Browsing by Author "Avissar, Roni"

Spatial variability of the Earth's surface has a considerable impact on the atmosphere at all scales and understanding the mechanisms involved in land-atmosphere interactions is hindered by the scarcity of appropriate observations. A measurement gap exists between traditional point sensors and large aircraft and satellite-based sensors in collecting measurements of atmospheric quantities. Point sensors are capable of making long time series of measurements, but cannot make measurements of spatial variability. Large aircraft and satellites make measurements over large spatial areas, but with poor spatial and temporal resolution. A helicopter-based platform can make measurements on scales relevant for towers, especially close to the Earth's surface, and can extend these measurements to account for spatial variability. Thus, the Duke University Helicopter Observation Platform (HOP) is designed to fill the existing measurement gap.

Because measurements must be made in such a way that they are as uncontaminated by the platform itself as much as is possible, it is necessary to quantify the aerodynamic envelope of the HOP. The results of an analytical analysis of the location of the main rotor wake at various airspeeds are shown. Similarly, the results of a numerical analysis using the commercial Computational Fluid Dynamics software Fluent are shown. The optimal flight speed for the sampling of turbulent fluxes is found to be around 30 m/s. At this airspeed, the sensors located in front of the nose of the HOP are in advance of the wake generated by the main rotor. This airspeed is also low enough that the region of high pressure due to the stagnation point on the nose of the HOP does not protrude far enough forward to affect the sensors. Measurements of differential pressures, variables and turbulent fluxes made while flying the HOP at different airspeeds support these results. No systematic effects of the platform are seen at airspeeds above about 10 m/s.

Processing of HOP data collected using the current set of sensors is discussed, including the novel use of the Empirical Mode Decomposition (EMD) to detrend and filter the data. The EMD separates the data into a finite number of Impirical Mode Functions (IMFs), each of which is unique and orthogonal. The basis is determined by the data itself, so that it need not be known a priori, and it is adaptive. The EMD is shown to be an ideal tool for the filtering and detrending of HOP data using data gathered during the Cloud and Land Surface Interaction Campaign (CLASIC).

The ability of the HOP to accurately measure atmospheric profiles of potential temperature is demonstrated. During experiments conducted in the marine boundary layer (MBL) and the convective boundary layer (CBL), HOP profiles are evaluated using profiles from an elastic backscatter lidar. The HOP and the lidar agree on the height of the boundary layer in both cases, and the HOP effectively locates other atmospheric structures.

Atmospheric sensible and latent heat fluxes, turbulence kinetic energy (TKE) and horizontal momentum fluxes are also measured, and the resulting information is used to provide context to tower-based data collected concurrently. A brief comparison made over homogeneous ocean conditions yields good results. A more exhaustive evaluation is made using short HOP flights made over an orchard during the Canopy Horizontal Turbulence Study (CHATS).

Land-surface heterogeneity (LSH) at different scales has significant influence on atmospheric boundary layer (ABL) buoyant and shear turbulence generation and transfers of water, carbon and heat. The extent of proliferation of this influence into larger-scale circulations and atmospheric structures is a topic continually investigated in experimental and numerical studies, in many cases with the hopes of improving land-atmosphere parameterizations for modeling purposes. The blending height is a potential metric for the vertical propagation of LSH effects into the ABL, and has been the subject of study for several decades. Proper assessment of the efficacy of blending height theory invites the combination of observations throughout ABLs above different LSH scales with model simulations of the observed ABL and LSH conditions. The central goal of this project is to develop an apt and thoroughly scrutinized method for procuring ABL observations that are accurately detrended and justifiably relevant for such a study, referred to here as Detrending with Empirical Mode Decomposition (DEMD).

The Duke University helicopter observation platform (HOP) provides ABL data [wind (u, v, and w), temperature (T), moisture (q), and carbon dioxide (CO₂)] at a wide range of altitudes, especially in the lower ABL, where LSH effects are most prominent, and where other aircraft-based platforms cannot fly. Also, lower airspeeds translate to higher resolution of the scalars and fluxes needed to evaluate blending height theory.

