10 percent of the ground truth (Bernstein et al., 2012). Based on the empirical finding that the average reflectance of
diverse material spectra (except for highly structured materials such as vegetation, mud, and shallow water) is not scene-dependent,
QUAC determines atmospheric correction parameters directly from the observed pixel spectra in a scene, without ancillary information
(https://www.l3harrisgeospatial.com/docs/backgroundquac.html). QUAC processing requires starting from an image without scene
edges, without clouds (or with faint clouds), without large marine areas and with at least three valid bands. If clouds are present, they
must be masked out previously. The result will be a reflectance image of the area under consideration after performing the following
processes automatically: (1) Apply the Normalized Difference Mud Index (NDMI) to exclude shallow water or mud pixels; (2) Select the
points with the highest reflectance in the near-infrared band; (3) Constrain the gain curve to be constant at wavelengths below 650 nm;
(4) Suppresses the effect of dense vegetation and (5) Removes points with higher reflectance due to clouds in the infrared band if
available.
As algorithms derived using remote sensing reflectance (Rrs) performed much better than using reflectance in estimating CDOM
(Chen et al., 2017), the surface reflectance was converted to remote sensing reflectance by dividing by π, because the water is
approximated as a Lambert body (Novoa et al., 2017; Coimbra et al., 2019; Kuhn et al., 2019; Wang et al., 2019; Olmanson et al., 2020
Sòria-Perpinyà et al., 2022). If necessary we used B1 to mask the clouds, masking the digital values higher than 15,000. The
time-window between image acquisition and field campaigns was extended to four days, considering three factors: (1) HRT
(Table A1)—calculated as the ratio between the reservoir’s average daily outlet and the reservoir’s total volume; (2) meteorological
changes—knowing that summer is the most stable period in the study area, we checked the values of temperature, wind, and rain
between the image and sample data in most lakes; and (3) CDOM stability—concentrations were stable on at least a short-term basis
(days to weeks), e.g., see Brezonik et al. (2015). Meteorological stability was checked at https://www.aemet.es/es/datos_abiertos/
AEMET_OpenData, and we did not observe days with rain, the wind did not exceed 10 km/h, and temperature was stable in its
daily oscillations.
The best CDOM retrieval algorithm using the empirical method was determined using three approaches, one technical, one
bibliographic (finding most frequently used algorithm for Landsat sensors), and one theoretical (Table 2). The technical approach was
based on the emission wavelength used to obtain CDOM fluorescence, 460 nm, and only B2 (452–512 nm) of the OLI sensor contained
data for this wavelength. Regarding the bibliographic approach, Matthews (2011) and Chen et al. (2022) found that the band ratio
most commonly used in various studies was the green/red ratio. This was because for relatively insensitive sensors, the reflectance in
the green region—inversely related to CDOM—is normalized by reflectance in the red region (Matthews, 2011). The green/red ratio
was used by Slonecker et al. (2016), using bands 2 and 3 of the ETM + sensor on board Landsat 7, and was also used by Kutser (2012).

Table 2
Three approaches used to obtain the best CDOM retrieval algorithm.

Approach Used bands

Technical B2
Theoretical B1/B4 and B2/B4
Bibliographic B3/B4

4
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

Therefore, the ratio between bands B3 (533-590 nm) and B4 (636-673 nm) was calculated. Finally, the theoretical approach was based
on the principle that the algorithms using ratios between the reflectance band in the blue region (400-500 nm) and the reflectance
bands in the green or red regions (500-700 nm) have been demonstrated to be well correlated with CDOM (Kutser et al., 1998).
However, empirical algorithms in complex inland waters can be significantly improved by selecting at least one band with a relatively
longer wavelength (>600 nm), especially when the water optical properties are heavily influenced by constituents originating from
terrestrial vegetation or in cases with high concentrations of chlorophyll and nonalgal particles due to present high back-scattering
within the longer wavelengths. In this case the use of longer wavelengths for CDOM estimation reduce the possible effects of par­
ticulate matter (Zhu et al., 2014). With OLI bands, we have two possibilities, the ratio between B1 (435-451 nm) and B4 (636-673 nm)
and the ratio between B2 (452-512 nm) and B4 (636-673 nm) used by Chen et al. (2022).
The process for the calibration and validation of the algorithms was developed in two steps. In the first step, the data was equally
divided into two data sets, one for calibration with 23 data and another for validation with 22 data, to have data sets with the same
range. Once we had two data sets, the algorithms were calibrated by performing a regression between the CDOM data and band
combinations using L8–OLI, finding the best-fitting function (linear, potential, exponential, or polynomial). The algorithms were
validated by means of five error statistics, using observed and estimated data: the coefficient of determination (R2), calculated by
adjusting the linear regression between field and estimated data; the root mean squared error (RMSE); the relative root mean squared
error (RRMSE); bias and the root mean squared log-based error (RMSElog), due to the fact that log-based error is generally used for
variables (e.g., CDOM and other ocean colour components) with logarithmic distributions (IOCCG and Lee, 2006).
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
∑N estimated 2
i=1 (xi − xmeasured )
RMSE = i
(2)
N

RMSE
RRMSE = ∑N measured
/ ×100 (3)
i=1 xi N
∑n ( )
xestimated − xmeasured
Bias = i=1 i i
(4)
N
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
∑N ( estimated )2
i=1 logi − logmeasured
RMSElog = i
(5)
N− 2

In the second step, the best-performing algorithm was recalculated using the entire database with the aim of obtaining a more robust
and accurate algorithm.

