Ghazal Farhani

Bio: Ghazal Farhani is an academic researcher from University of Western Ontario. The author has contributed to research in topics: Lidar & Ozone layer. The author has an hindex of 3, co-authored 6 publications receiving 16 citations.

Classification of lidar measurements using supervised and unsupervised machine learning methods

Abstract: . While it is relatively straightforward to automate the processing of lidar signals, it is more difficult to choose periods of "good" measurements to process. Groups use various ad hoc procedures involving either very simple (e.g. signal-to-noise ratio) or more complex procedures (e.g. Wing et al., 2018) to perform a task which is easy to train humans to perform but is time consuming. Here, we use machine learning techniques to train the machine to sort the measurements before processing. The presented methods is generic and can be applied to most lidars. We test the techniques using measurements from the Purple Crow Lidar (PCL) system located in London, Canada. The PCL has over 200,000 raw scans in Rayleigh and Raman channels available for classification. We classify raw (level-0) lidar measurements as "clear" sky scans with strong lidar returns, "bad" scans, and scans which are significantly influenced by clouds or aerosol loads. We examined different supervised machine learning algorithms including the random forest, the support vector machine, and the gradient boosting trees, all of which can successfully classify scans. The algorithms where trained using about 1500 scans for each PCL channel, selected randomly from different nights of measurements in different years. The success rate of identification, for all the channels is above 95 %. We also used the t-distributed Stochastic Embedding (t-SNE) method, which is an unsupervised algorithm, to cluster our lidar scans. Because the t-SNE is a data driven method in which no labelling of training set is needed, it is an attractive algorithm to find anomalies in lidar scans. The method has been tested on several nights of measurements from the PCL measurements.The t-SNE can successfully cluster the PCL data scans into meaningful categories. To demonstrate the use of the technique, we have used the algorithm to identify stratospheric aerosol layers due to wildfires.

Improved ozone DIAL retrievals in the upper troposphere and lower stratosphere using an optimal estimation method

Abstract: We have implemented a first-principle optimal estimation method to retrieve ozone density profiles using simultaneously tropospheric and stratospheric differential absorption lidar (DIAL) measurements. Our retrieval extends from 2.5 km to about 42 km in altitude, and in the upper troposphere and the lower stratosphere (UTLS) it shows a significant improvement in the overlapping region, where the optimal estimation method (OEM) can retrieve a single ozone profile consistent with the measurements from both lidars. Here stratospheric and tropospheric measurements from the Observatoire de Haute Provence are used, and the OEM retrievals in the UTLS region compared with coincident ozonesonde measurements. The retrieved ozone profiles have a small statistical uncertainty in the UTLS region relative to individual determinations of ozone from each lidar, and the maximum statistical uncertainty does not exceed a maximum of 7%.

Updated Climatology of Mesospheric Temperature Inversions Detected by Rayleigh Lidar above Observatoire de Haute Provence, France, Using a K-Mean Clustering Technique

Abstract: A climatology of Mesospheric Inversion Layers (MIL) has been created using the Rayleigh lidar located in the south of France at L’Observatoire de Haute Provence (OHP). Using criteria based on lidar measurement uncertainties and climatological mean gravity wave amplitudes, we have selected significant large temperature anomalies that can be associated with MILs. We have tested a novel approach for classifying MILs based on a k-mean clustering technique. We supplied different parameters such as the MIL amplitudes, altitudes, vertical extension, and lapse rate and allowed the computer to classify each individual MIL into one of three clusters or classes. For this first proof of concept study, we selected k = 3 and arrived at three distinct MIL clusters, each of which can be associated with different processes generating MILs in different regimes. All clusters of MIL exhibit a strong seasonal cycle with the largest occurrence in winter. The four decades of measurements do not reveal any long-term changes that can be associated with climate changes and only show an inter-annual variability with a quasi-decadal oscillation.

