|value||Concern over increasing droughts or so-called megadroughts in the western United States (Cook et al. 2004; 2010; Williams et al, 2013) have magnified in the public eye as many states are now facing severe water shortages, such as the current situation in the State of California (http://www.water.ca.gov/). As the effects of drought now and potentially into the future under climate warming are realized, concern has grown about the possibility that many streams will simply dry up. Stream drying or desiccation is likely to increase as stream flows across the west decline (Luce and Holden 2009; Luce et al. 2013). Loss of surface flows poses serious threats to aquatic species, potentially leading to irreversible extinctions, loss of connectivity, and decreased habitat availability. The prospect of complete water loss through stream desiccation is clearly a transformational consequence of drought. Our overarching objective in this work is to develop new methods for identifying and quantifying patterns of stream desiccation, and to apply broad-scale statistical models to predict patterns of stream desiccation across broad extents within the western United States. Through this process we will be able to identify the most important factors (including those tied to droughts such as snowpack, wildfire, warm temperatures) that drive stream desiccation, which in turn will prove valuable for a multitude of management decisions that depend on anticipating future patterns of drought as climate change is realized.
The first step in this effort is to develop new methods to identify where and when streams are drying. To do this, our project will build from existing efforts to model and map stream temperatures across the western United States, and ultimately across the Nation. This massive effort to acquire temperature data has been successful in modeling and mapping temperatures across the west (www.fs.fed.us/rm/boise), but these thousands of temperature records may also allow us to identify patterns of stream desiccation. Typically when a temperature record is interrupted by stream desiccation, the data are discarded or filtered out (e.g., Dunham et al. 2005), but if we can unlock the signature of stream desiccation from these data (i.e., identifying when recordings are actually of air temperature, not water), then we are able to unlock thousands of records that can be used for modeling and mapping spatial and temporal variability in water availability in response to drought and other influences. Although the prospect of obtaining information on stream desiccation from time series of temperatures represents a major opportunity, it is critical to develop valid, reliable procedures for this purpose. To this end, we will utilize networks of stream temperature loggers paired with resistors, which measure the presence of water in streams (Blasch et al. 2002, Jaeger and Olden 2012). We are in the process of deploying these networks in streams within the Great Basin desert and in National Parks extending from southern Oregon to southern California (>200 deployments beginning in FY 2014, with more planned in future FYs). These datasets will allow us to determine exactly when streams desiccate and how recorded temperatures change when this occurs. We will explore how to diagnose these changes in thermal regimes using a host of descriptors (Arismendi et al. 2013) applied at hourly to weekly time scales. We expect the thermal signature of stream desiccation may vary according to local climatic and hydrologic conditions, and that in some locations stream desiccation will be easy to diagnose, whereas in others a greater suite of indicators may be necessary, or when local factors (e.g., water diversion, groundwater extraction) complicate prediction from readily obtainable spatial data.
Ultimately, with patterns of stream desiccation established at locations (e.g., timing, duration, frequencies), we can develop models to predict patterns based on covariates linked to temporally and spatially variable climatic conditions (including factors that are magnified under droughts), spatial variation in fixed hydro-geologic factors, patterns of land use, or other relevant covariates (e.g., Wigington et al. 2012). The main products from this project would include 1) protocols for deployment of temperature and resistance loggers for use by field personnel; 2) methods for extracting information on stream desiccation from time series of temperature data in streams; 3) models to allow us to understand factors that influence stream desiccation and to predict patterns across broad extents and under different climatic (e.g., drought) or other scenarios, 4) fully documented database in compliance with the NCCWSC and USGS policies, 5) outreach through webinars, publications, and most importantly through direct engagement with practitioners who need products from this work.