Total organic carbon (TOC) is a basic index indicating the total amount of organic matter in water by carbon content, while chemical oxygen demand (COD) measures reducing matter in water samples by chemical method. Generally, the TOC is measured by a TOC analyzer. After removal of inorganic carbon by acidification (pH < 2), the organic carbon remained in the sample is converted to CO₂ via catalytic combustion and quantified by a non-dispersive infrared detector (Mu et al., 2016). COD measures the oxygen equivalent consumed by oxidizing organic matter to CO₂ with a strong oxidizing agent (potassium permanganate or dichromate) under acidic conditions. The average oxidation state of carbohydrate carbon in DOM is about 0. The average oxidation state of carbon in proteins and humic acids usually ranges from 0 to −2 (Stumm and Morgan, 1996). The average oxidation state calculated by TOC and COD can obtain information such as the occurrence and valence of organic matter in water samples (Eq. 1), which can be used in the study of DOM component migration and transformation in alpine water environment and regional carbon cycle (Stumm and Morgan, 1996; Mu et al., 2016; Rodriguez-Cardona et al., 2020). Oxidation State = 4 * TOC − COD TOC (1)

Total organic nitrogen (TON) is a basic index indicating the total amount of organic matter in water by nitrogen content, which is usually measured using a TON analyzer (combustion oxidation-chemiluminescence method). Total phosphorus (TP) is the sum of various chemical forms of granular and dissolved phosphorus in water, including orthophosphate, condensed phosphate, pyrophosphate, metaphosphate, and organic group combined phosphate. The water sample is usually dissolved with potassium persulfate or nitro-perchloric acid as an oxidant and then tested by spectrophotometry. N and P in water are derived from microbial metabolism, soil dissolution, anthropogenic sources, etc. (Qu et al., 2018). These processes are closely related to the formation and release of aquatic organic matter. The retention time of different molecular weight components in size-exclusion chromatography is different. Refining the molecular weight-related C and N composition by liquid chromatography combined with organic carbon and organic nitrogen detector (LC-OCD-OND) can potentially help explore the changing trend of DOM (Huber et al., 2011).

Figure 1 illustrates the principle and classification of spectral methods, ranging from ultraviolet and visible to infrared analysis. The UV-vis spectrum can detect the conjugated double bond structure, such as benzene ring and amide, generating n or π electron excitation. Different components in DOM have various molar absorption coefficients, and the absorbance follows Lambert-Beer’s law. On the peptide bond backbone of protein-like components, the carbonyl has π→π* transition at 190 nm and the peptide bond has n→π* transition at 210–220 nm. On the peptide side chain, the phenylalanine benzene ring, tyrosine phenolic group, and tryptophan indole group exhibit π→π* transitions at 255–270 nm, 270–285 nm, and 285–305 nm, respectively. The nucleic acids absorb light at around 260 nm due to π→π* transition of the bases (Rodger, 2013). Humic substances exhibit a broad spectrum of absorption from UV to visible range with relatively high absorbance at 250–300 nm, which could be attributed to the presence of multiple aromatic groups and complex inter-chromophore interactions (Del Vecchio and Blough, 2004).

Empirical secondary parameters (ratios of absorbance at different wavelengths) such as molecular weight index A₃₆₅/A₂₅₀, humification index A₄₀₀/A₃₀₀, hydrophobicity index A₂₅₀/A₂₀₄, aromaticity index A₂₁₀/A₂₅₄, and nonpolar index A₂₂₀/A₂₅₄ can be obtained by calculating absorbance data (Peuravuori and Pihlaja, 1997; Erlandsson et al., 2012; Al-Juboori et al., 2016). The larger these indices, the stronger their corresponding properties. In addition, the absorbance ratio can be obtained by dividing the absorbance by the TOC concentration, such as the specific UV absorbance at 254 nm (SUV₂₅₄) (Weishaar et al., 2003), representing aromaticity. Compared with the total amount analysis method, the UV-vis spectroscopy has the advantages of fast detection speed (consuming a few to 10 minutes), low detection limit (mg L^-1 level), and small sample volume (a few to 10 mL), which can meet the needs of DOM characterization in alpine water (Yan et al., 2016; Song et al., 2019). In addition, the sensitivity of DOM detection in alpine water can be improved by sample concentration or by using appropriate cuvettes to enlarge optical path difference. The UV-vis spectral analysis based on characteristic absorbance and secondary parameters can offer a quick glance at the composition and content of DOM in alpine water, and the detailed properties of DOM components can be cross-examined by other methods.

