On The Detection Of Aitken-Mode Particles In The Amazon Upper Troposphere By Ground-Based LIDARS

Abstract

The Amazonian upper troposphere (UT), between 9 and 15 km of altitude, is the place where new organic particles are formed. These particles, transported doward by convective downdrafts, replenish the cloud condensation nuclei (CCN) population in the Amazonian planetary boundary layer (PBL). This understanding is supported by in-situ measurements by the HALO and G1 aircrafts during recent field campaigns in the region. However, aircraft measurements are only available in short-term experiments, and cannot provide systematic observations of these UT aerosols.In this study, we investigate the possible use of a ground based lidar to detect these small particles. We performed a sensitivity analysis based on numerical simulations of a UV aerosol lidar instrument that was operational in the Amazon from 2011 to 2017. We looked at the probability of detection of the UT layer as a function of the aerosol number concentration in the UT and the instrument efficiency. In addition, we investigate the effect of underestimation and overestimation on the lidar ratio used for retrieving aerosol optical properties in the detection analysis.The signal was simulated for 355 nm by calculating the lidar equation for the observed atmospheric aerosol properties and adding a Poisson noise. The intrinsic optical properties were computed using a spherical scattering Mie code with number size distribution in the Aitken (reff = 23 nm) mode for UT aerosols, and a distribution with fine (reff = 85 nm) and coarse (reff = 236 nm) mode particles in the PBL, following the in-situ aircraft measurements. Refractive indexes were based on 8 years of AERONET measurements. The calculated mean aerosol lidar ratio in the PBL and the UT were 60 sr and 22 sr respectively. Due to the uncertainty in estimating the UT lidar ratio, we allowed it to vary with a standard deviation of 2 sr. Afterwards, the numerical concentration in the UT was increased by a factor of 1x to 16x, and the system efficiency by a factor of 1x to 10 000x. For each concentration and efficiency, 100 simulations were performed. The Klett-Fernald method was used for retrieving the aerosol backscattering and extinction coefficients, with uncertainties estimated with a Monte Carlo method. The UT layer was considered to be detected when the aerosol optical depth (AOD) of that layer was significant with a confidence of 99.73%.The results show that it is possible to detect these small particles in the Amazon UT layer with a ground-based lidar, even for the median concentration of 1250 cm-3 measured by the aircrafts. For a tropospheric lidar, like the one used as a reference for these simulations, the system efficiency would need to be 3000x larger (i.e. a 55x increase in the signal-to-noise ratio). This could be achieved by a time-averaging of 30 min combined with vertical bin lengths of 375 m, or also by modifying the hardware. Furthermore, aerosol lidar ratio bias used in the lidar signal inversion do not significantly hamper the detection of the UT layer. Although these results were based on a specific instrument, the methods and analysis developed along this study are applicable to any lidars and sites.