Cloud Cover Detection Research Articles

Empirical Verification of Satellite Data on Solar Radiation and Cloud Cover over the Baltic Sea

Abstract Empirical verification of the reliability of estimating the amount of solar radiation entering the sea surface is a challenging topic due to the quantity and quality of data. The collected measurements of total and diffuse radiation from the Multifilter Rotating Shadowband Radiometer (MRF-7) commercial device over the Baltic Sea were compared with the satellite results of using modeling data. The obtained results, also divided into individual spectral bands, were analyzed for usefulness in satellite cloud and aerosol detection. The article presents a new approach to assessing radiation and cloud cover based on the use of models supported by satellite data. Measurement uncertainties were estimated for the obtained results. To reduce uncertainty, the results were averaged to the time constant of the device, day, and month. The effectiveness of the method was determined by comparison against the SM Hel measurement point. The empirical results obtained confirm the effectiveness of using satellite methods for estimating radiation along with cloud-cover detection over the sea with the adopted uncertainty values. Significance Statement The difference in the amount of solar energy reaching the sea surface between cloudless and cloudy areas reaches tens of percent. Empirical results confirm the effectiveness of using satellite methods to estimate solar radiation along with cloud-cover detection. Over the sea in comparison to land, the amount of empirical data is limited. This research uses new empirical results of radiation to determine the accuracy of satellite estimation results. Experimental results show that the proposed method is effective and adequately parameterizes the detection of satellite image features.

SAFETRANS: a system for 3D visibility estimation and cloud cover detection.

The estimation of visibility is of significant importance in aviation safety and forms part of the measurements routinely collected in real time to provide safety guidelines and decisions. Our work concerns the creation and implementation of a lidar-based visibility estimation system as part of the SAFETRANS research program. We created a reproducible system to (1)support standard airport equipment, (2)serve simultaneously as a visibility meter and a ceilometer reporting on cloud cover, (3)offer increased accuracy and improved capabilities compared to standardized equipment currently in use while (4)requiring minimal user training to function. This work presents the visibility estimation and cloud cover algorithms and subsequently reports on results of field tests in a number of Greek airports under various atmospheric conditions.

Detection of Cloud Cover in Satellite Imagery Using Semantic Segmentation

Accurate detection of cloud cover is an important task in the field of Remote Sensing of the environment. Currently, a lot of development is going on in this field by using various methods. Some of the methods apply concepts of machine learning (ML) whereas some apply deep learning. Since the accuracy of ML being lower than deep learning, the latter is preferred. This paper also uses the method of deep learning to detect cloud cover using images of satellite. This paper proposes a modified U-Net based deep learning model for cloud cover detection in satellite images. The model proposed is not as accurate as the original model, but it compensates for it by reducing the time for learning. The accuracy of the model came out to be 89.73%.

Impact of cloud cover on local remote sensing – Piaśnica River case study

New satellite-based techniques open up new horizons to researchers and local communities. Concurrently, however, requirements and expectations with regard to satel-lite-based remote sensing products are increasingly higher. By relying on satellite-derived information, environmental observations can cover areas of a few to several metres resolution. Here we are dealing with freeof-charge and generally available sources of satellite-based information. The Piaśnica River mouth area was selected as an observation site representing a highly dynamic morphological transect. The paper compares products of cloud cover detection, supplied with data and available in the Copernicus database for a local area in the coastal zone of the Baltic Sea. The absolute difference did not exceed 5%, which confirms a high efficiency of the solutions offered. More than 96% of the clouded area determined for the Sentinel-2/MSI (Multispectral Instrument) was correctly identified when compared with supervised observations. The rate was lower (92%) for the Sentinel-3/OLCI (Ocean and Land Colour Instrument). It was eventually concluded that, at the local level, successful observations can be conducted using the cloud cover map supplied with the satellite data. At the same time, the analyses presented do not rule out further efforts to, e.g., increase the accuracy and speed of the analyses.