Fluorescence excitation-emission matrix (FEEM) detects the fluorophoric portion of DOM by exciting the electrons through n→π or π→π* transition, which then experience vibrational relaxation or internal conversion and emit fluorescence through π*→n or π*→π transition back to the ground state. Substances with high fluorescence intensity and fluorescence quantum yield often have structures of conjugated π bonds, rigid planes, heterocycles, and/or electron donating groups (-OH or -CN group). Substances with asymmetric structures, such as tryptophan/tyrosine in proteins and phenols in humus, have distinctive fluorescence properties (Peuravuori and Pihlaja, 1997; Mostofa et al., 2009). The quenching effect of halogen and metal ions in water samples may reduce the fluorescence intensity, and the pH and temperature during the detection may also change the properties of organic matter and affect the fluorescence intensity.

In the FEEM test, fluorescence signals can be scanned in the excitation wavelength (Ex) range of 200–650 nm and the emission wavelength (Em) range of 200–800 nm. For DOM identification, the occurrence of fluorescence in different excitation/emission (Ex/Em) wavelength positions indicates the occurrence of different DOM components (Supplementary Table SA1). At a given wavelength position the fluorescence intensity indicates the relative abundance of the corresponding DOM component. Protein-like substances usually exhibit fluorescence at a relatively low Em (<380 nm), with aromatic proteins having a lower Ex (<250 nm) than microbial byproduct-related proteins (250–360 nm) (Chen et al., 2003; Yu et al., 2020). Humus-like substances usually have fluorescence at a medium Em (380–520 nm), with the fulvic acid-like and humic acid-like subset excited at low and medium Ex (<250 nm and 250–420 nm), respectively (Chen et al., 2003; Mladenov et al., 2007). In contrast, the fluorescence of biological pigments (such as chlorophyll a, phycocyanin and phycoerythrin) occurs in notably high Em (>550 nm) regions with varied Ex from 400 to 630 nm (Nebbioso and Piccolo, 2013).

Rich information can be obtained from FEEM by different data analysis methods, including fluorescence regional integration (FRI), parallel factor analysis (PARAFAC), and two-dimensional correlation spectroscopic analysis (2DCOS). FRI divides the scanning matrix into five wavelength regions and calculates the contribution rate of fluorescence intensity in each wavelength region to the total fluorescence, which can be used to distinguish humic acid-like and protein-like components in alpine waters (Li YT. et al., 2021). The derived parameters of FRI include the hydrophobicity parameter, humification parameter, wavelength parameter, and Stokes shift parameter related to fluorescence energy (Xiao et al., 2020). PARAFAC can extract principal components from complex DOM and trace their spatiotemporal variations accordingly (García et al., 2020). The 2DCOS can spread the spectral changes caused by external disturbances to two dimensions, and explore the sensitivity relationship between spectral signal changes and external factors such as altitude, location, time, and meteorological conditions (Noda and Ozaki, 2004; Chen et al., 2019; Li YT. et al., 2021). FEEM has the advantages of high sensitivity (μg L^-1 level), small sample amount (less than 2–5 mL), fast measurement speed (consuming a few to 10 minutes), and rich information. The characteristics of microsample but high sensitivity can well fit the needs of alpine water analysis. Therefore, FEEM has been widely used in exploring DOM components in alpine waters.

