Cloud Screening Algorithm Research Articles

A New Cloud Screening Algorithm for Ground-Based Direct-Beam Solar Radiation

AbstractCloud screening of direct-beam solar radiation is an essential step for in situ calibration and atmospheric properties retrieval. The internal cloud screening module of a Langley analysis program [Langley Analyzer (LA)] used by the U.S. Department of Agriculture (USDA) UV-B Monitoring and Research Program (UVMRP) is used for screening the uncalibrated direct-beam measurements and for deriving Langley offset voltages for calibration of the UV version of the Multifilter Rotating Shadowband Radiometer (UV-MFRSR). The current LA cloud screening module utilizes data from extended clear-sky periods and tends to ignore shorter periods that typify periods of broken cloudiness, and as a result, fewer values are generated for sites with higher frequencies of cloudy days (cloudy sites). A new cloud screening algorithm is presented that calculates the total optical depth (TOD) difference between a target point and pairs of points, and identifies the target as cloudy if the mean TOD difference exceeds a certain threshold. The screening is an iterative process that finishes when no new cloudy points are found. The result at a typical clear/sunny site shows that values from partly cloudy days are consistent with those from cloud-free days, when the new method is employed. The new cloud screening algorithm picks up significantly more values at cloudy sites. The larger decrease of the annual mean value of at cloudy sites than at relatively clear sites suggests the potential for improving calibration accuracy at cloudy sites. The results also show that the new cloud screening method is capable of detecting clear points in short clear windows and in transitional regions.

Retrieval of Land Surface Temperature Diurnal Cycle model parameters from Kalpana-1 VHRR data over India

Abstract. Land Surface Temperature (LST) is an important parameter in the land surface processes on regional and global scale. The Land Surface Temperature Diurnal (LSTD) cycle of different land cover is an excellent indicator of the surface processes and their interaction with planetary boundary layer. The Kalpana-1 very high resolution radiometer (VHRR) LST product is available with 30 minute spatial resolution and 0.1 degree temporal resolution. A study was carried out with an objective to determine the LSTD parameters directly from K1-VHRR monthly averaged LST observations over Indian landmass. In this analysis, a harmonic function is fitted to LSTD from the K1-VHRR observations, where cosine term describing the effect of sun and exponential term represents decay of LST during nighttime. Using LSTD parameters, one can directly know the temperature amplitude, residual temperature and time of maximum temperature for each pixel. The LSTD parameters fitting accuracy in root mean square error (RMSE) and coefficient of determination (R2) ranges between 0.5–2.5 K and 0.90–0.99 respectively for most of the pixels over Indian landmass. These LSTD parameters may bring new insight for estimation of thermal inertia and also useful in cloud screening algorithms.

Method for validating cloud mask obtained from satellite measurements using ground-based sky camera.

Error propagation in Earth's atmospheric, oceanic, and land surface parameters of the satellite products caused by misclassification of the cloud mask is a critical issue for improving the accuracy of satellite products. Thus, characterizing the accuracy of the cloud mask is important for investigating the influence of the cloud mask on satellite products. In this study, we proposed a method for validating multiwavelength satellite data derived cloud masks using ground-based sky camera (GSC) data. First, a cloud cover algorithm for GSC data has been developed using sky index and bright index. Then, Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data derived cloud masks by two cloud-screening algorithms (i.e., MOD35 and CLAUDIA) were validated using the GSC cloud mask. The results indicate that MOD35 is likely to classify ambiguous pixels as "cloudy," whereas CLAUDIA is likely to classify them as "clear." Furthermore, the influence of error propagations caused by misclassification of the MOD35 and CLAUDIA cloud masks on MODIS derived reflectance, brightness temperature, and normalized difference vegetation index (NDVI) in clear and cloudy pixels was investigated using sky camera data. It shows that the influence of the error propagation by the MOD35 cloud mask on the MODIS derived monthly mean reflectance, brightness temperature, and NDVI for clear pixels is significantly smaller than for the CLAUDIA cloud mask; the influence of the error propagation by the CLAUDIA cloud mask on MODIS derived monthly mean cloud products for cloudy pixels is significantly smaller than that by the MOD35 cloud mask.