2.4. Image processing and algorithm retrieval

A total of 35 images were processed to study both, the CDOM degradation throughout the water bodies during the summer period,
and the annual temporal variation of CDOM degradation. Since the data for obtaining the algorithm were taken during the summer
period, the first study was carried out during the summer period on 16 different water bodies covering the entire study area. Sub­
sequently, the study of the annual variation was carried out on 4 reservoirs only, whose selection was based on their location within the
hydrological basin (right or left margin of the river) and their drainage basin area, which is directly related to the inflow of matter into
the system. Were selected two reservoirs belong to the left margin, one in the headwaters with <1000 km2 of watershed, Itoiz, and
another of in the main network with <25000 km2 of watershed, Barasona. A reservoir on the right margin, Cueva Foradada, with
<1000 km2 of watershed and in the headwaters. And a reservoir belonging to lower section of the main axe, with a watershed >25000
km2.
In order to calculate the degradation rates and compare the degradation time with those of previous studies, we use two equations
based on the percentage equation formula. To calculate the CDOM degradation rate, we applied one percentage equation (Eq. 6) using
the CDOM tail value as 100% and the CDOM in the dam as the value to calculate the remaining percentage of CDOM and subtracting
100 to obtain the degraded percentage. Furthermore, we applied equation (7), a simple percentage equation to calculate the time
necessary to reach this degradation rate, using the experiments of Vähätalo and Wetzel (2004) as a reference, in which 70-day solar
radiation decomposed 96% of the CDOM. Finally, we compared the calculated time with the HRT.
CDOMdam ×100
CDOMdegradation % = − 100 (6)
CDOMtail

CDOMdegradation ×70
Degradation time = (7)
96
To study the temporal variation in CDOM degradation, the CDOM values at the tail and dam of each image, as well as the CDOM
degradation percentage between these points (Eq. (6)), and the instantaneous direct solar radiation value are plotted. The instanta­
neous direct solar radiation (Direct) values coinciding with each L8 image were obtained from band Direct of the MCD18A1 product,
the Level-3 Terra and Aqua combined MODIS-derived surface radiation product, downloaded from the Level-1 and Atmosphere
Archive & Distribution System Distributed Active Archive Center (https://ladsweb.modaps.eosdis.nasa.gov/search/order/2).
Furthermore, to assess the strength of the association between the four variables plotted to show the annual variation, nonparametric

5
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

Spearman correlation coefficients were calculated because all data did not have a normal distribution.

3. Results
During the study, 45 samples were collected on 23 different reservoirs. The CDOM values were highly variable, ranging from ten to
hundreds of QSE, reflecting the diversity of environments studied. We observed an average of 58.93 μg L-1 QSE and a standard de­
viation of 22.55 μg L-1 QSE, a minimum of 18.48 μg L-1 QSE, and a maximum of 112.68 μg L-1 QSE. A summary of the descriptive
statistics for the measured variables are shown in Table 3. The calculated covariance values between the three variables were all
positive, indicating a direct relation among them.
Regarding to the relationship between the concentration of CDOM and CDOM absorption coefficient at 400 nm, as the two database
passed the Normality Test of Shapiro-Wilk, a linear regression is used (Fig. 2). The correlation was significantly and positively linearly
correlated (R2 = 0.63, p<0.001).
As a result of the algorithm development, the calibration relations results are presented in Table 4. According to the coefficient of
determination, the best result with a difference was obtained with the use of the B1/B4 ratio. The validation processes using 22 samples
with the B1/B4 ratio resulted in a coefficient of determination of 0.87, an RMSE of 8.14 μg L-1 QSE, an RRMSE of 13%, a bias of 3.70 μg
L-1 QSE, and an RMSElog of 0.07.
Fig. 3 shows the relationship between the B1/B4 ratio and the CDOM concentrations using all data. The resulting potential equation
(Eq. 8) is the one that was applied to the images in order to study the CDOM degradation processes throughout reservoirs and to
examine how this degradation varied over the year.
( )− 0.71
B1
CDOM= 66.983 × (8)
B4
A total of 16 reservoirs were studied during the summer period to understand the CDOM degradation throughout the water bodies.
For this purpose, we obtained the CDOM values of the reservoir tail and dam from the images (Table 5) and calculated the CDOM
degradation rate using equation (6) and degradation time using equation (7). The sequence presented in Table 4 was based on the use
of the HRT to observe the differences between short and longer HRTs. For short HRTs, the estimated degradation was similar to the
degradation observed experimentally at short HRTs, for example, one month, but for longer HRTs, the estimated degradation was
lower than the experimentally observed degradation. The homogeneity of the CDOM concentration in the El Val reservoir was
remarkable and was probably due to its hypereutrophic state.
To study the annual temporal variation in CDOM degradation, we processed images obtained throughout the year of the four
reservoirs, representing the entire study area: two situated on the left side of the river (Itoiz and Barasona), where the Pyrenees are
located; another situated on the left side (Cueva Foradada), where the Iberian mountain range is located; and another situated on the
main river (Mequinensa). Figs. 4–7 show the thematic CDOM maps of each processed image and a visual plot of the temporal variations
in the reservoir tail and dam CDOM values, their degradation rates and the Direct values. For all reservoirs, a peak of maximum
degradation can be observed in spring, coinciding with the minimum annual concentration of CDOM in the dam. In autumn and
winter, the degradation was lower and therefore the concentration reaching the dam was higher, except in the case of Mequinensa, the
largest reservoir.
The Spearman correlations results (Table 6) between the 4 variables used for the annual variation study (Tail CDOM, Dam CDOM,
CDOM degradation rate and Direct), using all values from 4 reservoirs studied, confirm statistically the trends of those variables in the
graphs of images 3, 4 and 5. Positive significant correlation exists between the CDOM degradation values and Direct and with the same
significance but opposite sign between the CDOM dam values and Direct. The same occurs between tail and dam CDOM values and the
CDOM degradation rates, establishing a positive relationship with tail CDOM values and a negative relationship with dam CDOM
values, both with the same significance.

