The measurement of fluxes such as evapotranspiration from complex canopies and terrain are challenging the technological resources of both the meteorological and hydrological sciences. Evapotranspiration is one of the critical variables in both water and energy balance models of the hydrological system. The hydrologic system is driven by the soil-plant-atmosphere continuum, and as such is a spatially distributed process. To this end, the Los Alamos National Laboratory's volume-imaging, scanning water-vapor Raman lidar has been shown to be able to estimate the latent energy flux at a point. The extension of this capability to larger scales over complex terrain represents a step forward. This work outlines the techniques used to estimate the spatially resolved latent energy flux over Cottonwoods in a riparian corridor in southern Arizona as part of SALSA.
The scanning Raman lidar was fielded along with an array of various point sensors along a reach of the San Pedro river in early August of 1997, and collected data over an 8 day period.
Preliminary analysis of the data has begun with the first part being an inspection of the vertical lidar scans of water vapor. Latent energy flux is computed from 25 m wide vertical profiles derived from vertical scans using Monin-Obukhov similarity theory, and initialized with u<sub>*</sub> estimates from a sodar. Lidar vertical scans were typically 500 to 600 m long and 75 m high at a range of 500 m. During SALSA, lidar vertical scans were also stepped every 5 degrees in azimuth, to form a 270 degree swath used to create 30 minute flux maps with 25 m horizontal resolution.
A cursory inspection of the data reveals some interesting properties of the canopy top when compared to the relatively flat grass area. The cottonwoods show extensive coherent plume structure directly over the tree canopy, while the grass area shows little convective features at this time. The means of the two profiles are roughly the same however, the range of the variation over the trees are about 2.5 times larger than over the grass. Secondly, there appears to be a steep highly unstable zone about 1 meter deep directly over the tree canopy that is not seen over the grass. Interestingly, the size of the turbulent structures in the profiles for both surfaces is about 10 to 12 meters in size. Coherent plumes and structures over the trees are two to three grams per kilogram higher than the background of 11 to 12 g/kg. Over the grass, the structures are somewhat weaker in intensity.
Presumably, the well watered trees are transpiring at a higher rate than the dry grass, giving rise to more intense turbulence, mixing, and coherent structures. The higher intensity turbulence over the trees would explain the variability of the Cottonwood profile. These hypothesis and observations will be tested with further data analysis in the near future, and maps of the latent heat flux will be generated from the grass and cottonwood biomes.