Hydraulic biotope mapping using terrestrial LiDAR and ADVP: linkages between water surface roughness, velocity and depth George Heritage, David Milan, Neil Entwistle 1. Aim a) Investigate the potential of Acoustic Doppler Velocity Profiling and terrestrial LiDAR for reach scale biotope mapping 7. Study site The study site was located on an active section of gravel bed river on the Wharfe between Buckden and Starbotton in North Yorkshire. The channel is up to 40m wide here and low and intermediate flows bifurcate around a large gravel bar. The river here drains a 72km 2 upland area composed of Carboniferous age Limestone and Millstone Grit and the flow regime is characterised by summer low flows often less than 1m 3 s -1 (the river comes close to drying up, flowing underground through the Limestone) rising to 3 m 3 s -1 to 5 m 3 s -1 during wetter weather. Winter flows are generally in excess of a couple of cumecs up to around 20m 3 s -1 . 2. Background Water surface flow type is increasingly being used to characterise instream habitat in the form of ‘physical’ or ‘hydraulic’ biotopes. Biotopes have also assumed importance in defining system biodiversity under the Water Framework Directive and in the development of typologies to underpin the ‘Habitat Quality Index’, providing a means of integrating ecological, geomorphological and water resource variables for management purposes. The full range of flow types are shown in Fig 1, where the character of the water surface adjusts along the flow strength continuum and varies with flow depth, velocity, substrate size, channel morphology and presence or absence of macrophytes and tree roots at bank edges. Low energy biotopes have flat water surfaces and include pools, glides and marginal deadwaters. Pools may be differentiated from glides through the later having shallower water, whilst marginal deadwaters are shallow water areas close to the channel edges. Runs are characterised by a low amplitude rippled water surface, whereas riffles are characterised by unbroken standing waves. Finally, broken standing waves characterise rapids and cascades. 3. Terrestrial LiDAR mapping of biotopes An LMS Z-210 scanning laser manufactured by Riegl Instruments (Fig 2a), was used to collect water surface coordinates for the study reach. The instrument works on the principle of ‘time of flight’ measurement using a pulsed eye-safe infrared laser source (0.9 μm wavelength) emitted in precisely defined angular directions controlled by a spinning mirror arrangement. Multiple scans, each containing several million data points, were taken from the study reached and subsequently merged using RISCAN pro postprocessing software. Fig 1. Biotope and water surface flow type metrics. On the top of the Figure, the water surface roughness delimeters for terrestrial LiDAR proposed by Milan et al. (2010) and applied in this study are presented. At the bottom of the Figure, Fr delimeters based upon data presented in Newson et al. (1998) are shown. These are applied to map surface flow types using the ADVP data. 5. LiDAR survey of the water surface Although water is a poor reflector of, ‘diffuse’ or ‘mixed’ laser returns are possible from rough water surfaces. Specular reflection occurs over flat ‘glass-like’ water surfaces such as those found on pools, glides and marginal deadwater biotopes and no or very few returns are obtained from these areas These biotopes are characterised by blank areas on the resultant water surface roughness (biotope) map. In shallow areas of the river channel dual returns are possible (Fig 3). Therefore it is important to separate first from last laser returns to ensure that the data captured represents that of the water surface and not a mixture of bed and water surface returns. Fig 3. a) Water surface survey using terrestrial LiDAR. Dual returns are possible in shallow areas of the river channel. In deep areas of the channel any laser that penetrates the water column gradually becomes absorbed. Separation of first from last returns ensures water surface data capture. b) Cross-check of LiDAR and independently surveyed water surface elevations. 4. Acoustic Doppler Profiler Survey A SonTek M5 ADVP coupled with a GPS unit and a floating hydroboard was employed during this study Fig 2b. The instrument was guided across the study reach using a rope attached to the hydroboard, permitting in the region of 10 000 data points to be collected in the reach; each position capturing information on depth, vertical velocity profile and position. Velocity and depth data obtained from the ADVP were used to calculate Froude number (Fr) using Fr = v/gd, where v is velocity, g is gravitational acceleration , and d is local water depth. Reynolds number (Re) was also calculated using Re = vd/n, where n is water viscosity at 20°C 6. Surface modelling of point cloud Water surface roughness maps are produced after post processing the dense point cloud obtained from the LiDAR survey. The standard deviation of elevations in a small moving window, equivalent to size of the smallest biotope present (0.4 m in diameter), passed over the point cloud, is attributed to a regular 0.1 m grid (Fig 4). Water surface roughness maps are then produced from the grid, and biotopes were differentiated based upon the water surface roughness delimeters presented in Fig 1. 