A multiscale approach to water quality variables in a river ecosystem

Abstract

Monitoring select ecosystem variables across time and interpreting the collected data are essential components of ecosystem assessment supporting management. Increasingly affordable sensors and computational capacity have made very large dataset assembly more common. However, these datasets initiate analytical challenges by their size and theoretical challenges due to the scale of the processes they encompass. Multiscale assessment of high temporal resolution water quality sensor data (temperature, in vivo chlorophyll a, colored dissolved organic matter) collected year-round was conducted for the Upper St. Lawrence River. Using numerical methods that directly integrate the concept of scale, we show that consideration of scale-dependent processes can lead to increased predictive power and a clearer understanding of ecosystem function. These results suggest that multiscale methods are not only an alternative way of approaching long-term data assessment, but also a necessity in order to avoid spurious interpretation. Consequently, the concept of scale as described here can be consistently integrated into long-term data studies to assist in the interpretation of high-resolution data that help describe natural phenomena in aquatic systems.