Dual Ice Crystal Imager (D-ICI): images of snow particles from Kiruna on 2014-10-19 with size, area, and fall speed measurements

Accurate predictions of snowfall require good knowledge of the microphysical properties of the snow ice crystals and particles. Shape is an important parameter as it influences strongly the scattering properties of the ice particles, and thus their response to remote sensing techniques such as radar measurements.
The fall speed of ice particles is another important parameter for both numerical forecast models as well as representation of ice clouds and snow in climate models, as it is responsible for the rate of removal of ice from these models.

A new ground-based in-situ instrument, the Dual Ice Crystal Imager (D-ICI), has been developed to determine snow ice crystal properties and fall speed simultaneously. The instrument takes two high-resolution pictures of the same falling ice particle from two different viewing directions.
Both cameras use a microscope-like set-up resulting in an image pixel resolution of approximately 4μm/pixel. One viewing direction is horizontal and is used to determine fall speed by means of a double exposure. For this purpose, two bright flashes of a light-emitting diode behind the camera illuminate the falling ice particle and create this double exposure and the vertical displacement of the particle provides its fall speed. The other viewing direction is close to vertical and is used to provide size and shape information from single-exposure images.
This viewing geometry is chosen instead of a horizontal one because shape and size of ice particles as viewed in the vertical direction are more relevant than these properties viewed horizontally as the vertical fall speed is more strongly influenced by the vertically viewed properties.
In addition, a comparison with remote sensing instruments that mostly have a vertical or close to vertical viewing geometry is favoured when the particle properties are measured in the same direction.

The instrument has been tested in Kiruna, northern Sweden (67.8N, 20.4E). First measurements from 2014-10-19 are presented with images and determined snow ice crystal properties.

The dataset is the basis of Kuhn, T. and Vázquez-Martín, S.: Microphysical properties and fall speed measurements of snow ice crystals using the Dual Ice Crystal Imager (D-ICI), Atmos. Meas. Tech., 13, 1273–1285, https://doi.org/10.5194/amt-13-1273-2020, 2020.

The data consist of images of individual snow crystals or snowflakes taken by the two cameras of D-ICI. Images from the top-view camera are in the folder named "20141018_180609_top" and the side-view images in the folder "20141018_180728_side". The folder of top-view images contains a subfolder called "detected" that contains results from image processing to detect particles and determine their edge, size (maximum dimension), and cross-sectional area (area inside boundary). These results consist of images where the background (first image in folder "20141018_180609_top") is removed (in folder "bck_removed"), black-and-white images where black marks detected snow particles (in folder "bw"), images representing the used Sobel gradient matrix (in folder "bw", file name contains "gm"), particle cut-out images (in folder "particles"), particle cut-out images with the particle edge visualized (in folder "borders"). In addition, the file "particles_output.txt" lists Particle name (image name including a running number starting at 1 for all particles found on the same image), area (in number of pixels), size (maximum dimension in number of pixels), and area ratio. The file "particles.log" is a log file from image processing containing information (described in the header), for example the maximum gradient of the Sobel gradient matrix.
The side-view images in the folder "20141018_180728_side" show double exposures of the snow particle, from which the fall speed is determined. Apart from very few exceptions where one of the cameras missed to take an image, each side-view image has a corresponding top-view image taken at the same time. However, the timestamps are not identical (but usually within one second) as the two computers saving the data were not exactly synchronized.

Four text files contain size, area and fall speed data derived from images and used for Area versus Maximum dimension and Fall speed versus Maximum dimension figures:

Size and area data:
20141018_all_particles.txt
Columns contain:
1: imagename
2: particlename
3: dmax/um (micrometre)
4: area/m2
5: Ar (area ratio)
6: m09/kg (particle mass determined from Schmitt and Heymsfield (2009), Eq.2.
7: date_time

Size and fall speed data:
20141018_fallspeed.txt
Columns contain:
1: speed_y/(m/s)
2: folder
3: particle_name
4: max_dimx (maximum width in x direction in pixels)
5: max_dimy (maximum width in y direction in pixels)
6: area(px)
7: area/m2
8: Dmax(px)
9: Dmax/um (micrometre)
10: area_ratio

Same data but only for particles of certain shapes is contained in:
droplets_20141018.txt: droplets
needles.txt: needles and agglomerates of needles
rimed.txt: rimed particles

References:
Schmitt and Heymsfield (2009), J. Atmos. Sci., 66 (7), pp2013-2028.