Automated Retrieval of Cloud Masks from the HJ-1 WVC Imagery

Robust and automated cloud discrimination is regarded as an important step toward the extraction and analysis of cloud-free surface features, especially for high-resolution images only covering visible and near-infrared spectrum, like data from the wide view CCD cameras (WVC) of HJ-1 minisatellites. A cloud screening algorithm designed for the HJ-1 (China) multispectral WVC data is presented using a combination of Tasseled Cap (TC) transformation and an unsupervised classification known as the iterative self-organizing data analysis (ISODATA). Two filters, one of which is cluster-based and the other uses a pixel-based analysis that employs greenness index and wetness index both extracted from the TC transformation, are successively driven to filter out cloud-pixels. The performance of the proposed algorithm is investigated for six HJ-1 WVC scenes with typical cloud coverage. Two comparisons, respectively, with one method based on the combination of the TC transformation and linear constrains (TCLC) and the other using the maximum-likelihood classification (MLC), are given in this study. The initial results from both comparisons show an impressive agreement, especially for regions covered by thick clouds or large thin clouds. Although this agreement decreases significantly for other complex situations, such as the presence of numerous small thin clouds, overall agreements still stay on an acceptable level. A considerable contribution, therefore, can be expected by this method for those high-resolution data with less frequent observations and a narrow spectrum cover.

Comparisons between VIIRS cloud mask performance results from manually generated cloud masks of VIIRS imagery and CALIOP-VIIRS match-ups

Binary cloud masks generated from the manual interpretation of imagery serve as one source of truth data in the evaluation of the Visible Infrared Imaging Radiometer Suite (VIIRS) cloud mask (VCM) algorithm in the Suomi National Polar-orbiting Partnership (S-NPP) Calibration/Validation (CalVal) Program of the Joint Polar Satellite System (JPSS). The other and primary source of truth data used to establish global VCM performance comes from match-up datasets of VIIRS and CALIOP (Cloud-Aerosol Lidar with Orthogonal Polarization) data that are collected by the NASA CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation) system. While manually generated cloud masks have long been used to quantitatively evaluate the performance of individual cloud-screening detection tests, complete cloud-screening algorithms, and even the results of cloud forecast models, there has not previously been the opportunity to compare results obtained with these cloud masks to an independent, highly accurate cloud dataset, as collected by CALIOP. Such an opportunity was afforded the S-NPP VCM CalVal Team as it prepared to assess the VCM algorithm’s performance to meet the JPSS Validation Phase-1 performance criteria milestone. This article provides an in-depth discussion on the use of data generated to evaluate VCM performance and the results obtained from comparisons to both manually generated cloud masks and CALIOP-VIIRS match-up datasets. Overall, the performance of the VCM algorithm performance is found to be consistent with each source of truth data, and while the evaluation of the VCM algorithm is still in progress, it is already satisfying the JPSS Level-1 System Requirements. However, the similarities in VCM performance using these two sets of performance results were surprising and that becomes the focus of this article. As an example, the probability of correct typing (PCT) with the VCM algorithm during daytime conditions over ocean, land, and desert backgrounds, was 96.5%, 94.4%, and 95.7% respectively, based upon the manually generated cloud masks. Similar results obtained using CALIOP-VIIRS match-up datasets were 95.0%, 93.9%, and 96.0% respectively. It is concluded that manually generated cloud masks, created from VIIRS imagery, provide unique insights into the VCM’s performance, which results in a robust CalVal program when augmented with results obtained from CALIOP-VIIRS match-up data.

Critical evaluation of cloud contamination in the MISR aerosol products using MODIS cloud mask products

Abstract. Using Terra Moderate Resolution Imaging Spectroradiometer (MODIS)-based cloud screening methods, the impacts of cloud contamination on the Terra Multi-angle Imaging Spectroradiometer (MISR) aerosol optical depth (AOD) product are evaluated. Based on one year of collocated MISR and MODIS data, this study suggests that cloud contamination exists in both over-water and over-land MISR AOD data, with heavier cloud contamination occurring over the high latitude southern hemispheric oceans. On average globally, this study shows that thin cirrus cloud contamination introduces a possible ~ 0.01 high bias for the over-water MISR AOD retrievals. Over the mid- to high-latitude oceans and Southeast Asia, this number increases to 0.015–0.02. However, biases much larger than this mean value are found in individual retrievals, especially in retrievals that are near cloud edges. This study suggests that cloud-clearing methods using observations from MISR alone, which has only visible and near-infrared channels, may not be sufficient for all scenarios. Measurements from MODIS can be applied to assist cloud-clearing of the MISR aerosol retrievals. Cloud screening algorithms based on multi-sensor approaches are feasible and should be considered for current and future satellite aerosol studies.