4. Discussion
Surface data reflects values in some cases that are similar to those rivers, with concentrations of hundreds of μg L-1 QSE, coinciding
with other studies on rivers such as a study by Chen and Gardner (2004), whose results showed concentrations of 100 μg L-1 QSE in the
Mississippi and Atchafalaya River plume regions, and Huang and Chen (2009), who obtained results reaching 200 μg L-1 QSE in the
wetlands of the Neponset River. Other reservoirs with lower concentrations (tens of μg L-1 QSE) exhibited results that were more
similar to those of lakes, coinciding with the data obtained for some lakes and reservoirs in tropical systems used in the comparisons of
Rochelle-Newall et al. (2014).
As expected, a positive covariance exists between CDOM and Chl_a and TSS, indicating that the variables tend to move in the same

Table 3
Descriptive statistics of the measured variables: chromophoric dissolved organic matter (CDOM), total suspended sediment (TSS), and chlorophyll a (Chl_a). N: data
number; Min.: minimum; Max.: maximum; SD: standard deviation.

N Min. Max. Average Median SD

-1
CDOM (μg L QSE) 45 18.48 112.68 61.92 64.64 25.55
TSS (mg L-1) 14 1.38 11.92 3.59 2.76 2.75
Chl_a (mg m-3) 45 1.00 18.14 4.31 2.49 4.21

6
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

Fig. 2. Linear relationship between the concentration of CDOM and CDOM absorption coefficient at 400 nm.

Table 4
Calibration and validation results.

Bands relation Calibration (N = 23) Validation (N = 22)

R2 R2 RMSE (μg L-1) RRMSE (%) Bias (μg L-1) RMSElog

B2 0.404 0.10 24.33 39 13.19 0.21

B1/B4 0.832 0.87 8.14 13 3.70 0.07
B2/B4 0.665 0.59 14.95 24 7.11 0.11
B3/B4 0.277 0.34 17.22 28 0.74 0.13

Fig. 3. CDOM algorithm, recalculated using the entire database.

direction. These variables are related with CDOM because the Chl_a is a proxy of phytoplankton quantity, and algae decomposition is
part of CDOM autochthonous source. And TSS are related with both autochthonous and allochthonous CDOM source, but specially
with allochthonous source when TSS is dominated by the dragging due to rain and thawing and the dragging of particulate matter due
to the effect of waves during reservoir outflow.
Determining a global relationship between absorbance and fluorescence CDOM measurements is challenging, due to the diverse
methodology to measure both. But in our case, for the 31 samples with both measures, the correlation was statistically significant
agreeing with Ferrari and Dowell (1998), Singh et al. (2010) and Liu et al. (2014) studies. And, in spite of using different methods to
measure absorbance and fluorescence values, although with a lower R2, the degree of significance was the same.
Regarding the developed algorithm, the relationship between CDOM concentration and the B1/B4 ratio is an inverse relationship,
so the higher concentration of CDOM, the lower the B1/B4 value, coinciding with other studies relating bands ratios to CDOM con­
centrations, such as: Kahru and Mitchell (2001); D’Sa and Miller (2003); Kallio et al. (2008); Kutser et al. (2005); Kutser (2012); Chen
et al. (2022). This relationship was to be expected, since CDOM absorbs radiation and therefore the signal reaching the sensor is lower.
However, according to the data of the present study, the best regression was potential, coinciding with the findings of Kutser et al.
(2005), Kutser (2012), Slonecker et al. (2016) and Chen et al. (2022).
Using all the data, we developed an algorithm for a wide range of CDOM concentrations. Considering the number of samples used
(45), this algorithm exhibited high performance, since the associated RMSElog of 0.07 was very low and similar to the RMSElog of 0.06

7
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

Table 5
CDOM values of the reservoir tail and dam, their degradation rate, the HRT and the corresponding CDOM degradation according to the experimental results (2004).

Name Image CDOM (μg/L QSE) HRT Degradation (%)

dd/mm/yyyy Tail Dam days Estimated Experimental

Albiña 11/09/2016 57 43 15 24 20
Oliana 01/09/2014 100 70 28 30 38
Mansilla 25/07/2016 66 38 29 42 40
Barasona 21/06/2017 53 28 34 47 47
Eugi 22/08/2017 42 30 35 30 48
Camarasa 21/06/2017 66 34 40 51 55
Ortigosa 25/07/2016 60 27 68 55 93
Mequinensa 30/07/2017 80 48 72 40 99
Yesa 22/08/2017 65 34 86 48 >96
Aalloz 22/08/2017 66 39 100 59 >96
Rialb 26/07/2018 95 70 117 27 >96
Itoiz 10/09/2018 55 38 126 31 >96
Urrunaga 11/09/2016 70 40 133 57 >96
Cueva foradada 12/06/2017 77 14 198 82 >96
El Val 03/08/2016 52 50 212 4 >96
La Tranquera 03/08/2016 80 34 256 58 >96

Fig. 4. Surface CDOM annual variation in Mequinensa reservoir in images (a-e), and the plot (f) representing the image CDOM estimated values in tail, dam, the
respective CDOM degradation rate expressed as a % and the Direct Solar Radiation (Direct) values of the same day.