精准的降雪预报需要对雪冰晶体与颗粒的微物理特性具备充分认知。冰晶的形状是一项关键参数，其对冰晶的散射特性具有显著影响，进而决定了冰晶在雷达探测等遥感技术中的响应特征。

冰晶的下落速度则是另一项关键参数，既对数值预报模型至关重要，也可用于气候模型中冰云与降雪的表征，因为该参数决定了此类模型中冰晶的移除速率。

研究人员研发了一款新型地基原位观测仪器——双冰晶成像仪（Dual Ice Crystal Imager, D-ICI），可同时测定雪冰晶体的特性与下落速度。该仪器可从两个不同观测视角，对同一颗正在下落的冰晶拍摄两张高分辨率图像。

两台相机均采用类显微镜成像架构，图像像素分辨率约为4μm/像素。其中一个观测视角为水平方向，通过双重曝光技术测定冰晶的下落速度；为此，相机后方的发光二极管（Light-Emitting Diode, LED）会发出两道强光，照亮下落中的冰晶以生成双重曝光图像，通过冰晶的垂直位移即可计算其下落速度。另一个观测视角接近垂直方向，通过单次曝光图像获取冰晶的尺寸与形状信息。

选择该观测几何布局而非水平布局，是因为垂直视角下观测到的冰晶形状与尺寸，相较于水平视角的相关特征更具实际意义——冰晶的垂直下落速度更易受垂直观测属性的影响。此外，当冰晶特性的观测方向与多数遥感仪器的垂直或近垂直观测几何一致时，更便于与此类遥感仪器开展对比分析。

该仪器已在瑞典北部基律纳（67.8°N，20.4°E）开展测试，本文展示了2014年10月19日的首批观测数据，包含冰晶图像与已测定的雪冰晶体特性。

本数据集是Kuhn与Vázquez-Martín的研究成果的支撑数据：《Microphysical properties and fall speed measurements of snow ice crystals using the Dual Ice Crystal Imager (D-ICI)》，发表于*Atmos. Meas. Tech.*，13卷，1273–1285页，https://doi.org/10.5194/amt-13-1273-2020，2020年。

本数据集包含D-ICI两台相机拍摄的单颗雪晶体或雪花图像。俯视相机拍摄的图像存储于文件夹"20141018_180609_top"中，侧视相机拍摄的图像则存储于"20141018_180728_side"文件夹。俯视图像文件夹包含一个名为"detected"的子文件夹，存储了图像处理后得到的颗粒检测结果，包括颗粒边缘、尺寸（最大维度）与横截面积（边界内面积）的计算结果。此类结果包含：已移除背景（背景图像为"20141018_180609_top"文件夹中的首张图像）的图像（存储于"bck_removed"文件夹）、以黑色标记检测到的雪颗粒的黑白图像（存储于"bw"文件夹）、表征索贝尔梯度矩阵（Sobel gradient matrix）的图像（存储于"bw"文件夹，文件名包含"gm"）、颗粒裁剪图像（存储于"particles"文件夹）、带有可视化颗粒边缘的颗粒裁剪图像（存储于"borders"文件夹）。此外，"particles_output.txt"文件记录了以下内容：颗粒名称（图像名称，包含同一张图像中所有检测到颗粒的从1开始的连续编号）、面积（单位：像素数）、尺寸（单位：像素数的最大维度）与面积比。"particles.log"为图像处理生成的日志文件，包含头部描述的各类信息，例如索贝尔梯度矩阵的最大梯度值。

"20141018_180728_side"文件夹中的侧视图像为雪颗粒的双重曝光图像，可用于计算下落速度。除极少数单台相机未拍摄图像的例外情况外，每一张侧视图像均对应同一时刻拍摄的俯视图像。但由于存储数据的两台计算机未完全同步，二者的时间戳并不完全一致（通常相差不超过1秒）。

以下四个文本文件包含从图像中提取的尺寸、面积与下落速度数据，用于绘制“面积-最大维度”与“下落速度-最大维度”相关图表：

尺寸与面积数据文件："20141018_all_particles.txt"
各列依次为：
1: 图像名称
2: 颗粒名称
3: dmax/μm（微米）
4: 面积/平方米
5: Ar（面积比）
6: m09/kg（根据Schmitt与Heymsfield 2009年的公式2计算得到的颗粒质量）
7: 日期与时间

尺寸与下落速度数据文件："20141018_fallspeed.txt"
各列依次为：
1: speed_y/(m/s)（y方向速度，单位：米/秒）
2: 文件夹名称
3: 颗粒名称
4: max_dimx（x方向最大宽度，单位：像素）
5: max_dimy（y方向最大宽度，单位：像素）
6: area(px)（面积，单位：像素）
7: 面积/平方米
8: Dmax(px)（最大维度，单位：像素）
9: Dmax/μm（微米）
10: 面积比

以下文件包含相同类型的数据，但仅对应特定形状的颗粒：
"droplets_20141018.txt"：液滴状颗粒
"needles.txt"：针状颗粒与针状团聚颗粒
"rimed.txt"：覆冰颗粒

参考文献：
Schmitt与Heymsfield (2009)，*J. Atmos. Sci.*，66卷第7期，第2013-2028页。