Satellite observed widespread decline in Mongolian grasslands largely due to overgrazing

The Mongolian Steppe is one of the largest remaining grassland ecosystems. Recent studies have reported widespread decline of vegetation across the steppe and about 70% of this ecosystem is now considered degraded. Among the scientific community there has been an active debate about whether the observed degradation is related to climate, or over-grazing, or both. Here, we employ a new atmospheric correction and cloud screening algorithm (MAIAC) to investigate trends in satellite observed vegetation phenology. We relate these trends to changes in climate and domestic animal populations. A series of harmonic functions is fitted to Moderate Resolution Imaging Spectroradiometer (MODIS) observed phenological curves to quantify seasonal and inter-annual changes in vegetation. Our results show a widespread decline (of about 12% on average) in MODIS observed normalized difference vegetation index (NDVI) across the country but particularly in the transition zone between grassland and the Gobi desert, where recent decline was as much as 40% below the 2002 mean NDVI. While we found considerable regional differences in the causes of landscape degradation, about 80% of the decline in NDVI could be attributed to increase in livestock. Changes in precipitation were able to explain about 30% of degradation across the country as a whole but up to 50% in areas with denser vegetation cover (P<0.05). Temperature changes, while significant, played only a minor role (r(2) =0.10, P<0.05). Our results suggest that the cumulative effect of overgrazing is a primary contributor to the degradation of the Mongolian steppe and is at least partially responsible for desertification reported in previous studies.

Failure and Redemption of Multifilter Rotating Shadowband Radiometer (MFRSR)/Normal Incidence Multifilter Radiometer (NIMFR) Cloud Screening: Contrasting Algorithm Performance at Atmospheric Radiation Measurement (ARM) North Slope of Alaska (NSA) and Southern Great Plains (SGP) Sites

Well-known cloud-screening algorithms, which are designed to remove cloud-contaminated aerosol optical depths (AOD) from Multifilter Rotating Shadowband Radiometer (MFRSR) and Normal Incidence Multifilter Radiometer (NIMFR) measurements, have exhibited excellent performance at many middle-to-low latitude sites around world. However, they may occasionally fail under challenging observational conditions, such as when the sun is low (near the horizon) and when optically thin clouds with small spatial inhomogeneity occur. Such conditions have been observed quite frequently at the high-latitude Atmospheric Radiation Measurement (ARM) North Slope of Alaska (NSA) sites. A slightly modified cloud-screening version of the standard algorithm is proposed here with a focus on the ARM-supported MFRSR and NIMFR data. The modified version uses approximately the same techniques as the standard algorithm, but it additionally examines the magnitude of the slant-path line of sight transmittance and eliminates points when the observed magnitude is below a specified threshold. Substantial improvement of the multi-year (1999–2012) aerosol product (AOD and its Angstrom exponent) is shown for the NSA sites when the modified version is applied. Moreover, this version reproduces the AOD product at the ARM Southern Great Plains (SGP) site, which was originally generated by the standard cloud-screening algorithms. The proposed minor modification is easy to implement and its application to existing and future cloud-screening algorithms can be particularly beneficial for challenging observational conditions.

