Characterization and spatiotemporal variations of fluorescent dissolved organic matter in leachate from old landfill-derived incineration residues and incombustible waste

Dissolved organic matter (DOM) influences the bioavailability and behavior of trace metals and other pollutants in landfill leachate. This research characterized fluorescent dissolved organic matter (FDOM) in leachate from an old landfill in Japan during a 13-month investigation. We employed excitation–emission matrix (EEM) fluorescence spectroscopy with parallel factor analysis (PARAFAC) to deconvolute the FDOM complex mixture into three fluorophores: microbial humic-like (C1), terrestrial humic-like (C2), and tryptophan-like fluorophores (C3). These FDOM components were compared with findings from other studies of leachate in landfills with different waste compositions. The correlations among EEM-PARAFAC components, dissolved organic carbon (DOC) concentration, and ultraviolet–visible and fluorescence indices were evaluated. The FDOM in leachate varied spatially among old and extended leachate collected in the landfill and leachate treatment facility. The FDOM changed temporally and decreased markedly in August 2019, November 2019, and April 2020. The strong positive correlation between HIX and %C2 (r = 0.87, ρ = 0.91, p < 0.001)) implies that HIX may indicate the relative contribution of terrestrial humic-like components in landfill leachate. The Fmax of C1, C2, and C3 and the DOC concentration showed strong correlations among each other (r > 0.72, ρ > 0.78, p < 0.001) and positive correlations with leachate level (r > 0.41, p < 0.001), suggesting the importance of hydrological effects and leachate pump operation on FDOM.


Introduction
In European countries and Japan, incineration has become the prevailing intermediate treatment for MSW since the 1970s [1][2][3].Incineration residues include bottom ash and fly ash, which are ultimately disposed of in sanitary landfills [2].Dissolved organic matter (DOM) in landfill leachate is a complex mixture, which can originate from incineration residues, incombustible waste combined with microbial activities in landfills, and soil covering landfills [2,[4][5][6][7].However, recognizing specific properties of DOM and tracing its origins in landfill leachate is challenging due to significant variations in the composition of solid waste, which serves as a source of DOM at landfill sites [4].The characteristics of DOM are considered to exhibit variations among sampling locations, potentially influencing the bioavailability, toxicity, and migration of trace metals and other pollutants to the surrounding environment [8,9].He et al. [10] emphasized the importance of treating DOM in landfill leachate for safe disposal.Old landfill sites will ultimately be abolished once decomposition has stabilized after a monitoring period [11].Therefore, there is a need for a practical understanding of the variations in DOM within landfills, to assist in decisions to landfill abolition.
Excitation-emission matrix (EEM) fluorescence spectroscopy has been used widely to characterize and monitor fluorescent dissolved organic matter (FDOM) in various environments, including marine water, freshwater, and wastewater [9,[12][13][14][15], owing to its relative simplicity, high sensitivity and specificity, and rapid analysis.Nonetheless, the EEM spectra of DOM cannot be easily interpreted due to overlaps in the spectra of different fluorophores [16].Various methods have been developed to characterize the fluorescence datasets of organic matter, such as the peak-picking technique, fluorescence regional integration (FRI), parallel factor analysis (PARAFAC), principal component analysis (PCA), and self-organizing maps (SOMs) [14,17,18].Among these methods, PARAFAC is a popular multivariate modeling technique that deconvolutes the EEM spectra of FDOM into distinct fluorescent components to quantify, relatively, each component [19].Although DOM characterization using EEM-PARAFAC has been widely reported in various environments [9,[12][13][14][15], including landfill leachate, most studies have investigated direct landfills containing unincinerated MSW, or a combination of MSW and incineration residues [9,17,20].In previous studies, FDOM was deconvoluted into 2 to 6 distinct components, emphasizing the differences in the number of PARAFAC components and their assignments among various landfills.In traditional Chinese landfills, including unburned waste, four EEM-PARAFAC-derived components were assigned, comprising one fulvic-like substance, two humic-like substances, and one protein-like compound [17].In another Chinese sanitary landfill site, four humic-like and two protein-like components were identified by PARAFAC [9], while leachate FDOM was deconvoluted into only one humiclike and one tryptophan-like component in a study from a stabilized Korean landfill [21].By contrast, the DOM in the leachate of landfills containing only incineration residues and incombustible waste remains poorly understood.Most investigations on spatiotemporal variations of FDOM have mainly focused on rivers [22][23][24] and lakes [25,26] rather than landfill leachate.Moreover, reports on DOM in landfills have focused mainly on improving leachate treatment [14,27].Only Lu et al. [28] have reported on the organic matter in leachate at multiple sites within a single landfill.Overall, studies on the spatial distributions have been conducted in relatively new landfills operating for only 3-15 years; the behavior of organic matter in older landfills (e.g., operating for 40 years) has not been reported.
