The case study of Dún Laoghaire Harbour in Ireland from 1925–1931 illustrates the challenges of accurately digitizing marigrams. The methodology involves digitizing the original marigram trace and the underlying grid at the nearest foot interval for each given hour to assess offsets. By subtracting the digitized foot interval from the known foot interval, differences are determined and then subtracted from each hourly trace value. This comparison between the original and adjusted data, where the upper and lower bounds were consistently aligned with the top line thickness during calibration, enhances the accuracy of final sea level data to the 10 mm level per year of data. The accompanying paper outlines the digitization and adjustment methodology, along with the compilation of key metadata. The attached metadata file in the dataset download explains the structure of the dataset, including its levels. The accurately digitized records from 1925 to 1931 can help us understand sea level changes over time.

There are three main levels of data and one folder file (Grid_Values) and a readme metafile (README-Dun_Laoghaire_Harbour.txt) describing all the data in detail. The dataset can be described as follows:i) Level_0_Raw_Images: This compressed zip file contains the Level_0_Raw_Images_Folder holding the raw marigram images in subfolders for each year (1925-1931), named by its respective year. Within each year's subfolder, there are two additional subfolders: Good_Quality_Images and Poor_Quality_Images, which separate the good and poorer quality images.ii) Level_1_Digitized_Data: This level (in a compressed zip file) consists of each year of digitized data only. It contains an Excel file for each year with columns for hour, height (units of height are feet relative to chart datum), and datetime. This is the raw digitized data only and has no adjustments made to the hourly intervals. Each year is named with _DL_Digitized.iii) Level_2_Adjusted_Data: This level (in a compressed zip file) consists of each year of adjusted data. The values were adjusted for offsets, where the nearest foot (interval) grid values were subtracted from the known feet values, and the offsets (differences) were subtracted from the digitized data accordingly at each hour. The units of the height column are feet relative to chart datum. It contains columns for hour, height, and datetime in Excel. This data is within 10 mm (1 cm) overall accuracy. Each year is named with _DL_Adjusted.iv) Grid_Values: This compressed zip file contains the Grid_Values. Within this compressed zip file, each yearly Excel file (named according to its year) includes the underlying grid values (nearest foot interval) digitized closest to each hourly trace value. The Excel file for each year comprise columns for hour, recorded foot interval (the foot interval recorded in WebPlotDigitizer), actual foot interval (the height in feet), offsets (the difference between the recorded and actual foot intervals), and datetime. Each year is named with _DL_Grid_Values. Despite not being one of the main levels, this data is vital as it accurately brings the Adjusted Data within 10 mm (1 cm) of overall accuracy.