8
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

Fig. 5. Surface CDOM annual variation in Barasona reservoir in images (a-h), and the plot (i) representing the image CDOM estimated values in tail, dam, the
respective CDOM degradation rate expressed as a % and the Direct Solar Radiation (Direct) values of the same day.

obtained by Zhu et al. (2014), who used the algorithm developed by Kutser et al. (2005) in seven samples with high CDOM con­
centrations in their evaluation of CDOM algorithms in complex freshwater environments. Our RRMSE also reached a good result, the
13% error obtained is lower than 26% obtained by Chen et al. (2022) using 66 samples and the same bands ratio (blue/red) and minor
than ~25–30% obtained by Olmanson et al. (2020) for three CDOM ranges (low, medium, high) using the red/green and infrar­
ed/green ratios in the bootstrap forest technique.
We obtained a robust algorithm by using summer period data due to the greater influence of dissolved organic matter on the
reflectance spectrum, which may be due to two factors. The first factor is the primary production, which decreases due to nutrient
depletion after the spring peak, reducing the influence of Chl_a on the reflectance spectrum, and the second factor is the presence of
allochthonous dissolved organic matter, which increases due to the outflow of water for agricultural uses. Outflow causes a sharp drop
in the reservoir levels, and variations in the volumes and heights of the reservoirs can generate excess particulate matter because of
coastline instability, which facilitates the dragging of particulate matter due to the effect of waves (Ordoñez, 2010).
Regarding the spatial variation due to the self-depuration effect of reservoirs, in addition to reducing the concentration of nutrients
and suspended solids, reservoirs regulate and reduce the variability of the flows of these variables (Margalef, 1983). Therefore, due to
the positive relation between Chl-a, TSS, and CDOM, the same occurs with the spatial variations in the CDOM concentration, since the
CDOM concentration values in the dam reservoir were mostly low in the different reservoirs studied in summer and throughout the
year. Regarding the degradation rates, we can observe that in reservoirs where the HRT was very low, probably solar radiation
degradation was predominant, as in the experiments of Vähätalo and Wetzel (2004). Our results were in accordance with those of the
experiments, as we can see in the example of the Barasona reservoir. Barasona CDOM concentration thematic map from 21 June 2017
(Fig. 5) show a CDOM concentration in the tail of 55 μg L-1 QSE, due to the entrance of the river with nutrients and organic matter. The
47% of CDOM concentration is degraded throughout the reservoir until reaching in the dam area a concentration of 28 μg L-1 QSE,

9
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

10 (caption on next page)

X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

Fig. 6. Surface CDOM annual variation in Cueva Foradada reservoir in images (a-f), and the plot (g) representing the image CDOM estimated values in tail, dam, the
respective CDOM degradation rate expressed as a % and the Direct Solar Radiation (Direct) values of the same day.

probably due to photobleaching. Barasona’s HRT was 34 days with a CDOM degradation rate of 47%, a degradation time similar to that
observed in the Vähätalo and Wetzel Vähätalo and Wetzel (2004) experiments, because if they obtain for 70-day exposure to solar
radiation a decomposition of the 96% of freshwater CDOM, this is equivalent to a decomposition of the 47% in 34 days.
In contrast, when HRT was higher, other factors such as runoff inputs or in situ production played a more important role.
Furthermore, in situ chromophoric degradation is masked by the constant input of new material in volumes that can be linked with rain
events (Raymond and Saiers, 2010). This occurred in the El Val reservoir, a hypertrophic reservoir where only 4% of CDOM was
degraded over an HRT of 212 days, which was probably due to autochthonous CDOM production. On the other hand, in the Itoiz (HRT
of 126 days and degradation rate of 31%) and Rialb (HRT of 117 days and degradation rate of 27%) reservoirs, the low degradation
rates could have been due to the presence of more recalcitrant CDOM. These reservoirs were located downstream and very close to
other reservoirs—Irabia and Oliana, respectively—so their CDOM characteristics depended on the outflow water depth.
Regarding the annual variation in CDOM degradation, previous studies in laboratory experiments have shown higher photo­
degradation in early spring (Gonsior et al., 2013; Kragh et al., 2022), which is in accordance with our results. Furthermore, we agree
with Kragh et al. (2022) regarding the higher degradation in spring compared to summer, because more autochtonous labile CDOM are
present in springtime water as consequence of the algae decomposition due to the spring primary production peak. In contrast, in
autumn and winter there was less degradation due to less photodegradation and the greater entry of allochthonous CDOM, which is
more recalcitrant, due to rain and thawing, as can be seen in the image of the Itoiz reservoir from 29 January 2018 (Fig. 7). Therefore,
the CDOM degradation rate variation is mainly and directly influenced, as evidenced by the statistical correlation, by photo­
degradation, however another very important factor in the annual variation was the reservoir flow (filling and emptying). For example,
in the case of the Mequinensa reservoir, in the image from 30 July 2017 (Fig. 4), the emptying of the reservoir in summer led to a
decrease in the CDOM degradation rate due to the presence of more recalcitrant material that washed up from the coastline. On the
contrary, as a consequence of stopping it from emptying, in the image from 19 November the degradation increased, and therefore
there was a reduction in the CDOM concentration in the dam. In the image from 19 January, an increase in the CDOM degradation rate
occurred because the sediment input from rainfall and thawing had not yet reached the dam. Something similar occurred in the
Barasona reservoir (Fig. 5), for which the image from 8 August 2017 showed the highest degradation rate, because organic matter
dragged into the reservoir by the incipient emptying process had not yet reached the dam, and the lowest CDOM concentration values
were recorded in summer in the dam. Once emptying stopped, as shown in the image from 16 September, degradation decreased but it
increased again when filling started, as shown in the image from 19 November. This was the reservoir in which the annual variation of
CDOM was most greatly affected by reservoir water management.
Given the studied variables annual variation and their correlation, in the Ebro basin reservoirs the CDOM degradation is directly
related to the Direct values. Only in moments with large inflows and outflows variations, associated with reservoir water management,
the CDOM degradation rates are more associated with the inflow of allochthonous material or with the resuspension of sediment due to
emptying. This lack of correlation in flow moments is due to the lack of correlation between the tail CDOM values and the Direct
values. Thus, while the dam CDOM values are directly related to the Direct values and thus to photobleaching, the tail CDOM values do
not have a significant relationship with the Direct values because they are more affected by the filling and emptying processes.
Using remote sensing techniques, we established an algorithm to estimate CDOM in Ebro river basin reservoirs to further explore
the study of the degradation of organic matter at the water surface, as affected by photo bleaching and microbial activity. In addition,
in this work we provide CDOM degradation rate data for the longitudinal profile of the reservoirs’ surfaces, calculated based on CDOM
values estimated by means of remote sensing. CDOM degradation rates and their annual variations under certain conditions of solar
radiation predominance and HRT were in agreement with data obtained in laboratory experiments.
Thus, even if it only applies to the water surface, this work represents a step forward in acknowledging the role of inland water in
the carbon cycle, demonstrating once again the usefulness of remote sensing in studies of surface water processes. Further studies of
this type should be carried out in other climatic conditions, but together with thematic maps that help us to identify the causes of
variation in CDOM concentrations, such as Chl_a and TSS which are variables directly related with CDOM. This would provide us with
more information about degradation rates and organic matter retention in reservoirs, since measurements of Chl_a would give us
information about autochthonous and more labile CDOM, whereas TSS would provide information about allochthonous and more
recalcitrant CDOM.