Optimal estimation method retrievals of stratospheric ozone profiles from a DIAL

Abstract: . This paper provides a detailed description of a first-principle optimal estimation method (OEM) applied to ozone retrieval analysis using differential absorption lidar (DIAL) measurements. The air density, detector dead times, background coefficients, and lidar constants are simultaneously retrieved along with ozone density profiles. Using an averaging kernel, the OEM provides the vertical resolution of the retrieval as a function of altitude. A maximum acceptable height at which the a priori has a small contribution to the retrieval is calculated for each profile as well. Moreover, a complete uncertainty budget including both systematic and statistical uncertainties is given for each individual retrieved profile. Long-term stratospheric DIAL ozone measurements have been carried out at the Observatoire de Haute-Provence (OHP) since 1985. The OEM is applied to three nights of measurements at OHP during an intensive ozone campaign in July 2017 for which coincident lidar–ozonesonde measurements are available. The retrieved ozone density profiles are in good agreement with both traditional analysis and the ozonesonde measurements. For the three nights of measurements, below 15 km the difference between the OEM and the sonde profiles is less than 25 %, and at altitudes between 15 and 25 km the difference is less than 10 %; the OEM can successfully catch many variations in ozone, which are detected in the sonde profiles due to its ability to adjust its vertical resolution as the signal varies. Above 25 km the difference between the OEM and the sonde profiles does not exceed 20 %.

Implementing Machine Learning Algorithms to Classify Postures and Forecast Motions When Using a Dynamic Chair

Abstract: Many modern jobs require long periods of sitting on a chair that may result in serious health complications. Dynamic chairs are proposed as alternatives to the traditional sitting chairs; however, previous studies have suggested that most users are not aware of their postures and do not take advantage of the increased range of motion offered by the dynamic chairs. Building a system that identifies users’ postures in real time, as well as forecasts the next few postures, can bring awareness to the sitting behavior of each user. In this study, machine learning algorithms have been implemented to automatically classify users’ postures and forecast their next motions. The random forest, gradient decision tree, and support vector machine algorithms were used to classify postures. The evaluation of the trained classifiers indicated that they could successfully identify users’ postures with an accuracy above 90%. The algorithm can provide users with an accurate report of their sitting habits. A 1D-convolutional-LSTM network has also been implemented to forecast users’ future postures based on their previous motions, the model can forecast a user’s motions with high accuracy (97%). The ability of the algorithm to forecast future postures could be used to suggest alternative postures as needed.

Pattern Recognition and Machine Learning

Abstract: Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Measurements of O3, NO2 and Temperature during the 2004 Canadian Arctic ACE Validation Campaign : Atmospheric chemistry experiment

Abstract: [1] The 2004 Canadian Arctic ACE Validation Campaign was conducted to provide correlative data for validating measurements from the Atmospheric Chemistry Experiment (ACE) satellite mission. These measurements were made at Eureka, Nunavut during polar springtime 2004. Six ground-based instruments were operated during the intensive phase of the campaign and ozonesondes and radiosondes were flown. During this time, ACE-FTS and ACE-MAESTRO were performing solar-occultation measurements over the Canadian Arctic. We report the first comparisons between campaign measurements and those from ACE, focusing on O 3 , NO 2 and temperature. Initial mean O 3 profiles from ACE-FTS and ACE-MAESTRO agree to within 20% between 10 and 30 km, and the NO 2 profiles agree to within 40% between 17 and 40 km, which is within the standard deviations. The ACE-FTS temperature profiles agree to better than 2.5 K with the radiosonde temperatures from 10 to 32 km and with the lidar temperatures from 17 to 45 km.

Retrieval of Temperature From a Multiple Channel Rayleigh-Scatter Lidar Using an Optimal Estimation Method

Abstract: The measurement of temperature in the middle atmosphere with Rayleigh-scatter lidars is an important technique for assessing atmospheric change. Current retrieval schemes for this temperature have several shortcomings, which can be overcome by using an optimal estimation method (OEM). Forward models are presented that completely characterize the measurement and allow the simultaneous retrieval of temperature, dead time, and background. The method allows a full uncertainty budget to be obtained on a per profile basis that includes, in addition to the statistical uncertainties, the smoothing error and uncertainties due to Rayleigh extinction, ozone absorption, lidar constant, nonlinearity in the counting system, variation of the Rayleigh-scatter cross section with altitude, pressure, acceleration due to gravity, and the variation of mean molecular mass with altitude. The vertical resolution of the temperature profile is found at each height, and a quantitative determination is made of the maximum height to which the retrieval is valid. A single temperature profile can be retrieved from measurements with multiple channels that cover different height ranges, vertical resolutions, and even different detection methods. The OEM employed is shown to give robust estimates of temperature, which are consistent with previous methods, while requiring minimal computational time. This demonstrated success of lidar temperature retrievals using an OEM opens new possibilities in atmospheric science for measurement integration between active and passive remote sensing instruments.

