| EVALUATION OF SENTINEL-2 LEVEL 1-C CLOUD MASK PRODUCT TO SUPPORT INFORMED SURFACE ANALYSES IN THE MOUNTAINOUS AREAS (CASE STUDY: the mountainous area of Sabalan, ARDABIL, IRAN) |
| Paper ID : 1029-SMPR |
| Authors: |
|
Jafar Jafarzadeh *1, Mahsa Nakhostin Rouhi2, Seyed Kazem Alavipanah3 1Department of Remote Sensing, Faculty of Geography and Environmental Planning, Tabriz University, Iran. 2Ph.D Student in Remote Sensing and GIS, Faculty of Geography, University of Tehran. 3Department of Remote Sensing and GIS, Faculty of Geography, University of Tehran, Iran |
| Abstract: |
| Unknown clouds can produce misleading results when analysing surface and atmospheric parameters. This is why cloud detection is one of the major problems with optic remote-sensed images. In this paper, the ability of Sentinel-2 L1C cloud masks to evaluate clouds on different types of biosphere such as vegetation, soil, residential areas and various cloudy conditions is investigated. Cloud reference mask was created to detect a misdiagnosis using a local threshold method, which decomposes separately from Sentinel-2 Band2 (0.490μm) and Band10 (1.375μm) decompositions. The histogram-based thresholds are localized by examining individual bands and natural colour combinations (B4B3B2). The analysis of cloud classification mistakes for our experimental areas allows us to obtain important results of the public value. In general, L1C cloud masks (average Omission Error, OE, 29.43%) were denied. This error (OE>45%) for dark-clouded images with a large transition area (between the cloud core and the bright spots) and cirrus clouds increased almost as a major source of removal errors (0.76 R2). Excessive concentrations are found in the presence of cavities inside the main cloudy area. Two other milestones were critical to the L1C cloud mask product. Detection of Sablan mountainous areas (OE>72%) was very inefficient due to the presence of large clouds and high water vapour. On the other hand, widespread cirrus clouds are created over the slopes of Mount Sablan on account of the wet and cold climate. According to the results, the users of Sentinel-2 L1C should make certain necessary precautions, such as performing the atmospheric corrections. |
| Keywords: |
| : Sentinel-2, Cloud Mask, Surface Analyses, Remote Sensing, Satellite Image, Cloud Detection |
| Status : Conditional Accept (Poster) |