Performance evaluation of selected cloud occlusion removal algorithms on remote sensing imagery

Removal of cloud cover from remote sensing satellite images is crucial to many optical remote sensing image data users because cloud cover can conceal important spatial information on an image data. This underscores the importance of making an informed choice in the selection of appropriate cloud cover detection and removal algorithms. For the purpose of large-scale training data, neural networks have been successful in many image processing tasks, but the use of neural networks to remove cloud occlusion in remote sensing imagery is still relatively evolving. The aim of this study is to evaluate the performance of two image restoration algorithms (The spatial attentive generative adversarial network and Convolutional autoencoder with symmetrical skip connection) used for the removal of cloud cover on remote sensing images. An open-source RICE dataset was used for the training and prediction of the models as each of them were implemented for the cloud cover removal. The evaluation metrics used to compare the models’ performance are the Structural similarity index ratio (SSIM), peak signal to noise ratio (PSNR), and the time taken for each model to complete its network training. After a successful completion of the network training using 80% of the data and the remaining 20% to test the networks, the spatial attentive generative adversarial network achieved the best performance on both the peak signal to noise ratio with a value of 26.3447 and the SSIM with a value of 0.8949 while the convolutional autoencoder generates a peak signal to noise ratio of 25.8257 and a SSIM of 0.6307. The result proves that SpaGAN is more effective for automatic cloud cover removal on remote sensing images and the improvement of the quality of the restored image when compared to CNN autoencoder.

Development of a new hybrid cloud cover detection method for all sky mapping

Local precise cloud detection techniques in the daytime are finding a wide range of applications in solar farms and for tracking the factory flue gases. Additionally, the cloud coverage is a measure for the site quality of astronomical observatories. The moon is itself a problem for astronomical observations. If there are astronomical objects near to moon, the light scattered by clouds from moon in the sky also has to be determined to extract from the observation schedule of robotic telescope. A novel hybrid method to deal with all-sky images in any conditions in daytime and nighttime is developed, and related algorithms are implemented in MATLAB. A GUI has been presented which interactively analyzes the images taken from all-sky cams. It plots a radar scan of the sky with clouds in polar coordinate systems. The systematic mean error is estimated at less than 3%.

Reply to Comment on “Comparison of Cloud Cover Detection Algorithms on Sentinel–2 Images of the Amazon Tropical Forest”

In their comments about our paper, the authors remark on two issues regarding our results relating to the MACCS-ATCOR Joint Algorithm (MAJA). The first relates to the sub-optimal performance of this algorithm under the conditions of our tests, while the second corresponds to an error in our interpretation of MAJA’s bit mask. To answer the first issue, we acknowledge MAJA’s capacity to improve its performance as the number of images increases with time. However, in our paper, we used the images we had available at the time we wrote our paper. Regarding the second issue, we misread the MAJA’s bit mask and mistakenly labelled shadows as clouds. We regret our error and here we present the updated tables and images. We corrected our estimation and, consequently, there is an increment in MAJA’s accuracy in the detection of clouds and cloud shadows. However, these increments are not enough to change the conclusion of our original paper.

Comment on “Comparison of Cloud Cover Detection Algorithms on Sentinel-2 Images of the Amazon Tropical Forest”

In their recent study, Sanchez et al. compared various cloud detection methods applied to Sentinel-2, specifically on images acquired over the Amazonian region, known for its frequent cloud cover. Comparison of cloud screening methods for optical satellite images is a complex task, which must take several parameters into account, such as the definition of a cloud, which can differ according to the methods, the different coding of the cloud and shadow masks, the possible dilation of masks, and also the way the method must be used to perform in nominal conditions. We found that the otherwise serious and useful comparison of cloud masks by Sanchez et al. is not fair to the real performances of MAJA cloud detection, for two reasons: (i) two thirds of the images used in the comparison were acquired before the launch of Sentinel-2B satellite, when the revisit of the Sentinel-2 mission was 20 days instead of five days for the nominal conditions of the mission, and (ii) there is an error in the understanding of how MAJA cloud masks are coded which also probably artificially degraded the results of MAJA as compared to the other methods.