The calibration methods for Multi-Filter Rotating Shadowband Radiometer: a review

The continuous, over two-decade data record from the Multi-Filter Rotating Shadowband Radiometer (MFRSR) is ideal for climate research which requires timely and accurate information of important atmospheric components such as gases, aerosols, and clouds. Except for parameters derived from MFRSR measurement ratios, which are not impacted by calibration error, most applications require accurate calibration factor(s), angular correction, and spectral response function(s) from calibration. Although a laboratory lamp (or reference) calibration can provide all the information needed to convert the instrument readings to actual radiation, in situ calibration methods are implemented routinely (daily) to fill the gaps between lamp calibrations. In this paper, the basic structure and the data collection and pretreatment of the MFRSR are described. The laboratory lamp calibration and its limitations are summarized. The cloud screening algorithms for MFRSR data are presented. The in situ calibration methods, the standard Langley method and its variants, the ratio-Langley method, the general method, Alexandrov’s comprehensive method, and Chen’s multi-channel method, are outlined. The reason that all these methods do not fit for all situations is that they assume some properties, such as aerosol optical depth (AOD), total optical depth (TOD), precipitable water vapor (PWV), effective size of aerosol particles, or angstrom coefficient, are invariant over time. These properties are not universal and some of them rarely happen. In practice, daily calibration factors derived from these methods should be smoothed to restrain error.

Development of a new data-processing method for SKYNET sky radiometer observations

Abstract. In order to reduce uncertainty in the estimation of Direct Aerosol Radiative Forcing (DARF), it is important to improve the estimation of the single scattering albedo (SSA). In this study, we propose a new data processing method to improve SSA retrievals for the SKYNET sky radiometer network, which is one of the growing number of networks of sun-sky photometers, such as NASA AERONET and others. There are several reports that SSA values from SKYNET have a bias compared to those from AERONET, which is regarded to be the most accurate due to its rigorous calibration routines and data quality and cloud screening algorithms. We investigated possible causes of errors in SSA that might explain the known biases through sensitivity experiments using a numerical model, and also using real data at the SKYNET sites at Pune (18.616° N/73.800° E) in India and Beijing (39.586° N/116.229° E) in China. Sensitivity experiments showed that an uncertainty of the order of ±0.03 in the SSA value can be caused by a possible error in the ground surface albedo or solid view angle assumed for each observation site. Another candidate for possible error in the SSA was found in cirrus contamination generated by imperfect cloud screening in the SKYNET data processing. Therefore, we developed a new data quality control method to get rid of low quality or cloud contamination data, and we applied this method to the real observation data at the Pune site in SKYNET. After applying this method to the observation data, we were able to screen out a large amount of cirrus-contaminated data and to reduce the deviation in the SSA value from that of AERONET. We then estimated DARF using data screened by our new method. The result showed that the method significantly reduced the difference of 5 W m−2 that existed between the SKYNET and AERONET values of DARF before screening. The present study also suggests the necessity of preparing suitable a priori information on the distribution of coarse particles ranging in radius between 10 μm and 30 μm for the analysis of heavily dust-laden atmospheric cases.

Retrieval and variability analysis of optically thin cloud optical depths from a Cimel sun-photometer

In this work we propose a technique to retrieve optical depths for optically thin clouds. This method complements the current performance of AERONET for optically thick cloud measurements using data eliminated by the cloud-screening algorithm that are not useful to derive aerosol properties, therefore inexpensively increasing the capabilities of the sun-photometers. The proposed method is based on the computation of the apparent cloud optical depths and on a forward scattering correction method that exploits state-of-the-art ice cloud scattering models. This complementary procedure is applied to Cimel sun-photometer measurements performed at the Évora Geophysics Centre (Portugal, 38.6°N, 7.9°W, 293m asl) included in the AERONET network in order to obtain a climatology of optical depths for optically thin clouds over middle-latitude continental regions of the southwestern Iberian Peninsula. Main features regarding annual and seasonal variability and the relative changes in the infrared radiative forcing from 2007 to 2010 are also reported and analyzed.

Cloud screening and quality control algorithm for star photometer data: assessment with lidar measurements and with all-sky images

Abstract. This paper presents the development and set up of a cloud screening and data quality control algorithm for a star photometer based on CCD camera as detector. These algorithms are necessary for passive remote sensing techniques to retrieve the columnar aerosol optical depth, δAe(λ), and precipitable water vapor content, W, at nighttime. This cloud screening procedure consists of calculating moving averages of δAe(λ) and W under different time-windows combined with a procedure for detecting outliers. Additionally, to avoid undesirable δAe(λ) and W fluctuations caused by the atmospheric turbulence, the data are averaged on 30 min. The algorithm is applied to the star photometer deployed in the city of Granada (37.16° N, 3.60° W, 680 m a.s.l.; South-East of Spain) for the measurements acquired between March 2007 and September 2009. The algorithm is evaluated with correlative measurements registered by a lidar system and also with all-sky images obtained at the sunset and sunrise of the previous and following days. Promising results are obtained detecting cloud-affected data. Additionally, the cloud screening algorithm has been evaluated under different aerosol conditions including Saharan dust intrusion, biomass burning and pollution events.