In addition, the humification index (HIX) and specific ultraviolet absorbance at 254 nm (SUVA 254 ) have been commonly used as indicators of the humic substance content, and the humification degree and aromaticity of DOM [29,30].SUVA 254 was calculated dividing the absorption coefficient at 254 nm (cm -1 ) by DOC concentration (mg C/L), and then multiplying by 100 (cm/m) [29], while HIX was calculated by dividing the area under the Em spectra 435-480 nm by the peak area 300-345 nm + 435-480 nm, at Ex 254 nm [30].Numerous studies have shown positive correlations between the HIX, SUVA 254 , and humic-like components derived from EEM-PARAFAC [31,32].Understanding humic substances or humus content and humification degree of organic matter is crucial for characterizing the landfill waste stability, the decomposition process and selecting suitable leachate treatment [33,34].However, He et al. [33] suggested that several humification indices showed no correlation with the F max of humic-like substances and were not suitable for characterizing humification.In the present study, we assessed whether the indicators HIX and SUVA 254 correlated with any EEM-PARA-FAC components and fluorescence parameters in landfill leachate.
In light of these gaps in the literature, our study aims to address the following research questions: What are the specific characteristics and variations of FDOM in the leachate of an old municipal landfill containing incineration residues and incombustible waste, which had been landfilled for 25-44 years?How do these characteristics compare with existing studies on landfill leachate?What are the relationships among fluorescence components, spectroscopic parameters, and dissolved organic carbon (DOC) concentration?
To address the aforementioned research questions, PARAFAC was applied to deconvolute the fluorescence signals into the underlying individual fluorescent components.Subsequently, these components were compared with findings from other studies of landfill leachate.The characterization and variations of EEM-PARAFAC components were evaluated alongside results derived from the FRI method, serving as a comparative method.Furthermore, the spatial and temporal variations in FDOM in leachate across the study site were investigated, and leachate DOM was assessed before and after treatment.Correlation and multilinear regression analysis were employed to understand the relationships among fluorescence components, spectroscopic parameters, and DOC concentration.Finally, the utility of indicators such as HIX and SUVA 254 , based on their correlation with humic-like components derived from EEM-PARAFAC, was evaluated and discussed.

Landfill site and sample collection
The landfill site, located in a city in Japan, contained approximately 75% of incineration residues (bottom ash and fly ash) and 25% of incombustible waste consisting mainly of concrete, asphalt, glass, and ceramics [3].Fig 1 presents a schematic overview of the landfill, leachatecollecting pipe system, and monitoring wells.In the landfill, we installed sampling devices containing a smart groundwater purging pump (Daiki Rika Kogyo, Saitama, Japan) at six monitoring wells to monitor leachate water quality.Over 13 months, from June 2019 to June 2020, 130 leachate samples were collected monthly at 10 sampling points.Polypropylene (PP) plastic bottles (Sanplatec, Osaka, Japan) were thoroughly cleaned and rinsed with Milli-Q water, followed by three additional rinses with the samples, prior to their use for sample collection.Six samples were collected in the monitoring wells (G1.1, G1.2, G2, G3, G4, and G5) located in five sections that covered most of the landfill area.The landfill operation periods varied among sections: G1.1 and G1. 2 (1976-1985), G2 (1977-1986), G3 (1985-1989), and G4 (1987)(1988)(1989)(1990)(1991)(1992)(1993) from the old disposal area, and G5 (1994)(1995) from the extended disposal area.Four additional leachate samples (L1, L2, L3, and L4) were collected at the leachate treatment facility.L1 was leachate collected from a pit located at the end of sections 4 and 5 in the old landfill area.L2 consisted of leachate collected from the extended landfill area, including G5. L3 comprised a mix of the untreated leachate downstream of L1 and L2.Finally, L4 was the leachate posttreatment.All samples were transported in iceboxes to the laboratory as quickly as possible.