Cloud Cover Avoidance in Satellite Systems

<span id="docs-internal-guid-41420025-7fff-e714-8d51-73d40b2b7340"><span>Cloud cover is primarily a major difficulty in the acquisition of optical satellite images and has a negative impact on the efficiency of data scheduling. Along with data scheduling, the computational power required is also increasing. Recent advances in an extensive variety of technologies have resulted in an explosion in the amount of data. Different methodologies have been used for Object detection in remote sensing images but it remains a challenge because of its diversity and complex backgrounds. In this paper, a cloud cover detection technique based on Convolutional Neural Networks is proposed for remote sensing images. The classifying model uses a neural network where the underlying features are used to classify the image as useful or not. Results illustrate that the proposed method outperforms other state of the art methods that exist. Once classified, it will be transmitted from the satellite to the earth giving the researchers only convenient pictures to study. This will help to save a massive amount of computation, expense and time.</span></span>

Comparison of Cloud Cover Detection Algorithms on Sentinel–2 Images of the Amazon Tropical Forest

Tropical forests regulate the global water and carbon cycles and also host most of the world’s biodiversity. Despite their importance, they are hard to survey due to their location, extent, and particularly, their cloud coverage. Clouds hinder the spatial and radiometric correction of satellite imagery and also diminishing the useful area on each image, making it difficult to monitor land change. For this reason, our purpose is to identify the cloud detection algorithm best suited for the Amazon rainforest on Sentinel–2 images. To achieve this, we tested four cloud detection algorithms on Sentinel–2 images spread in five areas of the Amazonia. Using more than eight thousand validation points, we compared four cloud detection methods: Fmask 4, MAJA, Sen2Cor, and s2cloudless. Our results point out that FMask 4 has the best overall accuracy on images of the Amazon region (90%), followed by Sen2Cor’s (79%), MAJA (69%), and S2cloudless (52%). We note the choice of method depends on the intended use. Since MAJA reduces the number of false positives by design, users that aim to improve the producer’s accuracy should consider its use.

ОПРЕДЕЛЕНИЕ ПАРАМЕТРОВ ОБЛАЧНОГО ПОКРОВА СИСТЕМАМИ АВТОМАТИЧЕСКОЙ ОБРАБОТКИ СПУТНИКОВЫХ ДАННЫХ

The paper discusses the results of comparing cloud cover properties determined by using polar orbiting satellite data (AVHRR/NOAA and MSU-MR/Meteor-M No. 2) for the European territory of Russia and Western Siberia. The cloud characteristics were computed by two threshold techniques: Complex Threshold Technique (CTT) (developed at the European Centre of the State Research Center ‗Planeta‘) and Cloud Cover Detection Technique (CCDT) (developed at the Siberian Centre of ‗Planeta‘). Pixel-by-pixel comparison was performed for very close in time satellite observations, and it showed that in spite of technical similarity of the two radiometers and little difference between both techniques used for the classifications, the results were not the same. The quality of the MSU-MR classification is significantly worse than that of the two AVHRR classifications. In fact, the MSU-MR derivation of cloud parameters fails in optically thin cirrus and altocumulus clouds, thus underestimating the cloud top height for multilayered clouds. As a result, the cloud top is found to be lower, warmer and less iced in comparison with both AVHRR estimates, regardless of the region and other conditions; on the contrary, the cloud top of low and middle clouds appears to be colder, higher and more iced according to MSU-MR data. The MSU-MR cloud mask is strongly underestimated at night during the cold period of the year. The CTT and CCDT‘s cloud top height, temperature and water phase retrieved by AVHRR data are quite close for both regions.

Cloud cover detection combining high dynamic range sky images and ceilometer measurements

This paper presents a new algorithm for cloud detection based on high dynamic range images from a sky camera and ceilometer measurements. The algorithm is also able to detect the obstruction of the sun. This algorithm, called CPC (Camera Plus Ceilometer), is based on the assumption that under cloud-free conditions the sky field must show symmetry. The symmetry criteria are applied depending on ceilometer measurements of the cloud base height. CPC algorithm is applied in two Spanish locations (Granada and Valladolid). The performance of CPC retrieving the sun conditions (obstructed or unobstructed) is analyzed in detail using as reference pyranometer measurements at Granada. CPC retrievals are in agreement with those derived from the reference pyranometer in 85% of the cases (it seems that this agreement does not depend on aerosol size or optical depth). The agreement percentage goes down to only 48% when another algorithm, based on Red-Blue Ratio (RBR), is applied to the sky camera images. The retrieved cloud cover at Granada and Valladolid is compared with that registered by trained meteorological observers. CPC cloud cover is in agreement with the reference showing a slight overestimation and a mean absolute error around 1 okta. A major advantage of the CPC algorithm with respect to the RBR method is that the determined cloud cover is independent of aerosol properties. The RBR algorithm overestimates cloud cover for coarse aerosols and high loads. Cloud cover obtained only from ceilometer shows similar results than CPC algorithm; but the horizontal distribution cannot be obtained. In addition, it has been observed that under quick and strong changes on cloud cover ceilometers retrieve a cloud cover fitting worse with the real cloud cover.

