HR: 09:20h
AN: H31I-06    [Abstracts]
TI: Interpolation of Water Quality Along Stream Networks from Synoptic Data
AU: * Lyon, S W
EM: sl336@cornell.edu
AF: Cornell University, Biological and Environmental Engineering, Ithaca, NY 14853 United States
AU: Seibert, J
EM: jan.seibert@natgeo.su.se
AF: Stockholm University, Physical Geography and Quaternary Geology, Stockholm, 106 91 Sweden
AU: Lembo, A J
EM: ajl53@cornell.edu
AF: Cornell University, Biological and Environmental Engineering, Ithaca, NY 14853 United States
AU: Walter, M T
EM: mtw5@cornell.edu
AF: Cornell University, Biological and Environmental Engineering, Ithaca, NY 14853 United States
AU: Gburek, W J
EM: wjg1@psu.edu
AF: USDA-ARS, Pasture Lab Building, University Park, PA 16802 United States
AU: Thongs, D
EM: dthongs@dep.nyc.gov
AF: New York City Department of Environmental Protection, 71 Smith Ave, Kingston, NY 12481 United States
AU: Schneiderman, E
EM: eschneiderman@dep.nyc.gov
AF: New York City Department of Environmental Protection, 71 Smith Ave, Kingston, NY 12481 United States
AU: Steenhuis, T S
EM: tss1@cornell.edu
AF: Cornell University, Biological and Environmental Engineering, Ithaca, NY 14853 United States
AB: Effective catchment management requires water quality monitoring that identifies major pollutant sources and transport and transformation processes. While traditional monitoring schemes involve regular sampling at fixed locations in the stream, there is an interest synoptic or `snapshot' sampling to quantify water quality throughout a catchment. This type of sampling enables insights to biogeochemical behavior throughout a stream network at low flow conditions. Since baseflow concentrations are temporally persistence, they are indicative of the health of the ecosystems. A major problem with snapshot sampling is the lack of analytical techniques to represent the spatially distributed data in a manner that is 1) easily understood, 2) representative of the stream network, and 3) capable of being used to develop land management scenarios. This study presents a kriging application using the landscape composition of the contributing area along a stream network to define a new distance metric. This allows for locations that are more `similar' to stay spatially close together while less similar locations `move' further apart. We analyze a snapshot sampling campaign consisting of 125 manually collected grab samples during a summer recession flow period in the Townbrook Research Watershed. The watershed is located in the Catskill region of New York State and represents the mixed forest-agriculture land uses of the region. Our initial analysis indicated that stream nutrients (nitrogen and phosphorus) and chemical (major cations and anions) concentrations are controlled by the composition of landscape characteristics (landuse classes and soil types) surrounding the stream. Based on these relationships, an intuitively defined distance metric is developed by combining the traditional distance between observations and the relative difference in composition of contributing area. This metric is used to interpolate between the sampling locations with traditional geostatistic techniques (semivariograms and ordinary kriging). The resulting interpolations provide continuous stream nutrient and chemical concentrations with reduced kriging RMSE (i.e., the interpolation fits the actual data better) performed without path restriction to the stream channel (i.e., the current default for most geostatistical packages) or performed with an in-channel, Euclidean distance metric (i.e., `as the fish swims' distance). In addition to being quantifiably better, the new metric also produces maps of stream concentrations that match expected continuous stream concentrations based on expert knowledge of the watershed. This analysis and its resulting stream concentration maps provide a representation of spatially distributed synoptic data that can be used to quantify water quality for more effective catchment management that focuses on pollutant sources and transport and transformation processes.
DE: 0430 Computational methods and data processing
DE: 1804 Catchment
DE: 1871 Surface water quality
DE: 1894 Instruments and techniques: modeling
SC: Hydrology [H]
MN: Fall Meeting 2005