DOC analysis
To prevent contamination from organic matter, pre-combusted glass fiber filters (Whatman GF/B, 1 μm; Whatman, Maidstone, UK) were used to separate solid and dissolved matter [35].
The DOC concentration was measured as non-purgeable organic carbon using the TOC-V CSH analyzer (Shimadzu, Kyoto, Japan).The detection limit (LOD) and quantification limit (LOQ) were 0.03 mg C/L and 0.1 mg C/L, respectively.The DOC concentration of each leachate sample was determined by averaging measurements conducted at least twice until the coefficient of variation (CV) was < 2%.

Spectroscopic analysis
Ultraviolet-visible absorbance scans were performed from 600 to 250 nm using a 1-cm quartz cuvette and UV-1280 spectrophotometer (Shimadzu) with a wavelength step size of 1 nm before fluorescence measurement to correct for inner filter effects (IFEs).The inner filter correction procedure is described in S1 Text of S1 File.Samples were allowed to stand at room temperature (20˚C) before the spectroscopic measurements.Fluorescence measurements were conducted using the RF-5301 PC spectrofluorophotometer (Shimadzu) and a quartz cuvette.The scanning ranges were 250-450 nm (5-nm increments) and 250-600 nm (5-nm increments) for the excitation and emission wavelengths, respectively, with a slit width of 5 nm for both excitation and emission monochromators and 150 W xenon lamp operated in the very fast wavelength speed mode.All samples were acidified to pH 2.0 ± 0.1 using HCl for both DOC and spectroscopic analysis.The fluorescence intensities were normalized to Raman units using the signal intensity of the Raman peak of Milli-Q (18.2 MO) water, which was also adjusted to pH 2.0 ± 0.1 using HCl at excitation/emission wavelengths (Ex/Em) of 350/350-450 nm.The relative precision of fluorescence intensity measurements was <2% based on repeated measurements.

Data analyses
PARAFAC.The EEMs were modelled by PARAFAC [12,36] in the MATLAB software 2018b environment (MathWorks, Natick, MA, USA).The principle and formula of PARAFAC are presented in S2 Text of S1 File.PARAFAC was conducted on 126 samples collected at 10 sites in the landfill and wastewater treatment facility from June 2019 to June 2020, using the drEEM toolbox (version 0.6.0)based on the tutorial of Murphy et al. [12].Four EEMs spectra were removed from the initial 130 EEMs spectra because one sample had an abnormal fluorescence intensity, and three samples had fluorescence intensities exceeding the upper maximum limit of 1000 AU [37].
In the procedure for PARAFAC analysis of EEMs (S1 Fig in S1 File), we first collected and pre-processed the data.Pre-processing consisted of EEM assembly, IFE correction, Raman normalization, scatter removal, interpolation, and signal normalization.Next, preliminary models of different components were run, outliers were removed, and model fits were evaluated.The FDOM components were identified and validated using jack-knifing, residual analysis, the core consistency diagnostic, and split-half validation [12,38,39].All three validation approaches (residuals and loadings analysis, core consistency, and split-half analysis) indicated that the three-component model was optimal.Finally, the results were obtained as the maximum fluorescence intensity, x ijk , and excitation and emission loadings for each sample.The detailed procedure and results are shown in S3 and S4 Texts, S1 Correlation analysis and multilinear regression analysis.The samples collected at 10 sampling sites from June 2019 to June 2020 were used for correlation analysis (sample size, n = 126).The maximum fluorescence intensity (F max ) and the relative percentage of each component derived from EEM-PARAFAC (%C1, %C2, %C3), fluorescent peak ratios (C:M, C:T, C:A, A:T, M:T) [40] were used as the input data for the correlation analysis.Several optical indices were added, including the specific ultraviolet absorbance at 254 nm (SUVA 254 ) [29], fluorescence index (FI) [41,42], humification index (HIX) [30], biological index (BIX) [43], and DOC concentration.The detailed descriptions of these optical indices are shown in S2 Table of S1 File.Both Pearson's r values and Spearman's rho were evaluated in correlation analysis because most of the data were not normally distributed [44].Multilinear regression analysis was used to predict the value of a dependent variable DOC concentration based on the values of independent or predictor variables (F max values of f C1, C2 and C3).These analyses were conducted on R version 4.3.3.