5. Conclusions
This study, like many previous ones, has demonstrated the usefulness of OLI sensor data in CDOM estimation, but has also
demonstrated its application for the study of the water surface CDOM degradation of a reservoir and over an annual period.
The degradation rates calculated in this work for reservoirs in the summer period, when the solar radiation predominated over
other factors, and with HRTs up to 30–40 days, were in agreement with those obtained in laboratory experiments on photobleaching,
whereas at higher HRTs other autochthonous or allochthonous factors started to have an influence.
The annual variation in the degradation rate exhibited a maximum value in spring, simply because more degradable CDOM was
present in springtime water, and showed a minimum value in autumn and winter, with less light degradation and the increased entry of
more recalcitrant allochthonous CDOM, coinciding with the results of other studies based on laboratory experiments. However,
depending on reservoir water management practices, this maximum could be displaced to the summer period. Therefore, the CDOM

11
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

12 (caption on next page)

X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

Fig. 7. Surface CDOM annual variation in Itoiz reservoir in images (a-d), and the plot (e) representing the image CDOM estimated values in tail, dam, the respective
CDOM degradation rate expressed as a % and the Direct Solar Radiation (Direct) values of the same day.

Table 6
Spearman correlation coefficients and p-values (in parenthesis).

N Dam CDOM CDOM degradation Direct

Tail CDOM 23 n.s. 0.7 (<0.001) n.s.

Dam CDOM 23 -0.7 (<0.001) -0.5 (<0.05)
CDOM degradation 23 0.5 (<0.05)

degradation rate variation is mainly and directly influenced, as evidenced by the statistical correlation, by photodegradation, however
another very important factor in the annual variation was the reservoir flow (filling and emptying).

Author contributions
Xavier Sòria-Perpinyà: Conceptualization, methodology, formal analysis, writing—original draft. Juan Soria: Conceptualization,
image processing, writing—review and editing, funding acquisition. Eduardo Vicente: writing—review and editing, project admin­
istration, Funding acquisition.

Funding
This work was carried out thanks to the projects regarding the monitoring and control of heavily modified water bodies studies
accomplished by the Ebro Basin Authority (Spanish Government) to E.V. and J.S.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to
influence the work reported in this paper.

Data availability

Data will be made available on request.

Acknowledgments
The authors are very grateful to Olga Kramer for her efforts in laboratory analyzes and to Maria Dolores Sendra for her efforts in the
field campaigns. We are also very grateful to the reviewers for their comments and feedback.

Appendix A

Table A.1
Summary of sampled reservoirs characteristics and field campaigns information. Symbols and abbreviations: max.: maximum; m.a.s.l.: meters above sea level.

Name Abreviation Depth Volume Elevation HRT Climate Visits Year Samples

m (max.) hm3 m.a.s.l. years

Albina ALB 25 6 605 0.04 Cfb 1 16 2

Alloz ALL 60 65 468 0.28 Cfb 1 15 1
Barasona BAR 60 92 447 0.10 Cfa 5 13, 14, 16, 17, 18 7
Baserca BAS 70 22 1434 0.11 Dfb 1 14 1
Camarasa CAM 91 113 376 0.11 Cfb 1 14 1
Casp CAS 46 82 230 0.82 Bsk 2 14, 15 2
Cueva Foradada CUE 43 22 580 0.55 Bsk 2 15, 18 3
Eugi EUG 43 21 628 0.10 Cfb 1 13 2
Itoiz ITO 107 418 588 0.35 Cfb 1 13 1
La Loteta LOT 34 105 288 0.40 Bsk 1 14 1
La Peña PEÑ 32 15 539 0.02 Csa 1 15 2
Mansilla MAN 72 68 930 0.08 Cfb 1 16 1
Mediano MED 72 436 528 0.23 Cfa 1 16 1
Mequinensa MEQ 79 1534 121 0.20 Bsk 4 13, 14, 15, 17 5
Oliana OLI 73 101 518 0.08 Cfa 1 14 1
Ortigosa ORT 54 33 995 0.19 Cfb 1 17 1
Pajares PAJ 61 35 1288 0.48 Dfb 1 16 1
Riba-roja RIB 34 210 70 0.03 Bsk 1 17 1
(continued on next page)

13
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

Table A.1 (continued )

Name Abreviation Depth Volume Elevation HRT Climate Visits Year Samples

m (max.) hm3 m.a.s.l. years

Sobrón SOB 33 20 511 0.14 Cfb 1 16 2

Terradets TER 10 23 372 0.03 Cfa 1 18 1
Uribarri-Ganboa URI 30 147 546 0.77 Cfb 1 16 1
Urrunaga URR 24 72 547 0.37 Cfb 2 16, 17 2
Yesa YES 61 447 489 0.24 Cfb 2 15, 17 5
Total 34 45

Table A.2
Images of Landsat Collection 1 Level 1 coincident with field campaigns.