Comparison of Cloud-Screening Methods Applied to GOSAT Near-Infrared Spectra

Several existing and proposed satellite remote sensing instruments are designed to derive concentrations of trace gases, such as carbon dioxide (CO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> ) and methane (CH <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> ), from measured spectra of reflected sunlight in absorption bands of the gases. Generally, these analyses require that the scenes be free of cloud and aerosol, necessitating robust screening algorithms. In this work, two cloud-screening algorithms are compared. One applies threshold tests, similar to those used by the MODerate resolution Imaging Spectrometer (MODIS), to visible and infrared reflectances measured by the Cloud and Aerosol Imager aboard the Greenhouse gases Observing SATellite (GOSAT). The second is a fast retrieval algorithm that operates on high-resolution spectra in the oxygen A-band measured by the Fourier Transform Spectrometer on GOSAT. Near-simultaneous cloud observations from the MODIS Aqua satellite are used for comparison. Results are expressed in terms of agreement and disagreement in the identification of clear and cloudy scenes for land and non-sun glint viewing over water. The accuracy, defined to be the fraction of scenes that are classified the same, is approximately 80% for both algorithms over land when comparing with MODIS. The accuracy rises to approximately 90% over ocean. Persistent difficulties with identifying cirrus clouds are shown to yield a large fraction of the disagreement with MODIS.

Tropical cirrus cloud contamination in sun photometer data

Abstract Cirrus clouds are endemic to Southeast Asia and are a source of potential bias in regional passive aerosol remote sensing datasets. Here, performance of the cloud-screening algorithm for the ground-based Aerosol Robotic Network (AERONET) sun photometer data is evaluated for cirrus cloud contamination at Singapore (1.30° N, 103.77° E). Using twelve months of concurrent AERONET Level 1.5 and 2.0 cloud-screened aerosol optical depth (AOD) data, and collocated Level 1.0 Micro-Pulse Lidar Network (MPLNET) measurements, we investigate the baseline AOD bias due to cirrus cloud presence. Observations are considered for a primary sample of all data and a secondary sample where AERONET data are restricted to a zenith viewing angle ≤ 45°. Cirrus clouds are present in zenith-viewing MPL profiles for 34% and 23% of these samples respectively. Based on approximations of cirrus cloud optical properties necessary to estimate cloud optical depth from the single-channel lidar signal, and assuming partial forward scattering of diffuse light by cirrus clouds into the sun photometer’s field of view, we estimate a range in AOD bias due to unscreened cloud presence of 0.034 to 0.060 and 0.031 to 0.055 ± 0.01 for the primary and secondary sample respectively. From the analysis of AERONET AOD for the angle-limited subset alone, we also derive a positive AOD bias of 0.034, which is comparable to the lower bounds for the estimated cloud bias reported for our datasets. These findings, which we attribute to the prevalence of cirrus clouds present from regional convection, are higher than previous reports of global AOD bias in the Moderate Resolution Infrared Spectroradiometer (MODIS) satellite-borne measurements due to residual cirrus cloud presence.

Investigation of GOSAT TANSO-CAI Cloud Screening Ability through an Intersatellite Comparison