Fluorescence regional integration (FRI).The FRI method divides the EEM spectra into two regions and characterizes FDOM by calculating the volume of each region [37,45].A detailed method description is presented in S5 Text of S1 File.The characterizations and variations of FDOM derived from the EEM-PARAFAC method were compared with those obtained from the FRI method.
All experimental and computational analyses described in this manuscript were conducted at Katsura Campus, Kyoto University in Japan.The peak positions of these components were compared with those in studies of FDOM spectra in other landfill leachate samples in China, Korea, and Belgium (Table 1).The waste compositions of some of these landfills consisted of MSW, sewage sludge, solidified sludge, domestic waste, and construction waste; however, the previous studies typically did not provide detailed information on the type of waste or type of landfill site.In previous research, leachate samples were collected in the landfill or during different stages of MSW management.By contrast, the waste composition of the landfill in the present study comprised mainly incineration residues and incombustible waste.Research on FDOM in the leachate of landfills containing primarily incinerated MSW remains poorly understood.A component similar to C1 in this research was reported as a fulvic acidlike substance related to aromatic and aliphatic groups of leachate-derived DOM [17,20,27] or microbial humic-like components derived from microbial activity in landfills [21,46,47].Oloibiri et al. [14] assigned compound C1 as humic-and fulvic-like components, with the first peak at 250/425 nm representing humic substances and the second peak at 320/425 nm representing fulvic-like compounds in DOM.However, the peak emission wavelength of C1 in this research (at 395 nm) was shorter than the values from other studies of leachate FDOM (400-425 nm).Only Zhang et al. [47] observed a similar peak position as C1 for the emission wavelength (390 nm); however, the excitation wavelength differed (280, 330 nm) compared with our findings (< 250, 320 nm).Similar component of C2 in this study were assigned as humiclike component [9,17,20] or terrigenous humic-like component [47] or terrestrial humic-like component [21,48].Similar component of C3 were assigned as protein-like component [9,17] or tryptophan-like component [20,27,46].

EEM-PARAFAC model and comparison with other studies
The OpenFluor database of PARAFAC aquatic fluorescence components (https://openfluor.lablicate.com/;[18]) was used to find matching peaks of EEM-PARAFAC components between the results from this study and those from other studies of a variety of aquatic environments.Matching peaks represented by excitation and emission minimum spectral scores > 0.95 were considered to be similar to the published spectra (S3 Table in S1 File).The fluorescence spectra of C1, C2, and C3 matched 75, 89, and 52 results, respectively, from other studies in the Open-Fluor database.The components similar to those in this study originated from various aquatic environments, including rivers [49], streams, groundwater, lakes [50], estuaries, seawater [51], freshwater, aquacultures [52], catchments, wetlands [53], and soil [54]; however, there was a lack of information on FDOM in landfill leachate.Most studies assigned C1 as a marine humiclike or microbial humic-like component, C2 as a terrestrially derived humic-like component, and C3 as a protein or tryptophan-like component.These findings suggest the ubiquitous presence of certain DOM fluorescence spectra across various environments [55] and reinforce the reliability of the three-component PARAFAC model in this study.A large traditional landfill [17] (Continued ) Ultimately, the three components were assigned: C1 (Ex/Em < 250, 320/395 nm), microbial humic-like peak M [37,56]; C2 (Ex/Em < 250, 350/450 nm), a mixture of terrestrial humiclike peaks A and C [37,57]; and C3 (Ex/Em 280/345 nm), a tryptophan-like peak T [28,37].The DOC concentration of leachate ranged from 2.6 to 38.2 (mg C/L), and the SUVA 254 ranged from 1.0 to 5.8 (L/mg C�m) (S4 Table in S1 File).Higher DOC concentrations were observed in the samples collected from the old disposal area, including G1.1 (12.6 ± 2.0 mg C/ L), G1.2 (15.3 ± 8.2 mg C/L), and G3 (18.4 ± 2.1 mg C/L).Compared with extended disposal area sample G5 (DOC concentration, 5.6 ± 1.8 mg C/L), the F max values of C1 and C3, along with the BIX, indicated a greater influence of microbial and biological activities on FDOM in the old disposal area.Among the sampling sites, G2, G4, and G5 had higher HIX values and lower BIX and