Development and Hardware Implementation of an Efficient Algorithm for Cloud Detection From Satellite Images

Detecting clouds in satellite imagery is becoming more important with increasing data availability which are generated by earth observing satellites. Hence, intellectual processing of the enormous amount of data received by hundreds of earth receiving stations, with specific satellite image oriented approaches, presents itself as a pressing need. One of the most important steps in previous stages of satellite image processing is cloud detection. While there are many approaches that compact with different semantic meaning, there are rarely approaches that compact specifically with cloud and cloud cover detection. In this paper, the technique presented is the scene based adaptive cloud, cloud cover detection and find the position with assumption of sun reflection, background varying and scattering are constant. The capability of the developed system was tested using dedicated satellite images and assessed in terms of cloud percentage coverage. The system used for this process comprises of Intel(R) Xenon(R) CPU E31245 @ 3.30GHz processor along with MATLAB 13 software and DSPC6713 processor along with Code Compose Studio 3.1.

Spatial and temporal variations of summer surface temperatures of wet polygonal tundra in Siberia - implications for MODIS LST based permafrost monitoring

The surface temperature of permafrost soils in remote arctic areas is accessible by satellite land surface temperature (LST) detection. However, the spatial resolution of satellite measurements such as the MODIS LST products is limited and does not detect the heterogeneities of the wet polygonal tundra landscape where surface wetness varies over distances of several meters. This paper examines the spatial and temporal variability of summer surface temperatures of a polygonal tundra site in northern Siberia using a ground based high resolution thermal imaging system. Thermal infrared images were taken of a 1000 m 2 polygonal tundra area in 10 min intervals from July to September 2008. Under clear sky conditions, the individual measurements indicate temperature differences of up to 6 K between dry and wet tundra surfaces and which can exceed 12 K when dry tundra and water surfaces are compared. These differences disappear when temperature averages are considered for intervals longer than the diurnal cycle; for weekly averages the spatial temperature variability decreases below 1 K. The exception is the free water surface of a shallow polygonal pond where weekly averaged temperature differences of 2.5 K are sustained compared to the tundra surface. The ground based thermal infrared images are upscaled to MODIS sized pixels and compared to available MODIS LST data for individual measurements and weekly averages. The comparisons show generally good agreement for the individual measurements under clear sky conditions, which exist during 20% of the studied time period. However, several erroneous measurements and large data gaps occur in the MODIS LST data during cloudy conditions, leading to biased weekly temperature averages inferred from the satellite observations. Based on these results the following recommendations are given for future permafrost temperature monitoring based on MODIS LST products: (i) high resolution surface water masks for the quality assessment in landscapes where lakes and ponds are frequent and (ii) reliable cloud cover detection in conjunction with a gap filling procedure for accurate temporal averages.

Automatic cloud amount detection by surface longwave downward radiation measurements

Naked‐eye observation of sky cloud cover has widely resisted automation. Automatic cloud cover detection systems suitable also for nighttime operation often demand large equipment investments and expensive data processing. An automatic partial cloud amount detection algorithm (APCADA) is presented, based only on accurate measurements of longwave downward radiation, temperature, and relative humidity at screen level height. APCADA provides cloud cover estimates every 10 min during daytime and nighttime and is applicable to radiation stations without knowledge of synoptic cloud observations. Naked‐eye observations from seven radiation sites spanning from arctic to tropical climates have been compared to APCADA estimates. Results show that about 86% of all cases agree within ±1‐octa cloud amount difference for sites with moderate climate, 82% for sites with arctic climate, and 78% for the site with tropical climate. For a maximum ±2‐octa cloud amount difference, average site percentages range from 90% up to 95%.