Detecting Multilayer Clouds From the Geostationary Advanced Himawari Imager Using Machine Learning Techniques

Tan, Zhonghui; Liu, Chao; Ma, Shuo; Wang, Xin; Shang, Jian Ku; Wang, Jianjie; Ai, Weihua; Yan, Wei

doi:10.1109/tgrs.2021.3087714

Cited by 19 publications

(13 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Sun-Mack et al (2017) and Minnis et al (2019) found that a larger separation distance can result in greater accuracy but at the expense of missing a significant number of actual ML clouds. Tan et al (2022) have demonstrated that detection accuracy is significantly compromised when the separation distance is less than 1 km.…”

Section: Methodsmentioning

confidence: 99%

“…Its impact on MLANN, examined by Sun-Mack et al ( 2017) and Minnis et al (2019), is similar to that from other studies. Tan et al (2022), for example, found that the probability of ML detection using a random forest method was greatest for separation differences of 3 km or more and that it dropped from values exceeding 0.8 to less than 0.5 for cloud gaps smaller than 1 km. Greater discrepancies in altitude between the upper and lower clouds increase the differences in the layer temperatures yielding stronger signals in the thermal channels.…”

Section: Dependence On Cloud Propertiesmentioning

confidence: 99%

“…While those results are quite encouraging, the approach needs further refinement and global coverage. Tan et al (2022) applied several machine-learning methods to a geostationary satellite dataset and found even greater accuracy with a random-forest technique. It is clear that artificial intelligence techniques are the most promising approach to detecting ML clouds from passive imagery.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Identification of ice-over-water multilayer clouds using multispectral satellite data in an artificial neural network

Sun-Mack,

Minnis,

Chen

et al. 2023

Preprint

View full text Add to dashboard Cite

Abstract. An artificial neural network (ANN) algorithm, employing several Aqua MODerate-resolution Imaging Spectroradiometer (MODIS) channels, the retrieved cloud phase and total cloud visible optical depth, and temperature and humidity vertical profiles is trained to detect multilayer (ML) ice-over-water cloud systems identified by matched 2008 CloudSat and CALIPSO (CC) data. The trained MLANN was applied to 2009 MODIS data resulting in combined ML and single layer detection accuracies of 87 % (89 %) and 86 % (89 %) for snow-free (snow-covered) regions during the day and night, respectively. Overall, it detects 55 % and ~30 % of the CC ML clouds over snow-free and snow-covered surfaces, respectively, and has a relatively low false alarm rate. The net gain in accuracy, which is the difference between the true and false ML fractions, is 7.5 % and ~2.0 % over snow-free and snow/ice-covered surfaces. Overall, the MLANN is more accurate than most currently available methods. When corrected for the viewing-zenith-angle dependence of each parameter, the ML fraction detected is relatively invariant across the swath. Compared to the CC ML variability, the MLANN is robust seasonally and interannually, and produces similar distribution patterns over the globe, except in the polar regions. Additional research is needed to conclusively evaluate the VZA dependence and further improve the MLANN accuracy. This approach should greatly improve the monitoring of cloud vertical structure using operational passive sensors.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Dependence On Cloud Propertiesmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Identification of ice-over-water multilayer clouds using multispectral satellite data in an artificial neural network

Sun-Mack,

Minnis,

Chen

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The RF algorithm is an ML method widely used in the inversion of meteorological elements and has been proven to perform well in applications such as the estimation of precipitation, detection of clouds, and inversion of PM2.5 concentrations [Oscar et al, 2020;Tan et al, 2021;Liu et al, 2021;Guo et al, 2021]. Thus, this study used this simple but promising RF algorithm to establish the nonlinear relationships among surface precipitation, satellite observations and atmospheric characteristics.…”

Section: Establishment Of Surface Precipitation Integration Algorithmmentioning

confidence: 99%

High-resolution typhoon precipitation integrations using satellite infrared observations and multi-source data