Fourier transform infrared (FT-IR) spectroscopy detects specific structures with dipole moments because their molecular vibrations and rotations are related to infrared light absorption. By analyzing the absorption wavenumber of the spectrum, information of chemical bonds (functional groups) corresponding to different molecules can be obtained. For example, characteristic wavenumbers useful for DOM analysis include 3,200–3,600 cm^-1 (hydrogen bonds), 2,850–2,960 cm^-1 (-CH), 1,690–1760 cm^-1 (carbonyls), 1,600–1,690/1,480–1,575/1,260–1,300 cm^-1 (amide I/II/III in proteins), ∼1,100 cm^-1 (C-O in polysaccharides), and ∼1,030 cm^-1 (C-O-C in polysaccharides or humus) (Krimm and Bandekar, 1986; Pretsch et al., 2011; Xiao et al., 2022). FT-IR is fast and non-destructive to the sample. FT-IR can obtain the molecular structure of DOM, but the characteristic peaks overlapping caused by the coexistence of multiple functional group absorption peaks often appear in the analysis of DOM samples. This method has not been frequently used for characterization of DOM in alpine waters.

Charged ions with different charge-mass ratios can be detected by mass spectrometry (MS). The inorganic salts in alpine water samples greatly challenge the analysis of ultra-low DOM concentration (Pan et al., 2023). Generally, the desalting, separation, and concentration of DOM components are prerequisites for DOM characterization. DOM can usually be extracted by ultrafiltration, freeze-drying, reverse osmosis combined with electrodialysis, or solid phase extraction methods (Kim et al., 2022). Fourier transform ion cyclotron resonance (FT-ICR) MS can deeply analyze the chemical structure of DOM in natural environments and water treatment processes with ultra-high mass resolution (the m/Δm ratio reaching 10⁴–10⁶) and sub-ppb level mass accuracy. It is suitable for accurately analyzing molecular weights and chemical compositions of various structural fragments in the complex DOM macromolecules in alpine water (He C. et al., 2022).

The full procedure for FT-ICR MS analysis is illustrated in Supplementary Figure SA1. Electrospray ionization can produce ions of various valence states and has high selectivity for polar acidic functional groups; however, it has seldom been used to study DOM (Persson et al., 2005). The FT-ICR MS results can be presented with a van Krevelen diagram to show the differences in DOM components from different sources, and the evolution of specific structures in DOM in different environmental processes can be found through molecular comparison (Pan et al., 2023). Each test requires an equivalent of 100–200 μg TOC mass. FT-ICR MS has high resolution and can provide a huge amount of information; however, its preprocessing is complex and the test is relatively expensive. Moreover, ionization of the complex DOM is often variable, likely to produce unrepeatable results. Quantitative analysis of DOM at molecular level based on FT-ICR MS is a significant but challenging task in future alpine water research.

Nuclear magnetic resonance (NMR) is a physical phenomenon of nuclear absorption and re-emission of electromagnetic radiation in the magnetic field. Before testing, the sample should be pretreated to ensure it does not contain paramagnetic substances and insoluble impurities to eliminate the resolution error (Emery et al., 2001; Hertkorn et al., 2016). NMR is highly selective and provides detailed picture information of the molecular structure of solid and liquid phase samples to distinguish the different connections of functional groups within the DOM molecule (Mao and Schmidt-Rohr, 2003). Measurement of the ¹H spectrum in conventional continuous wave methods requires more than 50 mg of samples. When testing the ¹³C spectrum, measuring the Fourier transform (FT) spectrum is necessary. The FT ¹³C spectrum requires 50–100 mg samples and the FT ¹H spectrum requires 1–10 mg samples (Simpson et al., 2011). NMR can distinguish multiple atoms in a single molecule or collections of molecules of the same type. However, since DOM bears complex compositions, there could be considerable overlap of signals in the NMR spectrum. NMR has been rarely used in the alpine water DOM analysis.

Chromatographic separation can achieve selective enrichment through the difference of charged properties. Combining chromatographic separation with optical detector and mass spectrometry can effectively characterize the DOM components in alpine water (Sandron et al., 2015; Maurischat et al., 2022). LC-OCD-OND requires no complex pretreatment, while DOM samples generally need to be filtered through a 0.22 µm mixed cellulose ester membrane. An LC-OCD-OND test often requires more than 50 mL of a sample with a TOC sample concentration of less than 5 mg L^-1. LC-OCD-OND can be applied to characterize natural organic matter to indicate the properties and changes of substances through different molecular weights (Huber et al., 2011). Currently, the coupling of chromatography and detector based on molecular weight and hydrophobicity has been widely used in non-alpine water analysis, showing great potential in the DOM analyses of alpine water.