Year WRS2 Image date Reservoir Sample date Difference Samples

Path Row

2013 200 30 26-jul ITO 23-jul 3 1

EUG 24-jul 2 2
199 31 20-ago MEQ 23-aug 3 1
198 31 14-sep BAR 17-sep 3 2
2014 199 30 22-jul BAS 23-jul 1 1
198 31 31-jul BAR 29-jul 1 1
199 31 07-ago LOT 06-aug 1 1
198 31 01-sep CAM 01-sep 0 1
OLI 02-sep 1 1
198 32 17-sep CAS 17-sep 0 1
199 31 24-sep MEQ 24-sep 0 1
2015 199 32 07-jun CUE 08-jun 1 2
200 30 30-jun PEÑ 30-jun 0 2
YES 30-jun 0 3
ALL 01-jul 1 1
198 31 03-ago MEQ 03-aug 0 2
CAS 05-aug 2 1
2016 201 30 23-jun ALB 21 jun 2 2
201 30 09-jul URR 12-jul 3 1
URI 12-jul 3 1
SOB 13-jul 4 2
201 31 25-jul PAJ 25-jul 0 1
MAN 26-jul 1 1
198 31 12-aug MED 9-aug 3 1
BAR 10-aug 2 1
200 30 05-jul YES 04-jul 1 2
2017 201 31 28-jul ORT 24-jul 4 1
201 30 28-jul URR 25-jul 1 1
198 31 30-jul MEQ 1-aug 2 1
RIB 2-aug 3 1
198 31 8-aug BAR 8-aug 0 2
2018 199 32 17-jul CUE 16-jul 1 1
198 31 17-jul TER 17-jul 0 1
198 31 26-jul BAR 23-jul 3 1

References
Ayoub, L.M., Hallock, P., Coble, P.G., Bell, S.S., 2012. MAA-like absorbing substances in Florida Keys phytoplankton vary with distance from shore and CDOM:
implications for coral reefs. J. Exp. Mar. Biol. Ecol. 420, 91–98. https://doi.org/10.1016/j.jembe.2012.03.026.
Bernstein, L.S., Jin, X., Gregor, B., Adler-Golden, S.M., 2012. Quick atmospheric correction code: algorithm description and recent upgrades. Opt. Eng. 51 (11),
111719, 111719.
Bidigare, R., Orldrusekh, M., Brooks, J., 1993. Influence of the Orinoco river outflow on distribution of algal pigments in the Caribbean Sea. J. Geophys. Res. 98 (2),
2259–2269. https://doi.org/10.1029/92JC02762.
Blough, N.V., Green, S.A., 1995. Spectroscopic characterization and remote sensing of non-living organic matter. In: Zepp, R.G., Sonntag, C. (Eds.), The Role of
Nonliving Organic Matter in the Earth’s Carbon Cycle, pp. 23–45.
Blough, N.V., Zepp, R.G., 1990. Effects of Solar Ultraviolet Radiation on Biogeochemical Dynamics in Aquatic Environments: Report of a Workshop, Marine Biological
Laboratory, Woods Hole, Massachusetts, October 23-26, 1989. Woods Hole Oceanographic Institution Technical Report, p. 194.
Brezonik, P.L., Olmanson, L.G., Finlay, J.C., Bauer, M.E., 2015. Factors affecting the measurement of CDOM by remote sensing of optically complex inland waters.
Remote Sens. Environ. 157, 199–215. https://doi.org/10.1016/j.rse.2014.04.033.
Chen, R.F., Gardner, G.B., 2004. High-resolution measurements of chromophoric dissolved organic matter in the Mississippi and Atchafalaya River plume regions.
Mar. Chem. 89 (1), 103–125. https://doi.org/10.1016/j.marchem.2004.02.026.
Chen, J., Zhu, W., Pang, S., Cheng, Q., 2022. Applicability evaluation of Landsat-8 for estimating low concentration colored dissolved organic matter in inland water.
Geocarto Int. 37 (1), 1–15. https://doi.org/10.1080/10106049.2019.1704071.
Chen, J., Zhu, W.N., Tian, Y.Q., Yu, Q., 2017. Estimation of colored dissolved organic matter from landsat-8 imagery for complex inland water: case study of lake
huron. IEEE Trans. Geosci. Rem. Sens. 55 (4), 2201–2212.

