Remole Sensing and Hydrology 2000 (Proceedings of a symposium held at Santa Fe, New Mexico, USA, April 2000). IAHS Publ. no. 267, 2001. 311 Spatial mapping of évapotranspiration and energy balance components over riparian vegetation using airborne remote sensing C. M. U. NEALE, L. E. HIPPS Departments of Biological and Irrigation Engineering, and Plants, Soils and Biometeorology, Utah State University, Logan, Utah 84322-4105, USA e-mail: cneale@cc.usu.edu J. H. PRUEGER, W. P. KUSTAS USDA-ARS National Soil Tilth Laboratory and National Hydrology Laboratory, USA D. I. COOPER Los Alamos National Laboratory, Los Alamos, New Mexico, USA W. E. EICHINGER University of Iowa, Iowa City, Iowa, USA Abstract High resolution, airborne multispectral imagery of a riparian system dominated by salt cedar (Tamarix spp) along the Rio Grande in New Mexico, USA, was used to determine the instantaneous évapotranspiration rates and spatially distributed energy balance components over the system. Comparisons of instantaneous spatially distributed upwind fluxes with values of groundbased measured fluxes using eddy correlation techniques and other micrometeorological instruments, were conducted for two different dates. Results show considerable differences between the fluxes that can be attributed to advection, canopy heat storage and wind variability. A careful footprint analysis will need to be conducted in the future to better match the ground-based and aircraft measurements. Key words advection; airborne multispectral imagery; canopy reflectance; energy balance fluxes; eddy correlation; évapotranspiration; footprint analysis; Rio Grande; Tamarisk; upwind fetch INTRODUCTION The mapping and monitoring of western river systems in the USA requires high spatial resolution imagery, due to their small-scale variability. The present suite of available satellite sensors does not have the required spatial resolution to capture this variability. Neale (1997) showed the usefulness of airborne multispectral imagery for classifying riparian vegetation and river corridors in general. In this paper, we use the latest version of the Utah State University (USU) airborne multispectral system (Neale & Crowther, 1994) to estimate the spatial energy balance components over a riparian vegetation system of the Rio Grande River in New Mexico, USA. METHODOLOGY The study site was located in the Bosque del Apache Wildlife Refuge in the riparian zone of the Rio Grande River (33 47'7"N, 106 52'38"W). It consisted of a large
312 C. M. U. Neale et al. stretch of salt cedar (Tamarix spp.) approximately 300 m wide and 1000 m long, oriented in the north-south direction, on the west side of the river. Two towers were placed in the centre of the Tamarisk groves, approximately 600 m apart and supported several micrometeorological instruments for flux and energy balance measurements. During the June 1999 field campaign, multiple over-flights of the experiment site were conducted on different dates and times, to acquire high spatial-resolution multispectral imagery using the digital airborne system from USU. The system consists of three Kodak Megaplus 4.2i digital cameras, with interference filters forming narrow spectral bands centred in the green (0.545-0.555 pm), red (0.665-0.675 um) and nearinfrared (0.790-0.810 um). This imagery was acquired with a pixel size of 0.5 m. The thermal imagery was acquired at approximately 1-m pixel resolution using an Inframetrics 760 scanner. Spectral images in the green, red and near-infrared bands were corrected for geometric distortions and for lens vignetting effects (Neale & Crowther, 1994) and then registered together forming 3-band images. The pixels were radiometrically calibrated to radiance using previously developed relationships and then transformed into reflectance using measurements of incoming irradiance over a barium sulphate standard reflectance panel with known bi-directional reflectance properties. The calibrated 3-band images were rectified to a 1:24 000 orthophotoquad and mosaicked to form a larger image (Fig. 1). Thermal imagery was mosaicked along the flight lines and rectified to the highresolution 3-band image mosaic. The digital numbers were transformed into apparent temperature values using the system calibration. The images were corrected for atmospheric effects using radiosonde observations acquired at the site during the remote sensing flights, and MODTRAN (Berk et al, 1989). The imagery was also corrected for emissivity assuming a Tamarisk canopy emissivity of 0.95. The approach in estimating the distributed fluxes was to use the high spatial resolution remotely sensed thermal and short wave multispectral airborne imagery to estimate components of the energy balance fluxes, specifically net radiation, soil heat flux and sensible heat fluxes. The distributed latent heat flux was then estimated from the energy balance equation. The red and near-infrared reflectance of the calibrated 3-band imagery was used to estimate the albedo of the Tamarisk grove using the method of Brest & Goward (1987). This resulting albedo image was used along with the measured incoming radiation over the site to obtain the distributed net short wave radiation. Incoming long-wave radiation was estimated using Aese & Idso (1978) with the average air temperature and vapour pressure measured from the instrument tower. The outgoing long-wave radiation from the site was estimated using the atmospherically corrected surface temperatures. The distributed net radiation for each pixel over the canopy was then obtained as the balance of the incoming and outgoing short-wave and long-wave components. The distributed soil heat flux was estimated using the approach proposed by Jackson et al. (1987) where G is estimated from Rn and the Normalized Difference Vegetation Index (NDVI). The distributed sensible heat flux layer was obtained from the aerodynamic sensible heat equation using the remotely sensed canopy temperature, air temperature from the tower measurements and the bulk aerodynamic resistance for neutral atmospheric conditions estimated from the turbulence measurements (Ubar/U* 2 ).
