Cleaning, standardization, and assessment of the accuracy and consistency of the Yellowstone National Park dataset (log book)

Abstract

Yellowstone National Park (YNP), Wyoming USA, contains over 10,000 geothermal features and 2 to 5 % of these features are geysers. Yellowstone has about half of the world’s geysers and the majority of YNP geysers are located in Upper geysers Basin. Beginning in 1970, details (time of eruption, height, duration, etc.) of about 25 geysers activities have been recorded in log books and later transcribed into an electronic dataset and posted on the park’s website. The data was collected by park rangers, visitors, and geyser enthusiasts, among others. The data collected by direct observation, camera, electronic, etc. The dataset contains a great deal of information that is relevant to scientists, educators and the public. However, the use of the dataset is severely limited without cleaning and standardization. Given the size, time span over which the data was collected, and the number of people involved in collecting the data, it’s inevitable that the data contains many inconsistencies. The dataset has been cleaned, standardized, reorganized in some parts and converted to a spreadsheet which makes the dataset much better suited for computations and analysis. The reorganization consists of two steps: step one was to remove text type information and extra information to a newly created column; and step two was to reorder the information in a set of records so that individual data entry is consistent with the column heading under which it should have been listed. The overall and monthly statistical summary of the data shows that interval and duration are both bimodal normally distributed, height is normally distributed and preplay display a Rayleigh type distribution. Comparison of the YNP and the electronic dataset was not feasible for all geysers and all variables; however, where it’s feasible such as the case for interval data, the two datasets are nearly identical.