AbstractIn this work, the Greenhouse Gases Observing Satellite (GOSAT) Thermal and Near-infrared Sensor for Carbon Observation–Cloud and Aerosol Imager (TANSO-CAI) cloud screening results, which are necessary for the retrieval of carbon dioxide (CO2) and methane (CH4) gas amounts from GOSAT TANSO–Fourier Transform Spectrometer (FTS) observations, are compared with results from Aqua/Moderate Resolution Imaging Spectroradiometer (MODIS) in four seasons. A large number of pixels, acquired from both satellites with nearly coincident locations and times, are extracted for statistical comparisons. The same cloud screening algorithm was applied to both satellite datasets to focus on the performance of the individual satellite sensors, without concern for differences in algorithms. The comparisons suggest that CAI is capable of discriminating between clear and cloudy areas over water without sun glint and also may be capable of identifying thin cirrus clouds, which are generally difficult to detect without thermal infrared or near-infrared bands. On the other hand, cloud screening over land by CAI resulted in greater cloudy discrimination than that by MODIS, whereas detection of thin cirrus clouds tended to be more difficult over land than water, resulting in incorrect identification of thin cirrus as clear. The amount of missed thin cirrus had a seasonal variation, with the maximum occurring in summer. The cloudy tendency of CAI over half vegetation is caused by lack of an effective threshold test that can be applied to MODIS. The statistical results of the comparison clarified the important points to consider when using the results of CAI cloud screening.

An automated day-time cloud detection technique applied to MSG-SEVIRI data over Western Europe

Since February 2003, the Spinning Enhanced Visible and Infrared Imager (SEVIRI) aboard the first Meteosat Second Generation (MSG) satellite has provided radiance data in 12 spectral bands for a full Earth hemisphere every 15 minutes. This high frame rate renders it an excellent tool for studies of atmospheric transport of pollutants, aerosol and clouds. TNO (Netherlands Organisation for Applied Scientific Research) is currently developing an algorithm for the retrieval of aerosol properties from MSG-SEVIRI observations over cloud-free scenes. This requires rigorous cloud screening for which a fast and stand-alone algorithm is developed. The detection technique described in this paper, which is based on the ATSR-2 (Along Track Scanning Radiometer 2) cloud screening algorithm, can be easily implemented, and satisfactorily identifies clouds. The study presented here focuses on Western Europe for the year 2006. Cloud detection results are compared to the KNMI/MF (Royal Netherlands Meteorological Institute/Meteo-France) and the Moderate Resolution Imaging Spectroradiometer (MODIS) cloud detection algorithms. According to the statistics, the results obtained with our algorithm show good agreement (>80%) with these data sets.

Aerosol Optical Depth measurements at 340 nm with a Brewer spectrophotometer and comparison with Cimel sunphotometer observations at Uccle, Belgium

Abstract. The Langley Plot Method (LPM) is adapted for the retrieval of Aerosol Optical Depth (AOD) values at 340 nm from Brewer#178 sun scan measurements between 335 and 345 nm (convoluted with the band pass function of the Cimel sunphotometer filter at 340 nm) performed in Uccle, Belgium. The use of sun scans instead of direct sun measurements simplifies the comparison of the AOD values with quasi-simultaneous Cimel sunphotometer values. Also, the irradiance at 340 nm is larger than the one at 320.1 nm due to lower ozone absorption, thus improving the signal to noise ratio. For the selection of the cloudless days (from now on referred to as calibration quality clear days), a new set of criteria is proposed. With the adapted method, individual clear sky AOD values, for which the selection criteria are also presented in this article, are calculated for a period from September 2006 until the end of August 2010. These values are then compared to quasi-simultaneous Cimel sunphotometer measurements, showing a very good agreement (the correlation coefficient, the slope and the intercept of the regression line are respectively 0.974, 0.968 and 0.011), which proves that good quality observations can be obtained from Brewer sun scan measurements at 340 nm. The analysis of the monthly and seasonal Brewer AODs at Uccle is consistent with studies at other sites reporting on the seasonal variation of AODs in Europe. The highest values can be observed in summer and spring, whereas more than 50% of the winter AODs are lower than 0.3. On a monthly scale, the lowest AOD are observed in December and the highest values occur in June and April. No clear weekly cycle is observed for Uccle. The current cloud-screening algorithm is still an issue, which means that some AOD values can still be influenced by scattered clouds. This effect can be seen when comparing the calculated monthly mean values of the Brewer with the AERONET measurements.