8. ADVP results ADVP data for water depth, velocity, Fr and Re are presented in Fig 6. Pool riffle morphology is clearly picked out by variations in flow depth, where a maximum flow depth of just over 2 m was recorded in a pool towards the centre of the reach (Fig 6a). A backwater area is also evident on the right bank at the tail of the gravel bar. One of the problems with using the instrumentation was the difficulty in moving it over shallow gravel riffle areas. Despite this, realistic depth recordings were made with values less than 0.1 m. Variations in mean water column velocity may also be seen in Fig 6b, with greatest velocities occurring over riffles with peaks in the order of 1.5 ms - 1 , and the lowest velocities occurring either in pools or in the backwater area where velocities below 0.1 ms -1 were found, supporting the expected pattern at low flow. Variations in Fr (Fig 6c) show lowest values in pool areas. Small areas of critical and supercritical flow are found on the riffle areas. Greatest turbulence as highlighted on the Re map (Fig 6d), is located in pools and glides, and is likely to reflect the more complex flow structure known to occur in these areas. 9. LiDAR results LiDAR – derived water surface roughness map for the study reach is shown in Fig 7a. A re-colour coded Fr map taken from Fig 6c is shown alongside in Fig 7b, where the colour classifications are based upon the Fr delimeters presented in Newson et al. (1998; Fig 1). Areas of the water surface roughness map that are blank, indicate that no laser returns were received from the water surface due to specular reflection. These areas of the channel have a flat water surface, and represent either pools, glides or marginal deadwaters / backwaters. These blank areas on Fig 7a appear to map fairly well on to the scarcely perceptible and smooth boundary turbulent flow areas depicted on the Fr map (Fig 7b). Three riffle areas can be identified from the water surface roughness map, where a mixture of water surface roughness characteristics indicated by green, blue and black colouration. According to the Milan et al. (2010) classification these areas represent a mix of riffle, run and cascades. A predominance of green, blue and black colours in these areas are also found on the Fr map, where the Newson et al. (1998) classification system indicates the surface flow type in these areas to be a mixture of rippled, unbroken and broken standing waves. Areas shown in grey on the water surface map indicate roughness in excess of the biotope delimeters. Fig 7. Biotope and water surface flow type maps using a) terrestrial LiDAR – derived water surface roughness data using Milan et al. (2010) water surface roughness delimeters for biotope classes, and b) Fr data obtained from ADVP measurements of velocity and depth. Fr water surface flow type delimeters are based upon that proposed by Newson et al. (2008). Key: P-pool, B-boil, Rf-riffle, Rn- run, Cs-cascade, Rp-rapid, SPT- scarcely perceptible flow, SBT- smooth boundary turbulent, RP- rippled, UBS-unbroken standing wave, BSW-broken standing wave, CH-chute. Fig 2. Hardware used in field study, a) Riegl LiDAR, b) SonTek M5 ADVP a) b) Fig 5. Location (a), aerial (b) and oblique (c) reach images of the River Wharfe study site. a) Fig 6. Hydraulic data obtained using the ADVP, a) water depth, b) mean column velocity, c) Fr and d) Re. Data was interpolated using TINs and maps were produced from a 0.1 m grid. Flow direction is from the top to bottom. 10. Conclusions Both terrestrial LiDAR and ADVPs, can be used to generate extremely detailed maps of instream habitat. Both techniques offer great potential for reach scale instream habitat mapping across the flow regime, and have a significant advantage over conventional methods as the field surveyor may operate the instruments safely from the bankside. ADVP and terrestrial LiDAR used in conjunction should improve current understanding of spatial and temporal shifts in habitat availability across the flow regime. This is of undoubted importance to river managers, who require a better understanding of the links between the flow hydrograph and instream habitat. Fig 4. Surface roughness characterisation using a moving window approach. REFERENCES Milan, D.J., Heritage, G.L., Large, A.R.G., Entwistle, N. 2010. Identification of hydraulic biotopes using terrestrial laser scan data of water surface properties. Earth Surface Processes & Landforms, 35,earlyview online. Newson MD, Harper DM, Padmore CL, Kemp JL, Vogel B. 1998. A cost-effective approach for linking habitats, flow types and species requirements. Aquatic Conservation: Marine and Freshwater Ecosystems 8: 431-446. a) b) c)