CAJ | 학술논문

如何确定合适的阈值来区分夜间辐射雾、晴空地表和中高云一直是雾检测研究的重点。圣巴巴拉DISORT大气辐射传输模型( Santa Barbara DISORT atmospheric radiative transfer,SBDART)可模拟雾顶亮度温度。基于该模型获取MODIS B20与B31波段的亮温差(brightness temperature difference,BTD),将其用于夜间辐射雾检测。以MODIS卫星数据为可行性试验数据,用国家卫星气象中心提供的地面验证数据进行验证,结果表明,使用该模型监测夜间雾的准确率达78.3%,误判率为21.7%,可靠性指标为0.643,Kappa系数为0.730。为进一步验证方法的稳定性,选取8景卫星序列图像进行时间序列分析,结果显示Kappa系数均值为0.744,说明应用当前阈值方法对MODIS夜间雾检测具有可适用性。该方法为夜间雾预报和夜间雾参数反演提供了有效的参考。
여하학정합괄적역치래구분야간복사무、청공지표화중고운일직시무검측연구적중점。골파파랍DISORT대기복사전수모형( Santa Barbara DISORT atmospheric radiative transfer,SBDART)가모의무정량도온도。기우해모형획취MODIS B20여B31파단적량온차(brightness temperature difference,BTD),장기용우야간복사무검측。이MODIS위성수거위가행성시험수거,용국가위성기상중심제공적지면험증수거진행험증,결과표명,사용해모형감측야간무적준학솔체78.3%,오판솔위21.7%,가고성지표위0.643,Kappa계수위0.730。위진일보험증방법적은정성,선취8경위성서렬도상진행시간서렬분석,결과현시Kappa계수균치위0.744,설명응용당전역치방법대MODIS야간무검측구유가괄용성。해방법위야간무예보화야간무삼수반연제공료유효적삼고。
How to obtain suitable threshold to distinguish radiation fog,clear sky surface and clouds is the focus of the study of fog detection. The Santa Barbara DISORT atmospheric radiative transfer( SBDART) model can simulate the fog top brightness temperature. In this paper,the authors obtained the brightness temperature difference( BTD) between MODIS B20 and MODIS B31 bands based on the model and applied it to the detection of radiation fog at night. The data used for feasibility test were from EOS MODIS satellite in the North China Plain on November 25, 2007,and ground validation data were from the National Satellite Meteorological Center. The varification results show that the accuracy of using the model to monitor the night time radiation fog (POD)is 78. 3%,the false alarm rate (FAR)is 21. 7%, the reliability index(CSI)is 0. 643,and the Kappa factor is 0. 730. To further validate the stability of the method,the authors selected the sequence of eight satellite images in northern China for the time series analysis. The results show that the mean value of reliability index is 0. 744,suggesting that the proposed method can serve as the foundation of night time fog forecasting and parameter inversion.