FI values (Fig 5).G4 and G5 also had the highest SUVA 254 values (4.4 ± 0.3 and 4.0 ± 1.2 L/mg C�m, respectively), revealing a higher fraction of aromatic humic substances in the FDOM at these sites (S4 Table in S1 File).Sample G4 had a higher C2 content but lower C1 and C3 contents, indicative of a prominent contribution of terrestrial humic-like components to FDOM.Moreover, samples G2, G4, and G5 had relatively lower C1 and C3 contents.At the leachate treatment facility, the DOM characteristic of sample L1 differed from those of L2, L3, and L4.Site L1, in which leachate was collected from pump wells in the old disposal area and leachate-collecting pipes from sections 4 and 5, had a higher average concentration and variation in the DOC concentration compared with L2, L3, and L4.This was likely the influence of the higher C1 and C3 concentrations in sample G3 from the old disposal area.Overall, the FDOM characteristics differed among sampling sites, in particular between the  The leachate treatment method was assessed, with a specific focus on the dissolved organic matter (DOM) characteristics of leachate samples L3 (untreated) and L4 (treated).The F max values of components C1, C2, C3, and DOC concentration, were 36.9RU, 26.4 RU, 22.1 RU, and 9.26 mg C/L respectively in L3.After treatment (L4), the F max values of C1 and C2, and DOC conc.statistically decreased (Student's t-test, 95% confidence level, n = 13 for each parameter) with averaged removal efficiencies over 13 months between L3 and L4 were 8.57%, 7.20%, and 6.4% respectively; however, the F max of C3 was not statistically different after treatment.During the leachate treatment, polyaluminum chloride (PACl) was added as a coagulant to the coagulation-sedimentation process.The coagulation-sedimentation was found to readily remove hydrophobic organic matter with high molecular weight and aromaticity compared to low molecular weight and hydrophilic substances [11,60].The presence of weakly acidic functional groups, such as carboxylic groups and phenolic groups, in humic substances may lead to their weak anionic polyelectrolyte-like properties, which, in turn, facilitate the process of flocculation [11,61].The analysis of FDOM variations revealed that the F max values of components C1, C2, C3, and DOC concentration in old landfills (G1.1, G1.2, G2, G3, G4), which were operated from 1976-1993, were higher than those observed in extended landfill (G5) operated from 1994-1995 (3.0, 2.8, 3.9, and 2.5 times, respectively).However, compared with the previous study [62] which demonstrated substantial changes in FDOM and ranges of DOC concentration depending on landfill ages, our research revealed relatively low levels of FDOM values and DOC concentrations across 10 different sampling points and 13 sampling months despite the landfill site's age 25-44 years.Considering the treatment of leachate DOM, these findings suggested it was not necessary to apply complicated or advanced leachate treatment processes.It may be possible to reduce monitoring and treatment activities, especially for leachate treatment in extended landfills, which could help save leachate operational costs.

Correlation analysis and multilinear regression analysis
The results of Pearson's correlation and Spearman 's correlation analysis are presented in Fig 6 .The F max values of C1 and C2 were significantly correlated (r = 0.85, ρ = 0.72, p < 0.001, n = 126), presumably because both C1 and C2 have similar properties of the humic-like components or potentially originate from similar sources.The F max values of C3 and C1 were also significantly and positively correlated (r = 0.83, ρ = 0.86, p < 0.001).Hudson et al. [63] showed that tryptophan-like components may be present as "free" molecules or may be bound to proteins, peptides, or humic structures.He et al. [64] suggested that the protein-like components in landfill leachate showing a strong correlation with humic-like components were bound mainly to humic-like components with supramolecular assembly structures.Parlanti et al. [65] suggested that co-occurring protein-like peaks and β peaks (Ex/Em 310-320/380-420 nm) were derived from biological activity.Thus, the strong correlations between the F max values of C1 and C3 in the landfill leachate in the present study were indicative of an association with microbial or biological activity.This was strengthened by the positive correlations of the BIX with the F max values of C1 (r = 0.55, ρ = 0.68, p < 0.001) and C3 (r = 0.59, ρ = 0.65, p < 0.001).Moreover, the F max of C3 was significantly and negatively correlated with the HIX (r = −0.82,ρ = 0.78, p < 0.001).
