Data to: A simple method to isolate fluorescence spectra from small dissolved organic matter datasets

a_rawdata.zip: Raw fluorescence data
Processing until here
1. Fluorescence data were measured on an Horiba AquaLog
- Fluorescence was measured as signal/reference beam
- Dark measurements were subtracted
- Internal excitation- and emission-correction factors were applied
- Blanks (ultrapure water) with identical instrument settings were subtracted
2. Inner filter effects were corrected with the absorbance-based method.
- Absorbance was measured on the same instrument
3. Fluorescence was normalized to the area under the Raman peak at 351 nm
4. Samples taken in 2019 were interpolated to fit the wavelength settings of samples taken and measured in 2020.

About the data
1. Each csv file contains the wavelength information in nm as column and row headers (columns=excitation, row=emission) and the fluorescence observations processed as described above.
2. Each file is named with its sample identifier.
3. The file "metadata_rawdata.csv" contains information on the sample type and all other useful information for the description of the samples detailed in the main text.
4. The zip-file contains a *.mat (MATLAB) file that holds data equivalent to the csv-files in a format that is compatible to the drEEM toolbox (dreem.openfluor.org)


b_processeddata.zip: Processed fluorescence data
Processing until here
1. Start: Raw fluorescence data
2. Rayleigh and Raman scatter was removed and replaced with missing numbers ("NaN")
3. Data 25nm below the calculated 1st order Rayleigh peak was zero'ed.
4. Negative fluorescence data was zero'ed.
5. After the excision of scatter, excitation and emission wavelengths were increased by 3nm. Some wavelengths were removed from the data.
6. Some data was replaced with "NaN" due to abnormal character (outliers)
7. The fluorescence data of each sample was divided by the 3/2th root of its standard deviation.

About the data
1. Each csv file contains the wavelength information in nm as column and row headers (columns=excitation, row=emission) and the fluorescence observations processed as described above.
2. Each file is named with its sample identifier.
3. The file "metadata_processeddata.csv" contains information on the sample type and all other useful information for the description of the samples detailed in the main text.
4. The zip-file contains a *.mat (MATLAB) file that holds data equivalent to the csv-files in a format that is compatible to the drEEM toolbox (dreem.openfluor.org)

c_PARAFACmodels.zip: Fluorescence PARAFAC models
Model fitting
1. Start: Processed fluorescence data.
2. Models were obtained with the following options
- Starts: 50
- Convergence: 1e-8
- Constraints: nonnegativity in all modes
- Initialization: random orthogonolized numbers

About the data
1. Each *.ods (OpenDocument Spreadsheet) file contains three spreadsheets with scores and loadings of each PARAFAC model.
2. For information on filenames, please refer to the file "metadata_modelscores.csv"
3. The zip-file contains a *.mat (MATLAB) file that holds data equivalent to the *.ods-files in a format that is compatible to the drEEM toolbox (dreem.openfluor.org)