A Universal Dynamic Threshold Cloud Detection Algorithm (UDTCDA) supported by a prior surface reflectance database

Sun, Lin; Wei, Jing; Wang, Jian; Mi, Xueting; Guo, Yamin; Yang, Leiku; Yang, Yikun; Gan, Ping; Zhou, Xuechang; Chen, Jia; Tian, Xinpeng

doi:10.1002/2015jd024722

Cited by 78 publications

(55 citation statements)

References 56 publications

(70 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To avoid the influence of clouds on surface reflectance, the monthly blue band data in 2013 were composited using the MOD09A1 product and the LSR database via the second minimum value synthesis method [30,31].…”

Section: Lsr Databasementioning

confidence: 99%

Validation and Accuracy Analysis of Global MODIS Aerosol Products over Land

et al. 2017

Self Cite

View full text Add to dashboard Cite

Land surface reflectance (LSR) and aerosol types are the two main factors that affect aerosol inversions over land. According to LSR determination methods, Moderate resolution Imaging Spectroradiometer (MODIS) aerosol products are produced using the Deep Blue (DB) and Dark Target (DT) algorithms. Five aerosol types that are determined from Aerosol Robotic Network (AERONET) ground measurements are used to describe the global distribution of aerosol types in each algorithm. To assess the influence of LSR and the method used to determine aerosol type from aerosol retrievals, 10-km global aerosol products that cover 2013 are selected for validation using Level 2.0 aerosol observations from 175 AERONET sites. The variations in the retrieval accuracy of the DB and DT algorithms for different LSR values are analyzed by combining them with a global 10-km LSR database. Meanwhile, the adaptability of the MODIS products over areas covered with different aerosols is also explored. The results are as follows. (1) Compared with DT retrievals, the DB algorithm yields lower root mean squared error (RMSE) and mean absolut error (MAE) values, and a greater number of appropriate sample points fall within the expected error (EE). The DB algorithm shows higher overall reliability; (2) The aerosol retrieval accuracy of the DB and DT algorithms decline irregularly as the surface reflectance increases; the DB algorithm displays relatively high accuracy; (3) Both algorithms have a high retrieval accuracy over areas covered by weak absorbing aerosols, whereas dust aerosols and continental aerosols produce a low retrieval accuracy. The DB algorithm shows good retrieval results for most aerosols, but a lower accuracy for strong absorbing aerosols.

show abstract

Section: Lsr Databasementioning

confidence: 99%

Validation and Accuracy Analysis of Global MODIS Aerosol Products over Land

et al. 2017

Self Cite

View full text Add to dashboard Cite

show abstract

“…MOD09 is the eight-day surface reflectance (ρ S ) product of MODIS/AQUA. The dataset was fused to the monthly average surface reflectance in January 2015 by using a minimum method [36]. These mature surficial reflectance data were used as input data in the radiative simulation.…”

Section: Satellite Datamentioning

confidence: 99%

Interference of Heavy Aerosol Loading on the VIIRS Aerosol Optical Depth (AOD) Retrieval Algorithm

Wang

Chen

et al. 2017

Remote Sensing

View full text Add to dashboard Cite

Aerosol optical depth (AOD) has been widely used in climate research, atmospheric environmental observations, and other applications. However, high AOD retrieval remains challenging over heavily polluted regions, such as the North China Plain (NCP). The Visible Infrared Imaging Radiometer Suite (VIIRS), which was designed as a successor to the Moderate Resolution Imaging Spectroradiometer (MODIS), will undertake the aerosol observations mission in the coming years. Using the VIIRS AOD retrieval algorithm as an example, we analyzed the influence of heavy aerosol loading through the 6SV radiative transfer model (RTM) with a focus on three aspects: cloud masking, ephemeral water body tests, and data quality estimation. First, certain pixels were mistakenly screened out as clouds and ephemeral water bodies because of heavy aerosols, resulting in the loss of AOD retrievals. Second, the greenness of the surface could not be accurately identified by the top of atmosphere (TOA) index, and the quality of the aggregation data may be artificially high. Thus, the AOD retrieval algorithm did not perform satisfactorily, indicated by the low availability of data coverage (at least 37.97% of all data records were missing according to ground-based observations) and overestimation of the data quality (high-quality data increased from 63.42% to 80.97% according to radiative simulations). To resolve these problems, the implementation of a spatial variability cloud mask method and surficial index are suggested in order to improve the algorithm.

show abstract

“…Sedano et al (2011) coupled pixel-based cloud seed identification and object-based cloud region growing to extract the clouds in SPOT imagery. To reduce the effects of atmospheric factors and complex surfaces, an a priori monthly surface reflectance database was established in Sun et al (2016) to support the universal dynamic threshold cloud detection algorithm (UDTCDA) algorithm for cloud detection in MODIS and Landsat-8 imagery. Zhai et al (2018) later proposed a unified method for cloud and cloud shadow detection in multi/hyperspectral images, based on spectral indices and spatial matching, with parameters which may need to be fine-tuned.…”

mentioning

confidence: 99%

Deep learning based cloud detection for medium and high resolution remote sensing images of different sensors

Shen

Cheng

et al. 2019

ISPRS Journal of Photogrammetry and Remote Sensing

237

102

View full text Add to dashboard Cite

Cloud detection is an important preprocessing step for the precise application of optical satellite imagery. In this paper, we propose a deep learning based cloud detection method named multi-scale convolutional feature fusion (MSCFF) for remote sensing images of different sensors. In the network architecture of MSCFF, the symmetric encoder-decoder module, which provides both local and global context by densifying feature maps with trainable convolutional filter banks, is utilized to extract multi-scale and high-level spatial features. The feature maps of multiple scales are then up-sampled and concatenated, and a novel multi-scale feature fusion module is designed to fuse the features of different scales for the output. The two output feature maps of the network are cloud and cloud shadow maps, which are in turn fed to binary classifiers outside the model to obtain the final cloud and cloud shadow mask. The MSCFF method was validated on hundreds of globally distributed optical satellite images, with spatial resolutions ranging from 0.5 to 50 m, including Landsat-5/7/8, Gaofen-1/2/4, Sentinel-2, Ziyuan-3, CBERS-04, Huanjing-1, and collected high-resolution images exported from Google Earth. The experimental results show that MSCFF achieves a higher accuracy than the 2 traditional rule-based cloud detection methods and the state-of-the-art deep learning models, especially in bright surface covered areas. The effectiveness of MSCFF means that it has great promise for the practical application of cloud detection for multiple types of medium and highresolution remote sensing images. Our established global high-resolution cloud detection validation dataset has been made available online

show abstract

A Universal Dynamic Threshold Cloud Detection Algorithm (UDTCDA) supported by a prior surface reflectance database

Cited by 78 publications

References 56 publications

Validation and Accuracy Analysis of Global MODIS Aerosol Products over Land

Validation and Accuracy Analysis of Global MODIS Aerosol Products over Land

Interference of Heavy Aerosol Loading on the VIIRS Aerosol Optical Depth (AOD) Retrieval Algorithm

Deep learning based cloud detection for medium and high resolution remote sensing images of different sensors

Contact Info

Product

Resources

About