In the present study, the SUVA 254 was not linearly correlated with the F max value of C1 or C2; however, it exhibited a weak positive correlation with %C2 (r = 0.44, ρ = 0.33, p < 0.001).This weak correlation could be due to the complexity of leachate and the variations in leachate characteristics at different sampling points.The presence of iron, colloids, and other constituents in samples may increase the UV absorbance at 254 nm [29,63].Broder et al. [66] showed that the DOM origin and dynamics could be tracked using different spectrofluoroscopic indices; however, this method could not display apparent seasonal or hydrologic dynamics in nonfluorescent fractions.The HIX was negatively correlated with F max values of C1 and C2, while it was positively correlated with % C2 (r = 0.87, ρ = 0.91, p < 0.001) and %C1(r = 0.26, ρ = 0.21, p < 0.05).Previous investigations consistently indicated strong positive correlations between the HIX and the relative percentage of terrestrial humic-like components [32,67,68], as well as microbially-derived humic-like substances [67,69], and fulvic acid-like substances https://doi.org/10.1371/journal.pone.0304188.g005[67].However, Lu et al. [67] reported a negative correlation between HIX and the proportion of microbial humic-like substances (r = −0.37,p = 0.03, n = 35).Gabor et al. [70] noted that fluorescence indices were originally developed for specific sample sets; therefore, they may be suitable only in certain environments, such as marine or soil environments.This could explain the weak correlation between HIX and %C1, which was presumed to originate from microbial activities in our study.Besides, HIX showed strong negative correlation with %C3 (r = −0.92,ρ = −0.91,p < 0.001) and strong positive correlations with A:T, M:T and C:T (r = 0.92, 0.85, 0.88, ρ = 0.91, 0.81, 0.84, all p < 0.001, respectively).He et al. [33] presented correlations among several humification indices and F max values of three PARAFAC components, concluding that several humification indices could not yield a full interpretation of the humic-like components in landfill leachate.However, our findings suggest that HIX and SUVA 254 should serve as indicators of percentage rather than F max value of terrestrial humic-like components in landfill leachate.
The DOC concentration was significantly, positively correlated with the maximum fluorescence intensities of C1 (r = 0.87), C2 (r = 0.89), and C3 (r = 0.78; all p < 0.001).Furthermore, the equation for the relationship between DOC concentration (conc.)and F max values of C2, and C3 can be expressed as follows: ðr ¼ 0:91; R 2 ¼ 0:83; p < 0:001; n ¼ 126; normalized dataÞ The variable F max value of C1 was removed from the equation to solve its multicollinearity with other independent variables based on the result of the variance inflation factor (VIF > 5).The result suggested humic-like component has a higher impact on the changes in DOC concentration than the protein-like component.The maximum fluorescence intensity of C2 and C3 could trace DOC concentration, which may contribute to the simplification of DOM measurement.There are few reports on the relationship between DOC and the F max of PARAFACderived components in landfill leachate, especially in Japan.However, researchers have investigated the relationship between fluorescence intensity and DOC concentration in freshwater ecosystems [63,[71][72][73] and marine environments [74].Different sources of DOM and DOC concentration-related fluorophores result in different gradients (i.e., the fluorescence intensity per gram of carbon) and correlation coefficients of the DOC-fluorescence intensity [71,72].The strongest correlation between fluorescence intensity and DOC concentration is typically observed at sites dominated by natural DOM that are less affected by anthropogenic activities [75].Moreover, a higher fluorescence per gram of carbon was reported in samples dominated by high-molecular-weight aromatic DOM, such as peat catchments and wetlands [71,72,75].In comparison, in the present study, we investigated a site highly influenced by anthropogenic activity, and we measured higher F max values for C1 and C3 than for C2, which had terrestrial humic-like peaks (S10 Fig in S1 File).S11-14 Figs in S1 File shows the relationships of the F max values of C1, C2, and C3 with the DOC concentration, as well as that between the UVA 254 and DOC concentration, among the samples.These correlations indicated that the temporal variations in the F max values of C1, C2, and C3 were affected by the same factor as the DOC and also highlight the importance of dilution effects.The F max of C2 (terrestrial humic-like component) had a higher positive correlation with the UVA 254 (r = 0.67, ρ = 0.71, p < 0.001) compared with the other components (S15 Fig in S1 File); however, this relationship was not strong enough to interpret all the changes in humic-like components in the dataset of landfill leachate FDOM.