14
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

Coble, P.G., 2007. Marine optical biogeochemistry: the chemistry of ocean color. Chem. Rev. 107 (2), 402–418. https://doi.org/10.1021/cr050350+.
Coble, P.G., Del Castillo, C.E., Avril, B., 1998. Distribution and optical properties of CDOM in the arabian sea during the 1995 southwest monsoon. Deep Sea Res. Part
II Top. Stud. Oceanogr. 45 (10), 2195–2223. https://doi.org/10.1016/S0967-0645(98)00068-X.
Coble, P.G., 1996. Characterization of marine and terrestrial DOM in seawater using excitation-emission matrix spectroscopy. Mar. Chem. 51 (4), 325–346. https://
doi.org/10.1016/0304-4203(95)00062-3.
Coble, P.G., Green, S.A., Blough, N.V., Gagosian, R.B., 1990. Characterization of dissolved organic matter in the Black Sea by fluorescence spectroscopy. Nature 348
(6300), 432–435. https://doi.org/10.1038/348432a0.
Coimbra, K.T.O., Alcântara, E., de Souza Filho, C.R., 2019. An assessment of natural and manmade hazard effects on the underwater light field of the Doce River
continental shelf. Sci. Total Environ. 685, 1087–1096. https://doi.org/10.1016/j.scitotenv.2019.06.127.
Cory, R.M., Ward, C.P., Crump, B.C., Kling, G.W., 201). Sunlight controls water column processing of carbon in arctic fresh waters. Science. 345(6199), 925-928. doi:
10.1126/science.1253119.
Devesa-Rey, R., Barral, M.T., 2012. Allochthonous versus autochthonous naturally occurring organic matter in the Anllóns river bed sediments (Spain). Environ. Earth
Sci. 66 (3), 773–782. https://doi.org/10.1007/s12665-011-1286-3.
D’Sa, E.J., Miller, R.L., 2003. Bio-optical properties in waters influenced by the Mississippi River during low flow conditions. Remote Sens. Environ. 84 (4), 538–549.
https://doi.org/10.1016/S0034-4257(02)00163-3.
[EC] European Communities, 2000. Directive 2000/60/CE of the European Parliament and of the Council of 23 October 2000 establishing a framework for
Community action in the field of water policy. DOUE L, núm 327, 1–73, 22-12-2000.
Ferrari, G.M., Dowell, M.D., Grossi, S., Targa, C., 1996. Relationship between the optical properties of chromophoric dissolved organic matter and total concentration
of dissolved organic carbon in the southern Baltic Sea region. Mar. Chem. 55 (3), 299–316. https://doi.org/10.1016/S0304-4203(96)00061-8.
Ferrari, G., Dowell, M., 1998. CDOM absorption characteristics with relation to fluorescence and salinity in coastal areas of the southern Baltic Sea. Estuar. Coast Shelf
Sci. 47, 91–105. https://doi.org/10.1006/ecss.1997.0309.
Gonsior, M., Schmitt-Kopplin, P., Bastviken, D., 2013. Depth-dependent molecular composition and photo-reactivity of dissolved organic matter in a boreal lake under
winter and summer conditions. Biogeosciences 10, 6945–6956. https://doi.org/10.5194/bg-10-6945-2013,2013.
Huang, W., Chen, R.F., 2009. Sources and transformations of chromophoric dissolved organic matter in the Neponset River Watershed. J. Geophys. Res.
Biogeosciences. 114, G00F05. https://doi.org/10.1029/2009JG000976.
IOCCG, 2006. Remote sensing of inherent optical properties: fundamentals, tests of algorithms, and applications. In: Lee, Z.P. (Ed.), Reports of the International
Ocean-Colour Coordinating Group, No. 5. (Dartmouth, NS: IOCCG), 1–126.
Jeffrey, S.T., Humphrey, G.F., 1975. New spectrophotometric equations for determining chlorophylls a, b, c1 and c2 in higher plants, algae and natural
phytoplankton. Biochem. Physiol. Pflanz. (BPP) 167, 191–194. https://doi.org/10.1016/S0015-3796(17)30778-3.
Kahru, M., Mitchell, B.G., 2001. Seasonal and nonseasonal variability of satellitederived chlorophyll and dissolved organic matter concentration in the California
Current. J. Geophys. Res. 106, 2517–2529. https://doi.org/10.1029/1999JC000094.
Kallio, K., Attila, J., Härmä, P., Koponen, S., Pulliainene, J., Hyytiäinen, U.M., Pyhälahti, T., 2008. Landsat ETM images in the estimation of seasonal lake water
quality in boreal river basins. Environ. Manag. 42, 511–522. https://doi.org/10.1007/s00267-008-9146-y.
Kallio, K., 2012. Water Quality Estimation by Optical Remote Sensing in Boreal Lakes; Monographs of the Boreal Environment Research 39; the Finnish Environment
Institute. Finland, Helsinki.
Kirk, J.T.O., 2010. Light and Photosynthesis in Aquatic Ecosystems, third ed. University Press, Cambridge, UK. https://doi.org/10.1017/CBO9781139168212.
Kragh, T., Sand-Jensen, K., Kristensen, E., Pedersen, O., Madsen-Østerbye, M., 2022. Removal of chromophoric dissolved organic matter under combined
photochemical and microbial degradation as a response to different irradiation intensities. J. Environ. Sci. 118, 76–86. https://doi.org/10.1016/j.
jes.2021.08.027.
Kuhn, C., de Matos Valerio, A., Ward, N., Loken, L., Sawakuchi, H.O., Kampel, M., et al., 2019. Performance of Landsat-8 and Sentinel-2 surface reflectance products
for river remote sensing retrievals of chlorophyll-a and turbidity. Remote Sens. Environ. 224, 104–118. https://doi.org/10.1016/j.rse.2019.01.023.
Kutser, T., 2012. The possibility of using the Landsat image archive for monitoring long time trends in coloured dissolved organic matter concentration in lake waters.
Remote Sens. Environ. 123, 334–338. https://doi.org/10.1016/j.rse.2012.04.004.
Kutser, T., Arst, H., Mäekivi, S., Kallaste, K., 1998. Estimation of the water quality of the Baltic Sea and lakes in Estonia and Finland by passive optical remote sensing
measurements on board vessel. Lakes Reservoirs Res. Manag. 3, 53–66. https://doi.org/10.1111/j.1440-1770.1998.tb00032.x.
Kutser, T., Pierson, D., Tranvik, L., Reinart, A., Sobek, S., Kallio, K., 2005. Using satellite remote sensing to estimate the colored dissolved organic matter absorption
coefficient in lakes. Ecosystems 8, 709–720. https://doi.org/10.1007/s10021-003-0148-6.
Kutser, T., Verpoorter, C., Paavel, B., Tranvik, L.J., 2015. Estimating lake carbon fractions from remote sensing data. Remote Sens. Environ. 157, 138–146. https://
doi.org/10.1016/j.rse.2014.05.020.
Lillesand, T., Kiefer, R.W., Chipman, J., 2014. Remote Sensing and Image Interpretation. John Wiley & Sons.
Liu, X., Zhang, Y., Shi, K., Zhu, G., Xu, H., Zhu, M., 2014. Absorption and fluorescence properties of chromophoric dissolved organic matter: implications for the
monitoring of water quality in a large subtropical reservoir. Environ. Sci. Pollut. Res. 21, 14078–14090.
Margalef, R., 1983. Limnología. Ediciones Omega, Barcelona.
Matthews, M.W., 2011. A current review of empirical procedures of remote sensing in inland and near-coastal transitional waters. Int. J. Rem. Sens. 32 (21),
6855–6899. https://doi.org/10.1080/01431161.2010.512947.
Mitchell, G.N., McDonald, A.T., 1991. Relationship between different methods of colour measurement in potable water. Meas. Control 24 (10), 306–309.
Novoa, S., Doxaran, D., Ody, A., Vanhellemont, Q., Lafon, V., Lubac, B., Gernez, P., 2017. Atmospheric corrections and multi-conditional algorithm for multi-sensor
remote sensing of suspended particulate matter in low-to-high turbidity levels coastal waters. Rem. Sens. 9 (1), 61. https://doi.org/10.3390/rs9010061.
Olmanson, L.G., Page, B.P., Finlay, J.C., Brezonik, P.L., Bauer, M.E., Griffin, C.G., Hozalski, R.M., 2020. Regional measurements and spatial/temporal analysis of
CDOM in 10,000+ optically variable Minnesota lakes using Landsat 8 imagery. Sci. Total Environ. 724, 138141 https://doi.org/10.1016/j.
scitotenv.2020.138141.
Ordoñez, S.J., 2010. Limnología del Embalse de Sau: relaciones del zooplancton, la clorofila y los sólidos en suspensión con el clima lumínico del agua. Universidad de
Barcelona [dissertation thesis].
Raymond, P.A., Saiers, J.E., 2010. Event controlled DOC export from forested watersheds. Biogeochemistry 100, 197–209. https://doi.org/10.1007/s10533-010-
9416-7.
Rochelle-Newall, E.J., Hulot, F.D., Janeau, J.L., Merroune, A., 2014. CDOM fluorescence as a proxy of DOC concentration in natural waters: a comparison of four
contrasting tropical systems. Environ. Monit. Assess. 186 (1), 589–596. https://doi.org/10.1007/s10661-013-3401-2.
Shoaf, W.T., Lium, B.W., 1976. Improved extraction of chlorophyll a and b from algae using dimethyl sulphoxide. Limnol. Oceanogr. 21, 926–928. https://doi.org/
10.4319/-lo.1976.21.6.0926.
Singh, S., D’Sa, E., Swenson, E., 2010. Seasonal variability in CDOM absorption and fluorescence properties in the Barataria Basin, Louisiana, USA. J. Environ. Sci. 22
(10), 1481–1490. https://doi.org/10.1016/S1001-0742(09)60279-5.
Slonecker, E.T., Jones, D.K., Pellerin, B.A., 2016. The new Landsat 8 potential for remote sensing of colored dissolved organic matter (CDOM). Mar. Pollut. Bull. 107
(2), 518–527. https://doi.org/10.1016/j.marpolbul.2016.02.076.
Sòria-Perpinyà, X., Delegido, J., Urrego, E.P., Ruíz-Verdú, A., Soria, J.M., Vicente, E., Moreno, J., 2022. Assessment of sentinel-2-MSI atmospheric correction
processors and in situ spectrometry waters quality algorithms. Rem. Sens. 14 (19), 4794. https://doi.org/10.3390/rs14194794.
Spencer, R.G., Bolton, L., Baker, A., 2007. Freeze/thaw and pH effects on freshwater dissolved organic matter fluorescence and absorbance properties from a number
of UK locations. Water Res. 41, 2941–2950.
Stedmon, C.A., Markager, S., Kaas, H., 2000. Optical properties and signatures of chromophoric dissolved organic matter (CDOM) in Danish coastalwaters. Estuar.
Coast Shelf Sci. 51, 267–278.

