Snow Depth Retrieval on Arctic Sea Ice Using Under-Ice Hyperspectral Radiation Measurements
Snow Depth Retrieval on Arctic Sea Ice Using Under-Ice Hyperspectral Radiation Measurements
复制标题
DOI：
10.3389/feart.2021.711306
复制
发表时间：
2021-12
期刊：
影响因子：
--
通讯作者：
P. Anhaus;C. Katlein;M. Nicolaus;Stefanie Arndt;Arttu Jutila;C. Haas
中科院分区：
文献类型：
--
作者：
P. Anhaus;C. Katlein;M. Nicolaus;Stefanie Arndt;Arttu Jutila;C. Haas
通过海冰和雪传播的辐射对大气-冰-海界面的能量分配具有重要影响。雪深和冰厚对于确定其时间和空间变化至关重要。使用自主机器人车辆测量透射辐射的冰下调查通常缺乏一致的雪深和冰厚测量，因此无法调查直接关系。雪和冰在透射辐射的光谱形状上留下了明显的特征。在这里，我们使用这些特征来检索雪深。 2018 年春季，使用遥控车辆在林肯海测量了陆地一级一年冰下的透射辐射率。获得了雪深和冰厚的同地测量结果。恒定的冰厚度、清澈的水条件和低冰内生物量使我们能够分离雪的光谱特征。我们使用两种基于冰下光谱的反演方法成功地反演了雪深，其中 1）归一化差异指数和 2）包括光谱雪和海冰消光系数的理想化两层辐射传输模型。检索到的消光系数与之前的研究一致。然后，我们将这些方法应用于 2018 年秋季北极中部固定的第一年冰和漂移的、融化池覆盖的多年冰的透射率和积雪深度的连续时间序列。这两种方法都允许雪深反演精度约为 5 厘米。我们的结果表明，大气变化和绝对光照水平对雪深反演有影响。
Radiation transmitted through sea ice and snow has an important impact on the energy partitioning at the atmosphere-ice-ocean interface. Snow depth and ice thickness are crucial in determining its temporal and spatial variations. Under-ice surveys using autonomous robotic vehicles to measure transmitted radiation often lack coincident snow depth and ice thickness measurements so that direct relationships cannot be investigated. Snow and ice imprint distinct features on the spectral shape of transmitted radiation. Here, we use those features to retrieve snow depth. Transmitted radiance was measured underneath landfast level first-year ice using a remotely operated vehicle in the Lincoln Sea in spring 2018. Colocated measurements of snow depth and ice thickness were acquired. Constant ice thickness, clear water conditions, and low in-ice biomass allowed us to separate the spectral features of snow. We successfully retrieved snow depth using two inverse methods based on under-ice optical spectra with 1) normalized difference indices and 2) an idealized two-layer radiative transfer model including spectral snow and sea ice extinction coefficients. The retrieved extinction coefficients were in agreement with previous studies. We then applied the methods to continuous time series of transmittance and snow depth from the landfast first-year ice and from drifting, melt-pond covered multiyear ice in the Central Arctic in autumn 2018. Both methods allow snow depth retrieval accuracies of approximately 5 cm. Our results show that atmospheric variations and absolute light levels have an influence on the snow depth retrieval.