Laser Beam Atmospheric Propagation Modelling for Aerospace LIDAR Applications

Abstract: Atmospheric effects have a significant impact on the performance of airborne and space laser systems. Traditional models used to predict propagation effects rely heavily on simplified assumptions of the atmospheric properties and their interactions with laser systems. In the engineering domain, these models need to be continually improved in order to develop tools that can predict laser beam propagation with high accuracy and for a wide range of practical applications such as LIDAR (light detection and ranging), free-space optical communications, remote sensing, etc. The underlying causes of laser beam attenuation in the atmosphere are examined in this paper, with a focus on the dominant linear effects: absorption, scattering, turbulence, and non-linear thermal effects such as blooming, kinetic cooling, and bleaching. These phenomena are quantitatively analyzed, highlighting the implications of the various assumptions made in current modeling approaches. Absorption and scattering, as the dominant causes of attenuation, are generally well captured in existing models and tools, but the impacts of non-linear phenomena are typically not well described as they tend to be application specific. Atmospheric radiative transfer codes, such as MODTRAN, ARTS, etc., and the associated spectral databases, such as HITRAN, are the existing tools that implement state-of-the-art models to quantify the total propagative effects on laser systems. These tools are widely used to analyze system performance, both for design and test/evaluation purposes. However, present day atmospheric radiative transfer codes make several assumptions that reduce accuracy in favor of faster processing. In this paper, the atmospheric radiative transfer models are reviewed highlighting the associated methodologies, assumptions, and limitations. Empirical models are found to offer a robust analysis of atmospheric propagation, which is particularly well-suited for design, development, test and evaluation (DDT&E) purposes. As such, empirical, semi-empirical, and ensemble methodologies are recommended to complement and augment the existing atmospheric radiative transfer codes. There is scope to evolve the numerical codes and empirical approaches to better suit aerospace applications, where fast analysis is required over a range of slant paths, incidence angles, altitudes, and atmospheric conditions, which are not exhaustively captured in current performance assessment methods.

A Raman Lidar Tropospheric Water Vapour Climatology and Height-Resolved Trend Analysis over Payerne Switzerland

Abstract: . Water vapour is the strongest greenhouse gas in our atmosphere, and its strength and its dependence on temperature lead to a strong feedback mechanism in both the troposphere and the stratosphere. Raman water vapour lidars can be used to make high-vertical-resolution measurements on the order of tens of metres, making height-resolved trend analyses possible. Raman water vapour lidars have not typically been used for trend analyses, primarily due to the lack of long-enough time series. However, the Raman Lidar for Meteorological Observations (RALMO), located in Payerne, Switzerland, is capable of making operational water vapour measurements and has one of the longest ground-based and well-characterized data sets available. We have calculated an 11.5-year water vapour climatology using RALMO measurements in the troposphere. Our study uses nighttime measurements during mostly clear conditions, which creates a natural selection bias. The climatology shows that the highest water vapour specific-humidity concentrations are in the summer months and the lowest in the winter months. We have also calculated the geophysical variability of water vapour. The percentage of variability of water vapour in the free troposphere is larger than in the boundary layer. We have also determined water vapour trends from 2009 to 2019. We first calculate precipitable water vapour (PWV) trends for comparison with the majority of water vapour trend studies. We detect a nighttime precipitable water vapour trend of 1.3 mm per decade using RALMO measurements, which is significant at the 90 % level. The trend is consistent with a 1.38 ∘ C per decade surface temperature trend detected by coincident radiosonde measurements under the assumption that relative humidity remains constant; however, it is larger than previous water vapour trend values. We compare the nighttime RALMO PWV trend to daytime and nighttime PWV trends using operational radiosonde measurements and find them to agree with each other. We cannot detect a bias between the daytime and nighttime trends due to the large uncertainties in the trends. For the first time, we show height-resolved increases in water vapour through the troposphere. We detect positive tropospheric water vapour trends ranging from a 5 % change in specific humidity per decade to 15 % specific humidity per decade depending on the altitude. The water vapour trends at five layers are statistically significant at or above the 90 % level.