Cloud base heights and atmospheric backscatter observations from expedition ARTofMELT, Arctic Ocean, 2023 — cloudnetpy input

Heather Guy, Sonja Murto, Michael Tjernström, Michail Karalis, John Prytherch, Ian Brooks

Cloud base heights and atmospheric backscatter data are obtained from a lidar ceilometer instrument deployed on the 7th top deck of the Swedish icebreaker Oden during the expedition ARTofMELT (Atmospheric Rivers and the Onset of Arctic Melt) in the Fram Strait pack ice during the spring transition, 8 May to 14 June of 2023. Measurements ran continuously during the whole expedition.

The observed data include direct cloud-base heights from thresholding backscatter, derived so-called 'cloud scenes' with cloud base height and cloudiness for multiple cloud layers from a clustering technique, and also the lidar backscatter profile directly from the instrument.

The files included in this dataset are specifically formatted to be used as input to the cloudnetpy retrieval algorithm. This additionally includes determination of the signal-to-noise ratio (SNR) and the calculation of a noise-screened attenuated backscatter based on a SNR threshold.

Citation

Heather Guy, Sonja Murto, Michael Tjernström, Michail Karalis, John Prytherch, Ian Brooks (2025) Cloud base heights and atmospheric backscatter observations from expedition ARTofMELT, Arctic Ocean, 2023 — cloudnetpy input. Dataset version 1. Bolin Centre Database. https://doi.org/10.17043/oden-artofmelt-2023-ceilometer-cloudnetpy-1

References

Murto S, Tjernström M, Karalis M, Prytherch J (2024a) Cloud base heights and atmospheric backscatter observations from expedition ARTofMELT, Arctic Ocean, 2023. Dataset version 1. Bolin Centre Database. https://doi.org/10.17043/oden-artofmelt-2023-ceilometer-1

Murto S, Tjernström M, Karalis M, Prytherch J (2024b) Cloud base heights and atmospheric backscatter observations from expedition ARTofMELT, Arctic Ocean, 2023 — raw data. Dataset version 1. Bolin Centre Database. https://doi.org/10.17043/oden-artofmelt-2023-ceilometer-raw-1

Tukiainen S, O’Connor E, Korpinen A (2020) CloudnetPy: A Python package for processing cloud remote sensing data. Journal of Open Source Software 5:2123. https://doi.org/10.21105/joss.02123

Kotthaus S, O’Connor E, Münkel C, Charlton-Perez C, Haeffelin M, Gabey AM, Grimmond CSB (2016) Recommendations for processing atmospheric attenuated backscatter profiles from Vaisala CL31 ceilometers. Atmospheric Measurement Techniques 9:3769–3791. https://doi.org/10.5194/amt-9-3769-2016

Data description

This dataset contains 39 NetCDF (.nc) files, one file per day, 2023-05-06 to 2023-06-13 inclusive. The total file size is about 2 GB, or ~54 MB per daily file.

Each file contains 28 variables with dimensions of time, cloud_layer, sky_condition_layer, andrange. A list of all variable names, with short explanations and units is provided below.

Note that the sky_condition and cloud_layer variables are equivalent to those in the dataset by Murto et al. (2024a), and the variable beta_raw is equivalent to the variable backscatter_profile in Murto et al. (2024a). The native 30 s time resolution data were used to generate this dataset.

Dimensions

• time Time (UTC)
• cloud_layer
• sky_condition_layer
• range Distance from instrument to centre of each range bin (m)

Variables

• All variables
  • day_of_year Time (decimal day of year)
  • year Year
  • month Month
  • day Day
  • hour Hour
  • minute Minute
  • second Time averaged to closest second
  • flag_cloudcode Code for number of cloud bases detected; as in Murto et al. (2024a)
  • vertical_visibility Vertical visibility given in case of obscured cloud base (at flag_cloudcode 4), else NaN (m)
  • highest_detected_signal Highest signal detected given in case of obscured cloud base (at flag_cloudcode 4), else NaN (m)
  • cloud_base_altitude Cloud base height of 1-3 cloud layers; NaN if no layer detected. Instrument height incorporated (m)
  • sky_condition_cloud_fraction Cloud fraction calculated with the sky condition algorithm. 0-8 = cloud coverage of up to 5 levels; 9 = obscuration; NaN = missing data or no detected layer (octal)
  • sky_condition_cloud_altitude Cloud layer height calculated with the sky condition algorithm. Cloud layer height given for 1-5 sky condition layers; NaN if no layer detected. Vertical visibility is reported as height if obscuration (at sky_condition_cloud_fraction 9). Instrument height incorporated (m)
  • beta_raw Attenuated backscatter coefficient (sr⁻¹ m⁻¹). Non-screened attenuated backscatter coefficient, equivalent to the variable backscatter_profile in Murto et al. (2024a)
  • wavelength Laser wavelength (nm)
  • height Ranges for the ceilometer backscatter profile, including the instrument height (m)
  • calibration_factor Attenuated backscatter calibration factor. Note that if calibration factor=1, this is equivalent to no calibration applied
  • altitude Altitude of site (m)
  • zenith_angle Zenith angle (degree). This is set at a single value for the purpose of the cloudnetpy algorithm and corresponds to how the ceilometer was configured assuming a level platform. In reality the zenith angle varied between 0-4 degrees when the ship was in motion. This will have essentially no impact on the data for the majority of uses.
  • beta_smooth Attenuated backscatter coefficient (sr⁻¹ m⁻¹). SNR-screened attenuated backscatter coefficient. Weak background smoothed using a 2D rolling mean window.
  • beta Attenuated backscatter coefficient (sr⁻¹ m⁻¹). SNR-screened attenuated backscatter coefficient
  • latitude Latitude of site (degree north)
  • longitude Longitude of site (degree east)
  • SNR Signal to noise ratio calculated following the method of Kotthaus et al. (2016) (Eq. 15). Minimum noise level is the median value of the nosie floor throughout the ARTofMELT campaign: 4.51e-14.

Comments

Measurement system

The data was obtained from a lidar ceilometer from Vaisala (Vaisala CL31). Clouds are detected by thresholds on backscatter radiation from an eye-safe lidar. When the lowest cloud layer is not covering the sky, multiple cloud layers can be detected. The instrument also provides a cloud scene product based on the last 10 minutes of data using a proprietary clustering algorithm that in addition to cloud base height also provide a cloud fraction for each layer. The original backscatter profile is also provided.

Quality control

A background noise removal algorithm has been applied to the backscatter profile using a signal-to-noise ratio (SNR) threshold following the methodology of Kotthaus et al. (2016).

The SNR was calculated by dividing the smoothed backscatter profile by the estimated noise floor for each time-step. The noise floor was calculated by the mean value plus one standard deviation of the atmospheric backscatter in the highest 300 m of the backscatter profile at each time-step.

A calculation of the relative variance was used to ensure that the data used to calculate the noise floor did not contain any cirrus clouds, and gaps were filled by linear interpolation.

Any values of the noise floor less than the median value of the entire campaign were set to the median vale (4.51e-14).

An optimal SNR threshold was selected based on the results of a the welch t-test acceptance plot when more than 90% of values are significantly different (p