Improvement in Determination of Ice Water Content from Two-Dimensional Particle Imagery. Part I: Image-to-Mass Relationships
Improvement in Determination of Ice Water Content from Two-Dimensional Particle Imagery. Part I: Image-to-Mass Relationships
复制标题
DOI：
10.1175/jam2398.1
复制
发表时间：
2006-09
影响因子：
3
通讯作者：
B. Baker;-R.;P. Lawson
中科院分区：
文献类型：
--
作者：
B. Baker;-R.;P. Lawson
天然云中的冰水含量是一个重要但又很难测量的量。过去的一些研究的目标是找到冰粒质量和长度之间的平均关系，以根据现场数据确定冰水含量，例如用二维成像探头常规记录的数据。在过去的研究中，通常的方法是测量在地面采集的冰晶数据集的最大长度L和质量M。对数据的对数进行线性回归分析，以估计形式ML的平均质量-长度关系。基于晶体习惯(形状)确定数据集的子集以及整个数据集的关系。在这项研究中，探索了使用宽度W、面积A和周长P的附加参数来确定质量的替代关系。与单独使用L相比，使用L、W、A和P的组合参数可以使质量测定的均方根误差减少50%。新参数的设计考虑了冰粒的形状，而不需要首先对晶体进行分类。一个有趣的结果是，当应用于测试数据集时，仅使用A也可以实现相同的均方根误差的降低。单独使用A有助于重新分析和改进从二维图像的现有大数据集中确定冰水含量的方法，因为A仅仅是数字图像中掩星像素数。调查了这项研究中可能的误差来源，以及首先将颗粒分离为结晶习惯的有用性。
Ice water content in natural clouds is an important but difficult quantity to measure. The goal of a number of past studies was to find average relationships between the masses and lengths of ice particles to determine ice water content from in situ data, such as those routinely recorded with two-dimensional imaging probes. The general approach in these past studies was to measure maximum length L and mass M of a dataset of ice crystals collected at a ground site. Linear regression analysis was performed on the logarithms of the data to estimate an average mass-to-length relationship of the form M L. Relationships were determined for subsets of the dataset based on crystal habit (shape) as well as for the full dataset. In this study, alternative relationships for determining mass using the additional parameters of width W, area A, and perimeter P are explored. A 50% reduction in rms error in the determination of mass relative to using L alone is achieved using a single parameter that is a combination of L, W, A, and P. The new parameter is designed to take into account the shape of the ice particle without the need to classify the crystals first. An interesting result is that, when applied to the test dataset, the same reduction in rms error is also shown to be achievable using A alone. Using A alone facilitates the reanalysis and improvement of the determination of ice water content from large existing datasets of two-dimensional images, because A is simply the number of occulted pixels in the digital images. Possible sources of error in this study are investigated, as is the usefulness of first segregating the particles into crystal habits.