Neural network cloud screening algorithm Part II: global synthetic cases using high resolution spectra in O&lt;sub&gt;2&lt;/sub&gt; and CO&lt;sub&gt;2&lt;/sub&gt; near infrared absorption bands in nadir and sun glint

In Part I a set of two layer feed-forward neural networks, trained via back propagation of sensitivities, was applied to a synthetic set of radiances in micro-windows of the near-infrared to make predictions of cloud water (cw), cloud ice (ci), effective scattering heights of cloud water and ice, (pcw and pci, respectively) and the column water vapor (w). A threshold test, using 2 g/m-2 for cloud water and 10 g/m-2 for cloud ice, was applied to the retrieved values to distinguish clear from cloudy scenes. In that work the discussion was limited to the nadir viewing geometry, and was applied only to land surfaces, excluding desert and snow and ice fields. Part II describes the extension to a set of high resolution radiances, as might be measured by a grating spectrometer from space, in both nadir and sun glint viewing geometries. Furthermore, results are given for all land surface types as well as scenes over ocean. Prior to neural network training, a Principal Component Analysis (PCA) is applied to the high resolution spectra, which consist of three bands centered at 0.76µm (O2 A-band), 1.61µm (weak CO2 band) and 2.06µm (strong CO2 band), each with 1016 channels. Analysis shows that the five leading EOFs together capture 99.9% of the variance in each band, reducing the data size by more than two orders of magnitude. Application of the trained neural networks to an independent data set, generated using CloudSat and Calipso cloud and aerosol profiles, as well as carbon dioxide profiles from a chemical transport model, were used to quantify the skill in the retrieval. The results vary significantly with surface type, viewing mode and cloud properties. Accuracies range from 7% to 100% (typically close to 75%), with confidence levels almost always greater than 90%.

The Relation between Moderate Resolution Imaging Spectroradiometer (MODIS) Aerosol Optical Depth and PM2.5 over the United States: A Geographical Comparison by U.S. Environmental Protection Agency Regions

Aerosol optical depth (AOD) acquired from satellite measurements demonstrates good correlation with particulate matter with diameters less than 2.5 µm (PM2.5) in some regions of the United States and has been used for monitoring and nowcasting air quality over the United States. This work investigates the relation between Moderate Resolution Imaging Spectroradiometer (MODIS) AOD and PM2.5 over the 10 U.S. Environmental Protection Agency (EPA)-defined geographic regions in the United States on the basis of a 2-yr (2005–2006) match-up dataset of MODIS AOD and hourly PM2.5 measurements. The AOD retrievals demonstrate a geographical and seasonal variation in their relation with PM2.5. Good correlations are mostly observed over the eastern United States in summer and fall. The southeastern United States has the highest correlation coefficients at more than 0.6. The southwestern United States has the lowest correlation coefficient of approximately 0.2. The seasonal regression relations derived for each region are used to estimate the PM2.5 from AOD retrievals, and it is shown that the estimation using this method is more accurate than that using a fixed ratio between PM2.5 and AOD. Two versions of AOD from Terra (v4.0.1 and v5.2.6) are also compared in terms of the inversion methods and screening algorithms. The v5.2.6 AOD retrievals demonstrate better correlation with PM2.5 than v4.0.1 retrievals, but they have much less coverage because of the differences in the cloud-screening algorithm.

Neural network cloud screening algorithm, Part I: A synthetic case over land using micro-windows in O&lt;sub&gt;2&lt;/sub&gt;&amp;gt; and CO&lt;sub&gt;2&lt;/sub&gt;&amp;gt; near infrared absorption bands with nadir viewing

A neural network is presented for estimating cloud water and ice paths, effective scattering heights of cloud water and ice, and column water vapor. The cloud water and ice are then used to classify scenes as either clear or cloudy using a simple threshold test of 2 gm−2 for water and 10 gm−2 for ice. Training of the neural networks was performed using high resolution spectra in micro-windows of O2 and CO2 near infrared absorption bands generated from an ensemble of analyzed meteorological fields from ECMWF and surface properties from MODIS. An independent test data set was generated using the same radiative transfermodel, but coupled with atmospheric profiles derived from CloudSat and Calipso data. Analysis indicates that the algorithmprovides approximately 75-90% accuracy with a 95-99% confidence level for classifying scenes as either cloudy or clear over land surfaces in nadir viewing geometry. These estimates are shown to be robust, in the sense that they are insensitive to realistic instrumental errors, errors in the meteorological analyses and surface properties, and errors in the simulations used for training.