The practical application of scintillometers in determining the surface fluxes of heat, moisture and momentum

This thesis has collated one review chapter and five experiments concerned with addressing the question, 'how successful is the scintillometer method in determining the surface fluxes of heat, moisture and momentum and under what circumstances does it appear to fail?' Answering this question is important as a workable scintillation method provides the meteorologist with spatial integrated measurements of the surface fluxes at kilometre scales. With such a tool, ground-truth validation of remote sensing systems is possible, water balance studies can be conducted at catchment scales and energy balance experiments extended over slightly non-homogeneous terrain. Using electromagnetic scintillation to infer turbulence quantities is a fairly recent development. Although our interest lies in estimating the surface fluxes Chapter 3 makes it very clear that the foundation of the scintillation method is deeply rooted in at times questionable combination of turbulence and wave propagation theory. The novice must appreciate the important steps and assumptions in the scintillation method. Purchasing a scintillometer off-the-shelf is no guarantee of reliable measurements.How well this thesis has answered the treatise depends to an extent on the relative performance of the scintillometer method against some benchmark. For these scintillation experiments this benchmark was the eddy covariance method, selected for convenience and familiarity. In many ways this selection is a compromise as different temporal and spatial band-widths are used by each method. Scintillation measures the ensemble average of spatial fluctuations in the refractive-index along the propagation path. The eddy covariance technique makes measurements at one point in space as a function of time. The time average of the eddy covariance method is considered to be an ensemble average. Depending on atmospheric stability the eddy covariance method may require several tens of minutes to integrate the energy from all eddy scales contributing to the surface fluxes. In comparison, the scintillometer can provide statistically stable data within minutes because it only measures in the inertial-convective subrange of frequencies. In light of such differences any comparison between methods should be made with a tongue-in-cheek approach with the state of the atmosphere and surface conditions carefully scrutinised to explain any discrepancies. The strength of using the eddy covariance technique as a comparison is because it identifies any marked deviations from the norm experienced by the scintillation method. Success is in comprehending what caused these deviations not through obtaining the perfect half-hour correlation between methods. The five experiments in this thesis table such deviations and propose explanations for their particular scintillometer type and application. What follows, are the salient facts gleamed from this research.The experiments using the inner scale meter and the semiconductor laser diode highlighted the pluses and minuses ofdependence for laser scintillometers. This dependence required a correction to the measured signal variance using the spectra of Hill (1978), but it also provided information on the magnitude ofand ultimately a measure of. In the case of the semiconductor laser diode scintillometer,was indirectly determined using a measurement of average windspeed and crop height. This approach proved successful in the calculation of. In contrast, the inner scale meter measuredusing the difference in received signal variances between a gas laser and a large aperture scintillometer having littledependence. This latter approach was particularly sensitive to small signal differences and caused considerable scatter in thecomparisons. Despite a coarse result for,still compared favourably to. In addition to thedependence the laser scintillometers suffered from signal saturation in the presence of strong turbulence. This path limited the laser scintillometers and consequently the inner scale meter to operation over short distances ( L < 100 m). This limitation was less than attractive for the purposes of path averaging fluxes at catchment scale.The near-infrared large aperture scintillometer was designed to overcome the shortcomings of the laser scintillometers. It is most suited to calculations ofand it successfully did so as a component of the inner scale meter. This scintillometer however performed poorly for the rice paddy experiment. Here<and at times up to 40% of the received signal variance could be attributed to correlated T-Q fluctuations. The rice paddy experiment highlighted the effect of absorption scintillations on the large aperture signal variance. This effect was also apparent in the test of the two-wavelength scintillometer at Ahipara (Chapter 7). Subsequent modification to the scintillometer's electronic filtering alleviated this problem. If large aperture near-infrared scintillometers are still being built based on the original design of Ochs and Cartwright (1980) then theoutput signal may contain the effect of absorption fluctuations in addition to refractive fluctuations.The optical wavelength scintillometers, whether they are the laser or the large aperture types, struggled to provide a measurement ofbecause they are less sensitive to humidity fluctuations than temperature fluctuations. This was confirmed by observations of the relative contributions made by,, andtoat visible to near-infrared wavelengths. The opposite was shown to be true at microwave wavelengths and so measuringrequires a microwave scintillometer. The two-wavelength combination of microwave and large aperture scintillometers proved successful in calculating bothand, provided the effect of low frequency path-averaged humidity fluctuations was filtered from the scintillometer signals. The microwave scintillometer was shown to be sensitive to inertial-convective fluctuations and capable of calculating.When mechanical turbulence is minimal and one is interested in unstable atmospheric conditions then the free convection formula developed for the two-wavelength scintillometer provided a reasonable estimate of. However measuring at height and under very unstable conditions means the scintillometer signal variance can be corrupted by the passage of the growing CBL. Because, the scintillometer signal variance can become very small and possibly undetectable from signal noise. Unpublished data from the Ahipara experiment showed the large aperture scintillometer operated at 10 m and can provide reliable estimates ofby also using a free convective scaling formula (De Bruin et al., 1995). This result is in-line with the observations of De Bruin et al. (1995) who also showed coarse measurements ofwere sufficient to ensure reasonable calculations ofand.Until microwave scintillometers were used in these experiments measurements were confined close to the ground. Microwave technology required installing the scintillometer at some minimum height to avoid surface reflection of the propagated signal. With increased height and distance so grew the requirements to preserve MOST. The valley experiment at Brancott was the first time the effect of advection was observed on the scintillation measurements. The scintillometer footprint was sensitive to the effects of the dry-to-wet transition and the entrainment of the dry and warm air into the newly formed surface boundary layer. Under these conditions the scintillometers could not distinguish the source of the additional signal variance and the scintillometer method failed. It was also highly unlikely under these circumstances the T-Q correlation held at the scintillometer beam height.In light of these summarised results we present here some recommendations.Laser scintillometers or scintillometers which incorporate lasers such as the inner scale meter are only useful over short distances as they are limited by signal saturation and a dependence on. Theoretical advances in describing the scintillometer signal variance in strong turbulence will still require a powerful and stable laser scintillometer to implement such advances if operation over kilometre distances is required.The near-infrared large aperture scintillometer is simple and inexpensive to construct. It has minimal dependence onand it can operate over kilometre distances without signal saturation. It probably is not suited to measurements over well-irrigated surfaces because it is most sensitive to temperature fluctuations. This scintillometer performs well in unstable conditions using free convective scaling.The microwave scintillometer and large aperture near-infrared scintillometer combination can provide reliable estimates ofand. They should be used over reasonable homogeneous terrain with sufficient fetch to guarantee reliable measurements. This two-wavelength combination can estimate bothandunder unstable conditions using free convective scaling.