Hydrological Modeling and Repeatability with Brokering

Data brokering aims to provide those in the hydrological sciences with access to relevant data to represent physical, biological, and chemical characteristics researchers need to accelerate discovery in their domain. Environmental models are useful tools to understand the behavior of hydrological systems. Unfortunately, parameterization of these models requires many different data sources from different disciplines (e.g., atmospheric, geoscience, ecology). In hydrological modeling, the traditional procedure for model initialization starts with obtaining elevation models, land-use characterizations, soils maps, and weather data. It is often the researcher's past experience with these datasets that determines which datasets will be used in a study, and often newer, more suitable data products exist. An added complexity is that various science communities have differing data formats, storage protocols and manipulation methods, which makes use by a non domain scientist difficult and time consuming. We propose data brokering as a means to address several of these challenges. We present two test case scenarios in which researchers attempt to reproduce hydrological model results using 1) general internet based data gathering techniques, and 2) a scientific data brokering interface. We show that data brokering increases the efficiency with which data are collected, models are initialized, and results are analyzed. As an added benefit, it appears brokering significantly increases the repeatability of a study.