What is a Spatial Stream Network?

Our team is using a modification of a predictive stream temperature model that incorporates the flow directionality, distance, and accumulation inherent in the dendritic network structure of rivers. This spatial stream network (SSN) based model uses statistical functions to derive predictor variables related to observed stream temperature data. 

SSN predictor variables used to describe relationships with geographically dispersed observed stream temperatures are integrated with geostatistical kriging techniques to make predictions throughout river networks with spatially explicit estimates of uncertainty. This allows us to explore the spatial properties of the data in relation to various in-stream processes and make projections based on future change scenarios.

How is this different than past modelling efforts?

Like other spatial statistical models, SSNs also consistently improve predictive performance relative to non-spatial models. Unlike past modeling efforts, ours will use a set of new and updated explanatory datasets and include process-based modifications to the SSN statistical model fitting process, such as exploring more localized weighting schemes for certain variables like forest canopy, as opposed to considering the entire reach contributing area as influencing the stream equally.

Also unlike past modeling efforts, this project draws from a unprecedentedly large network of sensors, simultaneously recording data across the entire basin over the course of three successive years. This is compared to past models for the basin which have pieced together isolated, disparate data from relatively few sites over large geographic areas.