Spatial mapping of évapotranspiration and energy balance over riparian vegetation 313 3740000 3739600 326000 326200 326400 326600 Scale 1:5000 0.1 0,1 0.2 100 0 100 200 300 400 Fig. 1 Three band image mosaic of the study area at Bosque del Apache wildlife refuge. RESULTS, DISCUSSION AND CONCLUSIONS Fluxes were estimated for two remote sensing overpasses on two different dates, 17 June (DOY 168) and 19 June (DOY 170), 1999. Figure 2 shows the distributed fluxes for DOY 170, over the site. In order to compare the distributed fluxes with the fluxes obtained from the micro-meteorological measurements, an upwind triangular polygon was digitized on the imagery centred around the average wind direction for the 10-min
314 C. M. U. Neale et al. flux-averaging period and with a vertex centred on the instrument tower. Flux statistics were extracted from the pixels within this polygon. Results are shown in Table 1. Table 1 Comparison of the measured (Micr.) and remotely sensed estimated fluxes (RS) in W m". DOY Time Wind dir. Rn Rn G G H H LE LE h 0 N RS Micr. RS Micr. RS Micr. RS Micr. 168 Mean 14:11 132 596 638 98 41-42 18 535 275 a 33 22 67 69 170 Mean 12:13 185 875 975 132 56 22 92 722 349 CJ 51 26 67 79
Spatial mapping of évapotranspiration and energy balance over riparian vegetation 315 The results show considerable differences between the measured fluxes and those estimated through the remote sensing technique. Reasons for this could be: (a) estimation of an H value using an instantaneously obtained canopy temperature but using an air temperature which was the average for a 10-min period; (b) comparison of instantaneous fluxes with 10-min average fluxes; (c) consideration of only neutral conditions in the aerodynamic resistance formulation; (d) variability in wind direction and speed throughout the 10- min period; (e) advection of energy from the surrounding areas; (f) unaccounted storage of energy within the dense Tamarisk canopy; and (g) errors in canopy temperature estimates. Variability in canopy temperature within the flux measurement footprints and the changing wind directions could have also played an important role in these mixed results. On DOY 170, a surge of water from a large storm upstream in the Rio Grande River had jumped the banks and was flowing beneath the Tamarisk canopy resulting in canopy temperatures that were cooler by as much as 4 C. To illustrate the significance of the variability, flux results for the one-hour period encompassing the remote sensing overpass of DOY 170, which occurred at 12:13 h are shown in Table 2. It is clear that there was a wind shift during the one-hour period, and that it was occurring during the remote sensing overpass at 12:13 h. The variability in measured H is apparent and could have been due to variability in what the upwind footprint to the flux tower was "seeing". Our data analysis indicates that depending on the wind direction and corresponding triangular footprint, H values varied considerably. A careful footprint analysis will be required to properly match the ground-based micro-meteorological fluxes and their area of contribution and integration time with the remotely sensed estimates. Though the comparisons of remotely sensed spatially distributed fluxes with ground-based fluxes were inconclusive, the imagery was able to capture the variability in canopy characteristics as well as surface temperature. Table 2 Summary of flux measurements from the north tower on DOY 170 for the 1-h period encompassing the remote sensing measurements. Period #(Wrn 2 ) LE (W m -2 ) Avg. r(c ) Ubar (m s" 1 ) U* (m s" 1 ) Wind direction 00-10 min 120.2 545 27.6 1.5 0.36 262 10-20 min 92.1 349 27.8 1.5 0.278 185 20-30 min 109.6 515 27.7 1.24 0.316 127 30-40 min -23.43 326 26.6 1.0 0.245 126 40-50 min 141.8 297.4 27.5 1.12 0.311 107 50-60 min -50.0 208 26.7 2.94 0.48 83 00-30 min 105.3 476 27.7 1.41 0.269 197 30-60 min 26.74 289.5 27.0 1.69 0.38 99 REFERENCES Aese, J. K. & Idso, S. B. (1978) A comparison of two formula types for calculating long-wave radiation from the atmosphere. Wat. Résout: Res. 14, 623-625. Berk, A., Bernstein, L. S. & Robertson, D. C. (1989) MODTRAN: a moderate resolution model for LOWTRAN 7. GL-TR-89-0122. Geophysics Laboratory, Bedford, Maryland, USA. Brest, C. L. & Goward, S. N. (1987) Driving surface albedo measurements from narrow band satellite data. Int. J. Remote Sens. 8, 351-367. Jackson, R. D, Moran, M. S., Gay, L. W. & Raymond, L. H. (1987) Evaluating evaporation from field crops using airborne radiometry and ground based meteorological data. Irrig. Sci. 8, 81-90. Neale, C. M. U. (1997) Classification and mapping of riparian systems using airborne multispectral videography. Rest. Ecol. 5,103-112. Neale, C. M. U. & Crowther, B. (1994) An airborne multispectral video/radiometer remote sensing system: development and calibration. Remote Sens. Environ. 49, 187-194. Report