Quantification of lidar uncertainties - Qlunc

What is Qlunc?

Qlunc is a python-based, open-source software, aligned with the Wind lidar Ontology by IEA TCP Wind Task52 , which can be used to estimate errors in wind velocity and wind direction estimations when probing the wind with lidars. The code has an objected-oriented structure; by using python objects and simulating real lidar components the code puts all together in modules to build up a lidar digital twin. The code is modular and offers the possibility of creating different lidar objects on parallel ( refer to Tutorial2.ipynb ). This enables seamless integration of lidar modules with distinct characteristics, allowing the simulation of various lidar devices.

Figure 1. General structure of Qlunc

Currently, Qlunc's framework addresses photonics, signal processing techniques and pointing uncertainties. In each module the Guide to the Expression of Uncertainty in Measurement (GUM) is applied to compute propagation of uncertainty. Qlunc estimates uncertainties of the line-of-sight wind velocity ($V_{LOS}$), the horizontal wind velocity ($V_{h}$), 3D wind vector ($V_{wind}$) and wind direction (Φ). It involves a comparative analysis the developed analytical model and Monte Carlo simulations. Additionally, the tool provides the capability to account for correlations among measurement angles uncertainties within a single lidar and across two and/or three lidar devices.

Creating a lidar device

The user creates the different lidar components by instantiating python classes, including its functional parameters and defining their specific uncertainty functions. Then, each module (also python objects) is "filled" with the corresponding components and their uncertainties are computed following uncertainty propagation method according to the GUM model. Once each component is "ensembled", building up the different modules, the modules are included into the lidar object.

Creating atmospheric conditions

The user creates also atmospheric scenarios to account for the different atmospheric conditions the lidar has to deal with. Power law exponent α, temperature and humidity are accepted, either single values or time-dependent variabilities of these inputs, taken from peripherals.

Qlunc (NEW!) available capabilities

Uncertainties in hardware

The flexibility of the code allows users, not only to assess global lidar uncertainty due to signal noise, but also to query uncertainties contributed by noise in specific modules or even single components.

🆕 Estimated uncertainties in $V_{wind}$ and Φ with information from 3D wind vector 🆕

🆕 Estimated uncertainties in $V_{LOS}$ , $V_{h}$ and Φ due to errors in pointing accuracy and focus distance 🆕

Considered as a major contributor to uncertainty in lidar estimations, the new Qlunc's add-on uses a combination of analytic and Monte Carlo approaches for estimating the intrinsic lidar uncertainty including:

• Hardware noise
• Speckle noise
• Bias in the sampling frequency
• Bias in the laser wavelength
• Quantisation and FFT-based signal processing
• Uncertainty in the lidar line-of-sight and horizontal wind speeds, as well as the uncertainty in the wind direction, all due to errors in pointing accuracy and focus distance
• Lidar unceratinty correlations assessment

Plots

• Photodetector signal-to-noise ratio and separate contributions due to shot noise, thermal noise, dark current noise and, if needed, trans-impedance amplifier noise.
• 🆕 Uncertainties in $V_{LOS}$ , $V_{h}$ and $V_{wind}$ with the wind direction
• 🆕 Uncertainties in $V_{LOS}$ with focus distance, elevation angle and azimuth angle for a fixed wind direction
• 🆕 Uncertainty in Φ lidar estimation
• 🆕 Uncertainty in vertical and horizontal measuring planes

Figure 2. Horizontal wind speed and wind direction uncertainties - Lissajous Pattern

How to use Qlunc

⚠️ Please downolad the latest release (V1.0).

Create an environment and install dependencies

1. Having Anaconda installed is a prerequisite if we want to work in a different environment than base, and it is recommended. Then, based on the requirements added to the environment.yaml file we build the new environment named Qlunc_Env by default.

2. In the Anaconda prompt, go to the directory where you have clone/download Qlunc and type:

conda env create -f environment.yml 
conda activate <envname>

3. Your environment is ready to rumble. You have now a new environment with all the packages needed to run Qlunc.

4. In case you don't want to create a new environment, just install the requirements listed in the environment.yml file.

Download or clone the repository to a local directory.

By downloading or cloning the repository user will get several folders within which Qlunc is organized.

First,

1. Create a folder named Qlunc_Output in '\Qlunc' main directory. Here the data will be saved.
2. Copy and paste the file Qlunc_inputs_Dual.yml, for a dual lidar solution, or Qlunc_inputs_Triple.yml, for a triple lidar solution, from TestFiles_Qlunc into Main and rename it to Qlunc_inputs.yml for a quick start/test. Otherwise, fill in the template in the same folder (TestFiles_Qlunc) and rename it to Qlunc_inputs.yml. Copy and paste this file into the Main folder.

The content of each folder in the repository is breafly explained here below. Further information can be found in the readme in the corresponding folder.

Main

Main is the core of Qlunc. It contains the scripts to create the classes describing the components, modules, general inputs of the lidar device and atmospheric scenarios, and instantiates the classes to build up the virtual lidar(s).

• Qlunc_Classes.py contains the code which creates the lidar components and modules. Each lidar module/component is assigned to a python class.
• Qlunc_Instantiate.py instantiate the lidar classes taking the values from Qlunc_inputs.yml.

UQ_Functions

• Contains the functions that compute the uncertainties from different devices, calculting also the uncertainty propagation corresponding to the different modules and the lidar uncertainty as well. Users can define their own functions to calculate specific module uncertainties, and combined/expanded uncertainties as well.

Utils

• Contains scripts meant to do different tasks. Contains funtions which interface directly with Qlunc and are necessary to compute calculations. Also contains Qlunc_Plotting.py, a script to automate plots and Qlunc_ImportModules.py to import the necessary python packages.
• The new functions implemented for this release estimating the uncertainty in $V_{LOS}$ and $V_{h}$ are allocated here.

TestFile_Qlunc

• Qlunc_inputs.yml: Human-friendly data file used as input to Qlunc. Copy and paste it into Main or
• Template_yaml_inputs_file.yml to create your own use case

Tutorials

• Contains 3 Jupyter Notebook-based tutorials with their corresponding yaml files and working examples. These tutorials are meant to serve as a guide to get familiar with Qlunc's routines, and show current capabilities of the framework. Users can find more information about these tutorials in the corresponding readme in the folder Tutorials.

Requirements

The environment.yml file summarises the python packages needed to run Qlunc

How to run Qlunc

1. Fill in the Qlunc_inputs.yml with the desired values for uncertainty estimation
2. Run Qlunc_Instantiate.py to instantiate the lidar classes
3. Type "QluncData = Lidar.Uncertainty(Lidar,Atmospheric_Scenario,cts,Qlunc_yaml_inputs)". A dictionary with relevant parameters about lidar(s) uncertainties and configuration(s) is stored in QluncData
4. Alternatively, user can execute Qlunc through a graphical user interface Qlunc_GUI.py in Main developed to ease the use of Qlunc

Author

Francisco Costa

Contributions

Contributions are very welcome! If you wish to:

• Colaborate: please contact the author
• Report issues or enhance the code: post an issue or make a pull request
• Seek for support: please contact the author

License

Qlunc is licensed under SD 3-Clause License

Citing and Contact

Francisco Costa García

University of Stuttgart - Stuttgart Wind Energy

email: costa@ifb.uni-stuttgart.de