近15年新疆逐日无云积雪覆盖产品生成及精度验证

Generation of daily cloudless snow cover product in the past 15 years in Xinjiang and accuracy validation

为降低云对MODIS逐日积雪覆盖产品MOD10A1和MYD10A1在新疆积雪实时监测与研究中的影响,引入交互式多传感器雪冰制图系统(interactive multi-sensor snow ice mapping system,IMS)等多源遥感数据和地面实测资料,综合时间滤波法、空间滤波法及多传感器融合法等不同的去云技术,建立基于多源数据的去云方法,生成新疆地区2002—2016年近15 a间逐日无云积雪覆盖产品数据,并利用实测资料对生成的产品数据进行精度评价及结果验证。结果表明,去云后积雪覆盖产品在新疆积雪覆盖的总体监测精度为90.61%,接近于去云前MODIS晴空积雪覆盖产品在新疆的总体监测精度(93.3%)。

MODIS snow product data constitute one of the most common data in the real-time monitoring and the research on snow cover; nevertheless,the cloud is the biggest factor affecting the application of MODIS snow cover products（ MOD10 A1 and MYD10 A1） to real-time monitoring and researching daily snow cover in Xinjiang region. By introducing such data as the interactive multi-sensor snow ice mapping system（ IMS） data and the meteorological station observation data and combining the existent cloud removal methods based on temporal filter method,spatial filter method and multi-sensor data fusion method,the authors established a new cloud removal method based on multi-source remote sensing data,with which the 15-year-long daily snow cover product in clear air of the study area from 2002 to 2016 was generated. In addition,the accuracy of the cloud removed product was evaluated with field experiment data. The results show that the overall monitoring accuracy of the new product after cloud removal reaches 90. 61% in the study area,which is close to the overall monitoring accuracy（ 93. 3%）of the MODIS clear air snow cover product before cloud removal in the study area.