A Monte Carlo Emissivity Model for Wind-Roughened Sea Surface

• Published in: Sensors (Basel, Switzerland) Vol. 19; no. 9; p. 2166
• Main Authors: Cheng, Jie, Cheng, Xiaolong, Meng, Xiangchen, Zhou, Guanhua
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 10.05.2019; MDPI
• Subjects: Accuracy; Bias; broadband emissivity; Computer simulation; Computing time; Efficiency; Emissivity; Geometry; Model accuracy; Monte Carlo model; Ray tracing; Remote sensing; sea surface emissivity; Sea surface temperature; SST; Temperature; thermal-infrared; Two dimensional models; Wind speed; Beijing China; United States > US; China; Monte Carlo model; thermal-infrared; broadband emissivity; sea surface emissivity; SST; ray-tracing
• ISSN: 1424-8220; 1424-8220
• DOI: 10.3390/s19092166

Sea surface emissivity (SSE) is a key variable in the estimation of sea surface temperature and the sea surface radiation budget. A physical base SSE model with adequate accuracy and acceptable computational efficiency is highly desired. This paper develops a Monte Carlo ray-tracing model to compute the SSE of a wind-roughened sea surface. The adoption of a two-dimensional continuous surface model and averaging the two polarization components in advance before ray-tracing gives the model acceptable computational efficiency. The developed model can output the contributions of direct emission and the reflected component to the effective emissivity. The contribution of the reflected component to the effective emissivity reaches 0.035 at an 80° emission angle for a wind speed larger than 10 m/s. The emissivity spectra and channel emissivities collected from two field campaigns and one set of outdoor measurements are used to validate the developed model. Statistical results indicate that the absolute value of bias or difference is less than 0.5% when the view angle is less than 65°, which means the retrieval accuracy of sea surface temperature (SST) is guaranteed from the view of SSE. When the view angle increases, the accuracy of the developed model degraded, especially at the view angle of 85°. Without considering this view angle, the absolute value of bias or difference is less than 0.016, and the root mean square difference (RMSD) is less than 0.018.