------------ The README describes Terra MODIS and HYSPLIT data used to study anthropogenic glaciation of supercooled liquid-water clouds. We sampled the time series of anthropogenic glaciation events for the years 2000 to 2021 for five different aerosol sources. The five sources were a copper smelter in Rouyn-Noranda (Latitude 48.2530°, Longitude -79.0163°), Canada; an oil refinery in Regina (Latitude 50.4849°, Longitude -104.5763°), Canada; metallurgical plant in Cherepovets (Latitude 59.1442°, Longitude 37.8482°), Russia; cement plants in Fokino (Latitude 53.4445°, Longitude 34.4092°) and Volsk (Latitude 52.0517°, Longitude 47.4355°), Russia. Additional information Hannes Keernik hannes.keernik@ut.ee Velle Toll velle.toll@ut.ee This work is funded by the Estonian Research Council grant PRG1726. ------------ DATASET ------------ The dataset contains datetimes, hand-logged polygons, and Terra MODIS satellite images for events of anthropogenic glaciation of clouds. Moreover, the Terra MODIS cloud properties for glaciation-affected and nearby unaffected areas are given. In addition, data for HYSPLIT aerosol plumes are given together with Terra MODIS satellite images overlain with aerosol plumes from HYSPLIT output. The dataset can be used to study the impact of glaciation on clouds downwind of industrial aerosol point sources. Creators: Hannes Keernik, Heido Trofimov, Peter Manshausen, Matthew Christensen, Jorma Rahu, Velle Toll Organisations: University of Tartu, University of Oxford, Pacific Northwest National Laboratory Rights-holder: University of Tartu ------------ TERMS OF USE ------------ Copyright 2024 University of Tartu. This dataset is licensed by the rights-holder under a Creative Commons Attribution 4.0 International Licence: https://creativecommons.org/licenses/by/4.0/. ------------ CONTENTS ------------ The six different data records include the following: 1) GIBS_images_with_polygons.zip This folder is divided into five subfolders, each corresponding to one of the studied pollution sources. These subfolders are named according to the respective locations of the pollution sources. Each subfolder contains a series of PNG images that show glaciation events downwind of pollution sources. The images are characterized by the following: - MODIS Channels: The images are RGB images using data from MODIS channels 3, 6, and 7. - Polygon Indication: The yellow polygon was manually drawn around each glaciation case. This yellow polygon represents the area affected by the pollution source. The green polygon represents the nearby unaffected area. The red dot marks the location of the pollution source. - The PNG files are named according to the date of the satellite image. Images for the cases included in the final analysis (after the filtering process detailed in the paper) are in the subfolders. These subfolders do not contain images for all initial cases. 2) GIBS_images_with_polygons_and_hysplit.zip This folder is divided into five subfolders, each corresponding to one of the studied pollution sources. These subfolders are named according to the respective locations of the pollution sources. Each subfolder contains a series of PNG images that show glaciation events downwind of pollution sources. The images are characterized by the following: - MODIS Channels: The images are RGB images using data from MODIS channels 3, 6, and 7. - Polygon Indication: The polygon given with the red dashed line was manually drawn around each glaciation case. This polygon represents the area affected by the pollution source. The red dot marks the location of the pollution source. - HYSPLIT Model Output: The output of the HYSPLIT dispersion model is overlain on the images as blue dots. These dots represent aerosol dispersion. - Lines: Two straight lines are drawn on each image: the red line connects the pollution source to the centre of the hand-logged polygon and the blue line connects the pollution source to the centre of the HYSPLIT model output. The angle between these two lines (given in degrees) is included in the title of each image. - The date and time of the satellite image snapshot and the time for which the HYSPLIT model output was used are both indicated in the image title. - The PNG files are named according to the date and time of the satellite image. Images for the cases included in the final analysis (after the filtering process detailed in the paper) are in the subfolders. These subfolders do not contain images for all initial cases. 3) hand_logged_polygons_json.zip This folder is divided into five subfolders, each corresponding to one of the studied pollution sources. These subfolders are named according to the respective locations of the pollution sources. Each subfolder contains json files, which contain the hand-logged polygons (the coordinates for the polygons are given) representing glaciation events downwind of the pollution source. The name of the json file indicates the date and time of the Terra MODIS satellite image from which the glaciation event was visually identified. Only the JSON files representing the cases included in the final analysis (after the filtering process detailed in the paper) are in the subfolders. These subfolders do not contain all the initial cases. 4) hysplit_files.zip This folder is divided into five subfolders, each corresponding to one of the studied pollution sources. These subfolders are named according to the respective locations of the pollution sources. Each subfolder contains HYSPLIT model output files generated to correspond closely in time to the identified glaciation event downwind of the pollution source. The filename indicates the date and time for which the HYSPLIT model output was calculated. Only the JSON files representing the cases included in the final analysis (after the filtering process detailed in the paper) are in the subfolders. These subfolders do not contain all the initial cases. The notebook "read_hysplit.ipynb" digests HYSPLIT model output. 5) initial_cases_tables.zip This folder contains five csv files, each corresponding to one of the studied pollution sources and named according to the respective locations of the pollution sources. Each csv file contains three data columns: date [DD.MM.YYYY], CCN-track ("1" if a polluted cloud track induced by aerosols serving as cloud condensation nuclei was identified on a given date, 0 otherwise) and Glaciation-event ("1" if a glaciation event was identified on a given date, 0 otherwise). These tables contain all the initial cases. 6) lists_of_glaciation_cases_and_cloud_properties.zip modis_units.csv lists units of analysed MODIS data variables For each studied location, a data file lists the MODIS-derived properties for glaciation-affected areas defined by the hand-logged polygons in corresponding json files (pol) and for nearby unaffected areas (unpol). hysplit_angle_diff is the difference in azimuths (angle relative to the north) between HYSPLIT aerosol dispersion plumes and hand-logged glaciation-affected areas. MODIS cloud product is described here: https://modis.gsfc.nasa.gov/data/dataprod/mod06.php MODIS Cloud Optical Properties are described here: https://atmosphere-imager.gsfc.nasa.gov/sites/default/files/ModAtmo/MODISCloudOpticalPropertyUserGuideFinal_v1.1_1.pdf ------------ METHOD and PROCESSING ------------ For each day in the years 2000 to 2021 and for each location, we visually checked the Terra MODIS daytime image for the presence of a glaciation event and CCN-induced perturbation on cloud droplet numbers. As Terra is a polar orbiter, there was one daytime image per day for each location. We detected 104 glaciation events for Rouyn-Noranda, 67 events for Regina, 63 events for Cherepovets, 89 events for Fokino and 52 events for Volsk. After the data screening described in the paper, 84 glaciation events for Rouyn-Noranda, 59 events for Regina, 40 events for Cherepovets, 74 events for Fokino and 41 events for Volsk were included in the final analysis of Terra MODIS data. For all the glaciation events, we hand-logged the areas affected by anthropogenic emissions. We simulated the spatial distributions of the emitted aerosol particles for Rouyn-Noranda, Regina, Cherepovets, Fokino and Volsk for the years 2000 to 2021 using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) version 5 concentration plume model. The model predicts the concentration of aerosol particles in 60-minute time steps. The model is run in the forward direction for 24 hours prior to the Terra satellite overpass time. During each time step, 2500 particles are released from the industrial aerosol hot spot. The location of the plume is largely dictated by the input meteorological conditions. We use state-of-the-art European Centres for Medium Range Weather Forecasts (ECMWF) ERA5 reanalysis data as input for HYSPLIT. ------------