Zhao

Liu

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Abstract. Typhoon-related precipitation over land can result in severe disasters such as floods and landslides, and satellites are a valuable tool to estimate surface precipitation with high spatial-temporal resolutions. This study develops a high-resolution surface precipitation integration framework to combine observations from geostationary Fengyun-4A/ Himawari-8 (F4/H8) satellite radiometers, high-density rain-gauge observations (or Integrated Multisatellite Retrievals for GPM (IMERG) data) and atmospheric reanalysis (ERA5) based on a random forest (RF) algorithm. The RF methods fuse cloud and atmospheric features from radiometric observations and reanalysis information, and the intensity and spatial distribution of rainfall can be revealed by high-density rain-gauge or IMERGE observations. We take three typhoons made landfall in South China in 2018 as examples. Both F4- and H8-based results using rain-gauge data as the predictand show excellent results with correlation coefficients (R) with the references ~0.75 and probabilities of detection (POD) as large as 0.98, higher than current satellite-only results. However, if the IMERG data are used as the predictand, the corresponding R and POD drop to ~0.5 and 0.93, respectively, due to the uncertainties related to IMERG retrievals. By carefully choosing the predictors, the RF algorithm can clearly summarize the information from satellite observations, surface observations and atmospheric reanalysis, resulting in precipitation results that are highly consistent with the actual ground observations. Our proposed integration framework can not only reconstruct hourly surface precipitation estimation datasets at high-spatial resolution for historical typhoon studies but also potentially monitor the fine- resolution surface precipitation of frequent typhoons in near-real time.

show abstract

“…Researchers have commenced extensive studies using infrared data to more accurately monitor cloud evolution at night, eliminating the need for visible data and achieving successful research outcomes [15]. Tan et al [16] proposed a nighttime cloud classification algorithm based on Himawari-8 satellite channel data and machine learning algorithms. They utilized data from 5 infrared channels, 3 brightness temperature difference (BTD) datasets, and latitude/longitude information as training features.…”

Section: Introductionmentioning

confidence: 99%

A Machine-Learning-Based Study on All-Day Cloud Classification Using Himawari-8 Infrared Data

Fu,

Mi,

Han

et al. 2023

Remote Sensing

View full text Add to dashboard Cite

Clouds are diverse and complex, making accurate cloud type identification vital in improving the accuracy of weather forecasting and the effectiveness of climate monitoring. However, current cloud classification research has largely focused on daytime data. The lack of visible light data at night presents challenges in characterizing nocturnal cloud attributes, leading to difficulties in achieving continuous all-day cloud classification results. This study proposed an all-day infrared cloud classification model (AInfraredCCM) based on XGBoost. Initially, the latitude/longitude, 10 infrared channels, and 5 brightness temperature differences of the Himawari-8 satellite were selected as input features. Then, 1,314,275 samples were collected from the Himawari-8 full-disk data and cloud classification was conducted using the CPR/CALIOP merged cloud type product as training data. The key cloud types included cirrus, deep convective, altostratus, altocumulus, nimbostratus, stratocumulus, stratus, and cumulus. The cloud classification model achieved an overall accuracy of 86.22%, along with precision, recall, and F1-score values of 0.88, 0.84, and 0.86, respectively. The practicality of this model was validated across all-day temporal, daytime/nighttime, and seasonal scenarios. The results showed that the AInfraredCCM consistently performed well across various time periods and seasons, confirming its temporal applicability. In conclusion, this study presents an all-day cloud classification approach to obtain comprehensive cloud information for continuous weather monitoring, ultimately enhancing weather prediction accuracy and climate monitoring.

show abstract

Detecting Multilayer Clouds From the Geostationary Advanced Himawari Imager Using Machine Learning Techniques

Cited by 19 publications

References 37 publications

Identification of ice-over-water multilayer clouds using multispectral satellite data in an artificial neural network

Identification of ice-over-water multilayer clouds using multispectral satellite data in an artificial neural network

High-resolution typhoon precipitation integrations using satellite infrared observations and multi-source data

A Machine-Learning-Based Study on All-Day Cloud Classification Using Himawari-8 Infrared Data

Contact Info

Product

Resources

About