Several factors may control the characteristics and quantities of FDOM.Stedmon and Cory [48] found that seasonal changes in FDOM characteristics in a catchment were driven by soil microbial activity and rapid drainage of precipitation.Precipitation data were collected at the leachate treatment facility, and seasonal variations in Fmax values of EEM-PARAFAC components and DOC concentration were analyzed; however, no direct statistical effect of precipitation or statistically significant differences among seasons were observed.Therefore, these data were not presented in this research.Since the landfill site was closed and covered by layers of soil, rainfall could penetrate through the soil cover to the waste layers and affect leachate volume.The leachate level was affected by precipitation and leachate pump operation in the landfill.The leachate level was measured from the surface of the leachate to the ground surface (S5 Table in S1 File).The correlation analysis revealed positive correlations of the leachate level with C1 (r = 0.57), C2 (r = 0.50), and C3 (r = 0.41; all p < 0.001) across the whole dataset of leachate collected in the landfill (n = 72) (S16 Fig in S1 File).These results indicated an influence of a hydrological factor on the variations in the F max values of C1, C2, and C3.For instance, higher leachate levels would be expected to dilute the leachate concentration, although this does not necessarily imply that less precipitation causes higher concentrations of C1, C2, and C3 because pump operation at the leachate treatment facility also affects the leachate level in the landfill.When the leachate level rises significantly during the rainy season, the pumped drainage system is operated to reduce inundation, altering the landfill leachate level.

Comparison results between EEM-PARAFAC and FRI methods
Furthermore, the findings derived from EEM-PARACFAC concerning the characterization, variations, and proportions of humic-like and protein-like components, as well as their correlations with HIX, BIX, and SUVA 254 , were subjected to a comparative analysis against the results obtained through fluorescence regional integration (FRI).Both approaches yielded comparable findings, allowing for a rough estimation of FDOM variations.However, EEM-PARAFAC exhibited advantages in resolving FDOM into distinct fluorescent components and mitigating peak overlap.Further details on these findings are provided in the S6 Text, S6

Conclusions
We evaluated the DOM characteristics in the leachate of an old landfill in Japan from June 2019 to June 2020 using EEM-PARAFAC and correlation analysis.The DOC concentration ranged from 2.6 to 38.2 mg C/L and the SUVA 254 from 1.0 to 5.8 L/mg C�m.EEM-PARAFAC identified three fluorescent components: C1, microbial humic-like compounds; C2, terrestrial humic-like compounds; and C3, tryptophan-like compounds.The FDOM of landfill leachate varied spatially among the old disposal area, extended disposal area, and leachate treatment facility.Temporal trends showed particularly pronounced changes in August 2019, November 2019, and April 2020.Leachate DOM was assessed before and after treatment, indicating lower levels of DOM compared to leachate from other landfills.The findings may contribute to selecting suitable leachate treatment methods and landfill site management strategies, especially when considering the increasing popularity of incineration for waste management.The characterization of DOM in leachate from landfills containing incineration residues is highly relevant in this context.Correlation analysis supported the possible origin of EEM-PARAFAC components related to microbial and biological activity or terrestrial matter.F max values of C1-C3 showed strong correlations with DOC, a novel finding in landfill leachate that may simplify DOM measurement.The strong positive correlation between HIX and %C2 suggests that HIX could serve as an indicator of the percentage of terrestrial humic-like components in landfill leachate.The findings help characterize FDOM and clarify its relationship with optical indices and DOC concentration in the leachate of a landfill containing incineration residues and incombustible waste, which is different from previous studies.The study may provide a useful reference when comparing DOM quality and FDOM properties in leachate in different contexts.