15
X. Sòria-Perpinyà et al. Remote Sensing Applications: Society and Environment 33 (2024) 101094

Vähätalo, A.V., Wetzel, R.G., 2004. Photochemical and microbial decomposition of chromophoric dissolved organic matter during long (months–years) exposures.
Mar. Chem. 89 (1), 313–326. https://doi.org/10.1016/j.marchem.2004.03.010.
Vodacek, A., Blough, N.V., DeGrandpre, M.D., Peltzer, E.T., Nelson, R.K., 1997. Seasonal variation of CDOM and DOC in the Middle Atlantic Bight: terrestrial inputs
and photooxidation. Limnol. Oceanogr. 42, 674–686. https://doi.org/10.4319/lo.1997.42.4.0674.
Wang, D., Ma, R., Xue, K., Loiselle, S.A., 2019. The assessment of Landsat-8 OLI atmospheric correction algorithms for inland waters. Rem. Sens. 11 (2), 169. https://
doi.org/10.3390/rs11020169.
Wetzel, R.G., 2001. Limnology. W.B. Saunders Company, Philadelphia.
Zhu, W., Yu, Q., Tian, Y.Q., Becker, B.L., Zheng, T., Carrick, H.J., 2014. An assessment of remote sensing algorithms for coloured dissolved organic matter in complex
freshwater environments. Remote Sens. Environ. 140, 766–778. https://doi.org/10.1016/j.rse.2013.10.015.

16