Using Distributed Temperature Sensing (DTS) and LDCA to Parse Thermal Forcings in Small Creeks

English Abstract:

Temperature has long been used as an indicator of ecosystem health and suitability for aquatic species, particularly in sensitive areas crucial to the persistence of declining fish populations. Typically, stream temperature surveys have long duration but only at point locations, limiting the precision of efforts to predict stream temperatures or understand broader climate linkages, and single–logger data give no insight into the spatial heterogeneity of thermal conditions often exploited by biota. Distributed Temperature Sensing (DTS) provides temperature data at high spatial and temporal resolution up to 5–km in length, allowing for detailed assessment of a creek’s heat budget. Rather than calculating a detailed energy balance from a single site or using a statistical approach, here we describe a hybrid method that uses Least Dependent Component Analysis (LDCA) capable of taking advantage of DTS data density in time and space. The method identifies distinct thermal components in the stream’s heat budget, using only temperature data and an algorithm based on mutual information that parses signals in the temperature data into dominant components or groups of parameters that force temperature signals in similar ways. These signals can be interpreted as sets of heat-flux elements sharing coordinated (non-independent) dynamics, both simplifying the number of heat budget components as well as the number thermally forcing stream temperatures. Comparing these components to meteorological data and fluvial system structure allowed us to relate the groups back to causal heating and cooling mechanisms, which can be tested directly with targeted heat-budget studies. We applied this method to a small, arid–land creek, and found that a minimum of three distinct components were necessary to describe the thermal heterogeneity of a 1–km reach. We could also estimate a spatial response profile of each component, yielding insight into possible links between stream geomorphology and function. This method shows promise to aid with siting and defining detailed heat–budget studies, determining the dimensionality of heat budgets in natural streams, and more broadly for associating thermal components to fluvial structure and processes.