Our findings revealed that fluorescence spectroscopy is a useful, sensitive, and quick analytical technique for identifying and monitoring variations in different FDOM components in this specific type of leachate.Future work on fractional quantification and characterizations of leachate's FDOM based on molecular weight is necessary to enhance our understanding of the environmental behavior of leachate, given its heterogeneity.Hereafter, based on the spatiotemporal distribution of FDOM screened and demonstrated in this research, we plan to evaluate DOM at the molecular level, chemical structural analysis (e.g., Orbitrap-LC-MS/MS) and the effects of DOM on the behavior of toxic chemicals, based on the conclusions drawn from this study.

Fig 2
Fig 2 illustrates the contours of the three fluorescent components deconvoluted by PARAFAC: Component 1 (C1) at Ex/Em < 250, 320/395 nm, Component 2 (C2) at Ex/Em < 250, 350/450 nm, and Component 3 (C3) at Ex/Em 280/345 nm.The peak positions of these components were compared with those in studies of FDOM spectra in other landfill leachate samples in China, Korea, and Belgium (Table1).The waste compositions of some of these landfills consisted of MSW, sewage sludge, solidified sludge, domestic waste, and construction waste; however, the previous studies typically did not provide detailed information on the type of waste or type of landfill site.In previous research, leachate samples were collected in the landfill or during different stages of MSW management.By contrast, the waste composition of the

Figs 3 and 4
Figs 3 and 4 respectively show the spatial and temporal variations in the maximum fluorescence intensities of C1, C2, and C3, as well as the DOC concentrations, of the 10 sampling points from June 2019 to June 2020.The F max of each component is proportional to the concentration in the sample [59].Fig 5 depicts the variation in optical indices in the same datasets.The DOC concentration of leachate ranged from 2.6 to 38.2 (mg C/L), and the SUVA 254 ranged from 1.0 to 5.8 (L/mg C�m) (S4 Table in S1 File).Higher DOC concentrations were observed in the samples collected from the old disposal area, including G1.1 (12.6 ± 2.0 mg C/ L), G1.2 (15.3 ± 8.2 mg C/L), and G3 (18.4 ± 2.1 mg C/L).Compared with extended disposal area sample G5 (DOC concentration, 5.6 ± 1.8 mg C/L), the F max values of C1 and C3, along with the BIX, indicated a greater influence of microbial and biological activities on FDOM in the old disposal area.Among the sampling sites, G2, G4, and G5 had higher HIX values and lower BIX and FI values (Fig5).G4 and G5 also had the highest SUVA 254 values (4.4 ± 0.3 and 4.0 ± 1.2 L/mg C�m, respectively), revealing a higher fraction of aromatic humic substances in the FDOM at these sites (S4 Table in S1 File).Sample G4 had a higher C2 content but lower C1 and C3 contents, indicative of a prominent contribution of terrestrial humic-like components to FDOM.Moreover, samples G2, G4, and G5 had relatively lower C1 and C3 contents.At the leachate treatment facility, the DOM characteristic of sample L1 differed from those of L2, L3, and L4.Site L1, in which leachate was collected from pump wells in the old disposal area and leachate-collecting pipes from sections 4 and 5, had a higher average concentration and variation in the DOC concentration compared with L2, L3, and L4.This was likely the influence of the higher C1 and C3 concentrations in sample G3 from the old disposal area.Overall, the FDOM characteristics differed among sampling sites, in particular between the old and extended disposal areas of the landfill, and between the landfill and leachate treatment facility.At most sampling sites, the F max values of C1, C2, and C3 followed similar trends over time, with higher values in June 2019, October 2019, December 2019, March 2020, and May 2020 and lower values in August 2019, November 2019, January 2020, and April 2020 (Fig4).The temporal variation in FDOM was likely partially influenced by variations in precipitation and leachate pump operation, leading to changes in leachate levels.