Regional hydromorphological characterization with continuous and automated remote sensing analysis based on VHR imagery and low‐resolution LiDAR data

With high-resolution topography and imagery in fluvial environments, the potential to quantify physical fish habitat at the reach scale has never been better. Increased availability of hydraulic, temperature and food availability data and models have given rise to a host of species and life stage specific ecohydraulic fish habitat models ranging from simple, empirical habitat suitability curve driven models, to fuzzy inference systems to fully mechanistic bioenergetic models. However, few examples exist where such information has been upscaled appropriately to evaluate entire fish populations. We present a framework for applying such ecohydraulic models from over 905 sites in 12 sub-watersheds of the Columbia River Basin (USA), to assess status and trends in anadromous salmon populations. We automated the simulation of computational engines to drive the hydraulics, and subsequent ecohydraulic models using cloud computing for over 2075 visits from 2011 to 2015 at 905 sites. We also characterize each site's geomorphic reach type, habitat condition, geomorphic unit assemblage, primary production potential and thermal regime. We then independently produce drainage network-scale models to estimate these same parameters from coarser, remotely sensed data available across entire populations within the Columbia River Basin. These variables give us a basis for imputation of reach-scale capacity estimates across drainage networks. Combining capacity estimates with survival estimates from mark-recapture monitoring allows a more robust quantification of capacity for freshwater life stages (i.e. adult spawning, juvenile rearing) of the anadromous life cycle. We use these data to drive life cycle models of populations, which not only include the freshwater life stages but also the marine and migration life stages through the hydropower system. More fundamentally, we can begin to look at more realistic, spatially explicit, tributary habitat restoration scenarios to examine whether the enormous financial investment on such restoration actions can help recover these populations or prevent their extinction.

Section: Upscaling Site‐scale Information and Network Modelingmentioning

confidence: 99%

Upscaling site‐scale ecohydraulic models to inform salmonid population‐level life cycle modeling and restoration actions – Lessons from the Columbia River Basin

Wheaton

Bouwes

McHugh

et al. 2017

“…The release of Google's Earth Engine cloud computing platform in 2010 has provided the research community with both access to remotely sensed imagery and the computational power required to process such data at the global scale (Gorelick et al, 2017). Demarchi et al, 2016;Demarchi et al, 2017). To date, this research has largely been centred on the use of data from NASA's Landsat and ESA's Copernicus programmes but neither platform offers the spatial resolution necessary to resolve the types of features used in this study.…”

Section: Discussionmentioning

confidence: 99%

Google Earth as a data source for investigating river forms and processes: Discriminating river types using form‐based process indicators

Henshaw

Sekarsari

Zolezzi

et al. 2019

Google Earth provides potential for exploiting an enormous reservoir of freely‐available remotely sensed data to support river science and management. In this paper, we consider how the platform can support investigation of river physical forms and processes by developing an empirically‐based reach‐scale classification of semi‐natural European single thread to transitional rivers. Using strict reach and image selection criteria, we identified 194 reaches of 68 rivers for analysis. Measurements of channel dimensions and counts of in‐channel and floodplain features, standardised for reach length and channel width where necessary, were used to derive a series of geomorphologically‐relevant process indicators. A suite of multivariate analyses were then applied to this data set, resulting in the discrimination of five river types: laterally stable, laterally active sinuous‐meandering; transitional (near‐braided); bedrock; and cascade/step dominated. The results of the classification were tested by examining the characteristics and distribution of the river classes in relation to known independent controls of river form including reach‐scale energy and valley confinement conditions. Our results show that if methods of data extraction are carefully developed, physically meaningful river reach discrimination can be achieved using Google Earth. Although there are limits to the types of information that can be extracted such that field investigations cannot always be avoided, there is enormous potential to mine Google Earth across different space and time scales, supporting the assembly of large, reliable data sets relevant to river forms and processes in a very cost‐effective way. © 2019 John Wiley & Sons, Ltd.

“…Demarchi et al, 2017;Bizzi et al, 2019). Numerous scientific and management applications emerge from this work.…”

Section: Examples Of Applications Of the Confinement Toolmentioning

confidence: 99%

“…While coverage of HRT is ever growing (with nationwide coverages emerging in some smaller nations), the reality in most locations on earth (including most of the United States) is that HRT is on the horizon as opposed to being readily available today. Demarchi et al, 2017;Bizzi et al, 2019). > 100 km 2 ).…”

Section: Introductionmentioning

confidence: 99%

Mapping valley bottom confinement at the network scale

O'Brien

Wheaton

Fryirs

et al. 2019

In this article, we demonstrate the application of a continuous confinement metric across entire river networks. Confinement is a useful metric for characterizing and discriminating valley setting. At the reach scale, valley bottom confinement is measured and quantified as the ratio of the length of channel confined on either bank by a confining margin divided by the reach length. The valley bottom is occupied by the contemporary floodplain and/or its channel(s); confining margins can be any landform or feature that makes up the valley bottom margin, such as bedrock hillslopes, terraces, fans, or anthropogenic features such as stopbanks or constructed levees. To test the reliability of calculating confinement across entire networks, we applied our geoprocessing scripts across four physiographically distinct watersheds of the Pacific Northwest, USA using freely available national datasets. Comparison of manually digitized and mapped with modeled calculations of confinement revealed that roughly one‐third of reaches were equivalent and about two‐thirds of the sites differ by less than ±15%. A sensitivity analysis found that a 500 m reach segmentation length produced reasonable agreement with manual, categorical, expert‐derived analysis of confinement. Confinement accuracy can be improved (c. 4% to 17% gains) using a more accurately mapped valley bottom and channel position (i.e. with higher‐resolution model inputs). This is particularly important when differentiating rivers in the partly confined valley setting. However, at the watershed scale, patterns derived from mapping confinement are not fundamentally different, making this a reasonably accurate and rapid technique for analysis and measurement of confinement across broad spatial extents. © 2019 John Wiley & Sons, Ltd.