Data management for the Cold Land Processes Experiment: improving hydrological science

Abstract

Abstract:The Cold Land Processes Field Experiment (CLPX) produced 37 data sets totalling about one terabyte in volume. This presented a considerable challenge for data managers. Two unique aspects of the CLPX data management process include participation of data specialists in field data collection and identification of one person, the 'data wrangler', to coordinate the acquisition of the entire collection of the experiment data. These unique aspects increased the quality and completeness of the overall CLPX collection. Data managers were guided by basic data management principles from the literature, including consideration of potential future users and applications and an imperative to keep data as simple and as flexible as possible. These principles guided data management decisions about data formats, data integration, data presentation, and documentation. This principle-based approach improved the usability and long-term viability of CLPX data. It is recommended that data managers be involved early in the planning of future large-scale field experiments and that data managers develop appropriate principle-based rubrics for guidance through persistent data-management challenges.

Evaluation of forest snow processes models (SnowMIP2)

“…Although much was learnt from the experimental design of these previous studies, strong efforts were made in SnowMIP2 to increase both the number of models and their diversity. The fact that 33 models (Table 1) participated in SnowMIP2 reflected a combination of: (1) enthusiasm and trust of the snow modeling community building on the success of previous studies, (2) flexibility in criteria required for model participation, for example, model purpose, adherence to Assistance for Land‐surface Modeling Activities (ALMA) conventions [ Henderson‐Sellers et al , 1995, 1993; Polcher et al , 2000], and (3) timely interactions between the intercomparison organizers, data providers and participants to aid the data management process [ Parsons et al , 2004]. As well as increased participation, it was important to increase the range of snow environments each model was evaluated against.…”

Designating user communities for scientific data: challenges and solutions

“…The Cold Land Processes Experiment included data management considerations in the design of the experiment, and data specialists participated in the actual field data collection. We found that this led to a more complete and consistent data set and potentially reduced the number of ambiguous data values by fifteen to twenty percent (Parsons, Brodzik & Rutter, 2004). In other words, by being directly involved in the experiment we reduced the uncertainty of the data collected.…”

Making data useful for modelers to understand complex Earth systems

“…I developed the initial sea ice product comparison page at NSIDC in response to requests from sea ice researchers like Dr. Holland (Parsons and Duerr 2005). I led the data management effort for the Cold Land Processes Experiment, the source of much of Dr. Liston's data (Parsons et al 2004). I met Dr. Rastetter at an Arctic synthesis workshop.…”

“…The devil is in the details. These details can be a syntactical issue like which delimiter to use between columns (a real issue in CLPX (Parsons et al 2004)) to semantic details like what exactly is meant by "temperature." These issues are not unique to meteorological station data, but they apply to many regular, pointbased, field measurements of the environment.…”