Description of Data Sets:

Elevation data to generate the watershed map was retrieved from Iowa Department of Natural Resources. Precipitation data was retrieved from The National Weather Service Advanced Hydrologic Prediction Service. NDVI was calculated on Arcgis Pro using the Landsat 8 images downloaded from USGS. Land cover data was retrieved from National Agricultural Statistics Service of United States Department of Agriculture.

The data used in heatmaps is biogeochemical composition data generated in our lab, as a part of multi-institutional NSF-funded projects, IML-CZO (Intensively Managed Landscape-Critical Zone Observatory); and CINet (Critical Interface Network). Particulates in a river in Clear Creek watershed during multiple storm events were collected and analyzed for their compound-level biogeochemical compositions. I analyzed the collected particulate samples. Biomarker compounds in each sample were identified and quantified to generate concentration data. The concentration data was scaled and used to construct heatmaps using R to visualize. The original concentration data is available in Hydroshare at http://www.hydroshare.org/resource/2fbdb2aefd264e539b3234acb6898706.

Scientific Potential of Presentation:

During storm events, rivers get muddy. However, 'Where do the particulates come from?' still remains as a big question. Rivers are an essential component of the global carbon cycle connecting land and aquatic environments. They play an important role in integrating and transporting carbon, while this role is most active during storm events. Thus, understanding their role during storm events is critical to understand the regional and global biogeochemical carbon cycling and integrate rivers in the global C model more appropriately. In addition, this knowledge will be useful for watershed management and planning and preserving ecosystem health in the watershed under our changing climate scenario that we estimate to have more frequent and intense rainfalls in the future.

To analyze storms, scientists collect riverine particulates at different times during storm events and analyze their biogeochemical compositions to track sources of the particulates and transport mechanisms. Sampling and chemical analysis strategies have been developed with time to better capture the rapid and complex phenomenon. First, to capture the rapid composition changes of riverine particulates, more high-temporal-resolution sampling has been employed. Second, to obtain more specific information about mobilized and transported sources, more source-specific indicators (biomarkers: source-specific compounds) have been used. As a consequence, they are left with a big data set with a large number of high-temporal-resolution samples and a large number of source-specific compounds that can make its interpretation more time-consuming and challenging.

My visualization technique helps to illustrate the complexity of the phenomenon and find any hidden patterns in the large data set. By converting biomarker concentrations into colors and re-ordering biomarkers that are statistically similar close together, the biomarker heatmap instantaneously shows the complicated biogeochemical composition changes of riverine particulates. The statistical grouping of biomarkers also provides useful information on sources and their relative importance changes during storm events. While this is one of the most popular data mining techniques among microbiologists and medical communities, it has never been used to describe temporal variations in geochemical compositions of earth samples to my knowledge. Above all, my biomarker heatmap can be an excellent alternative way to interpreting a big matrix with full of numbers or more conventional line or bar graphs that is often painful to interpret.