To confirm noninterference of the main rotor downwash with the HOP sensors, and also to identify optimal airspeeds, analytical, numerical, and observational studies are presented. Analytical analysis clears the main rotor downwash from the HOP nose at airspeeds above 10 m s-1. Numerical models find an acceptable range from 20-40 m s-1, due to a growing compressed air preceding the HOP nose. The first observational study finds no impact of different HOP airspeeds on measurements from ~18 m s-1 to ~55 m s-1 over a stable marine boundary layer (MBL). Another set of observations studies HOP and tower data, using the Duke University Mobile Micrometeorological Station (MMS) over an MBL, and concludes that HOP sensible heat (SH), latent heat (LE), and carbon dioxide (F_CO2) fluxes align well with MMS findings. The HOP sensors provide ABL data at 40 Hz, as well as a real-time display of θ for in-flight ABL height estimation. Sensor calibration and alignment procedures indicate usable ABL measurements.

HOP data are especially susceptible to the spurious influence of platform motion on ABL data, largely due to the low-altitude and low-airspeed capabilities of the HOP. For example, HOP altitude motion in the presence of a lapse rate can cause spurious T fluctuations. Empirical mode decomposition (EMD) can separate HOP data into a set of adaptive and unique intrinsic mode functions (IMFs), often with physical meaning. DEMD aims to correct for spurious contributions to HOP data, while merging EMD with a correlation analysis to adjust data without eliminating relevant ABL dynamics.

To evaluate DEMD efficacy, two-dimensional synthetic T fields with simulated turbulence over a prescribed lapse rate are sampled with altitude fluctuations similar to HOP flights, and with a wide range of T perturbation and sampling path parameter variations. DEMD recovers the prescribed lapse rate within 1% on average for the 552 test cases passing the filtering criteria. The method is further evaluated via application to vertical cross sections taken from the Ocean-Land-Atmosphere Model (OLAM) large-eddy simulation (LES) results, where DEMD shows improved accuracy of SH recovery.

DEMD is applied to three low-altitude HOP flight legs flown on 19 June 2007 during the Cloud and Land Surface Interaction Campaign (CLASIC), both as an example of practical application and to compare DEMD to the initially proposed method (Holder et al. 2011, hereafter H11). H11 dictates the elimination of correlated IMFs, along with other subtle differences from DEMD, which also eliminates any ABL motions embedded in those IMFs. As suspected, the H11 method produces marked reductions of variances and turbulence kinetic energy (TKE) and substantial deviations in SH, LE, and F_CO2 compared to DEMD. DEMD detrends without unnecessary elimination.

DEMD is vital for ensuring accurate scalars and fluxes from HOP data, and a strategy for future research is presented that integrates properly detrended observations from the CLASIC HOP dataset with OLAM simulations to explore LSH effects on ABL processes and evaluate blending height theory.

While very important from an economical and societal point of view, estimating precipitation in the western United States remains an unsolved and challenging problem. This is due to difficulties in observing and modeling precipitation in complex terrain. This research examines this issue by (i) providing a systematic evaluation of precipitation observations to quantify data uncertainty and (ii) investigating the ability of the Ocean-Land-Atmosphere Model (OLAM) to simulate the winter hydroclimate in this region. This state-of-the-art, non-hydrostatic model has the capability of simulating simultaneously all scales of motions at various resolutions.

This research intercompares nine precipitation datasets commonly used in hydrometeorological research in two ways. First, using principal component analysis, a precipitation climatology is conducted for the western U.S. from which five unique precipitation climates are identified. From this analysis, data uncertainty is shown to be primarily due to differences in (i) precipitation over the Rocky Mountains, (ii) the eastward wet-to-dry precipitation gradient during the cold season, (iii) the North American Monsoon signal, and (iv) precipitation in the desert southwest during spring and summer. The second intercomparison uses these five precipitation regions to provide location-specific assessments of uncertainty, which is shown to be dependent on season and location.

Long-range weather forecasts on the order of a season are important for water-scarce regions such as the western U.S. The modeling component of this research looks at the ability of the OLAM to simulate the hydroclimate in the western U.S. during the winter of 1999. Six global simulations are run, each with a different spatial resolution over the western U.S. (360 km down to 11 km). For this study, OLAM is configured as for a long-range seasonal hindcast but with observed sea surface temperatures. OLAM precipitation compares well against observations, and is generally within the range of data uncertainty. Observed and simulated synoptic meteorological conditions are examined during the wettest and driest events. OLAM is shown to reproduce the appropriate anomaly fields, which is encouraging since it demonstrates the capability of a global climate model, driven only by SSTs and initial conditions, to represent meteorological features associated with daily precipitation variability.