Kaixu Bai

Smart Information Reconstruction via Time-Space-Spectrum Continuum for Cloud Removal in Satellite Images

Cloud contamination is a big obstacle when processing satellite images retrieved from visible and infrared spectral ranges for application. Although computational techniques including interpolation and substitution have been applied to recover missing information caused by cloud contamination, these algorithms are subject to many limitations. In this paper, a novel smart information reconstruction (SMIR) method is proposed, in order to reconstruct cloud contaminated pixel values from the time-space-spectrum continuum with the aid of a machine learning tool, namely extreme learning machine (ELM). For the purpose of demonstration, the performance of SMIR is evaluated by reconstructing the missing remote sensing reflectance values derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) on board the Terra satellite over Lake Nicaragua, where is a very cloudy area year round. For comparison, the traditional backpropagation neural network algorithms will also be implemented to reconstruct the missing values. Experimental results show that the ELM outperforms the BP algorithms by an enhanced machine learning capacity with simulated memory effect embedded in MODIS due to linking the complex time-space-spectrum continuum between cloud-free and cloudy pixels. The ELM-based SMIR practice presents a correlation coefficient of 0.88 with root mean squared error of 7.4E - 04sr-1 between simulated and observed reflectance values. Finding suggests that the SMIR method is effective to reconstruct all the missing information providing visually logical and quantitatively assured images for further image processing and interpretation in environmental applications.

Global validation of FY-3A total ozone unit TOU total ozone columns using ground-based Brewer and Dobson measurements

Analysis of the accuracy and variability of total ozone columns (TOC) has been conducted by many studies, while the TOC observations derived from the total ozone unit (TOU) on board the Chinese FengYun-3A (FY-3A) satellite platform are notably less well documented. Therefore, in this present study, we mainly focus on the global-scale validation of TOU-derived total ozone column data by comparing them with spatially and temporally co-located ground-based measurements from the well-established Brewer and Dobson spectrophotometer for the period July 2009 through December 2011. The results show that TOU-derived total ozone column data yields high accuracy, with the root mean square error less than 5% in comparison with ground-based measurements. Meanwhile, TOU underestimates Brewer measurements by 1.1% in the Northern Hemisphere and overestimates Dobson total ozone 0.3% globally. In addition, TOU-derived total ozone shows no significant dependence on latitude in comparison with either Brewer or Dobson total ozone measurements. Nevertheless, a significant dependence of TOU-derived total ozone is observed on the solar zenith angle (SZA) in comparison with both Brewer and Dobson, demonstrating that TOU underestimates at large SZA and overestimates at small SZA. Finally, the dependence of satellite – ground-based relative difference for total ozone values shows fair agreement when total ozone values are in the range 250–450 Dobson units (DU). Overall, the Chinese FY-3A/TOU performs well on total ozone retrieval with high accuracy, and the total ozone data derived from the TOU can be used as a reliable data source for ozone monitoring and other atmospheric applications.

Spectral Information Adaptation and Synthesis Scheme for Merging Cross-Mission Ocean Color Reflectance Observations From MODIS and VIIRS

Obtaining a full clear view of coastal bays, estuaries, lakes, and inland waters is challenging with single satellite sensor observations due to cloud impacts. Cross-mission sensors provide the synergistic opportunity to improve spatial and temporal coverage by merging their observations; however, discrepancies originating from the instrumental, algorithmic, and temporal differences should be eliminated before merging. This paper presents the Spectral Information Adaptation and Synthesis Scheme (SIASS) for generating cross-mission consistent ocean color reflectance by merging 2012-2015 observations from Moderate Resolution Imaging Spectroradiometer and Visible Infrared Imaging Radiometer Suite over Lake Nicaragua in Central America, where the cloud impact is salient. The SIASS is able to not only eliminate incompatibilities for matchup bands but also reconstruct spectral information for mismatched bands among sensors. Statistics indicate that the average monthly coverage of a merged ocean color reflectance product over Lake Nicaragua is nearly twice that of any single-sensor observation. Results show that SIASS significantly improves consistency among cross-mission sensors by mitigating prominent discrepancies. In addition, reconstructed spectral information for those mismatched bands help preserve more spectral characteristics needed to better monitor and understand the dynamic aquatic environment. The final implementation of SIASS to map the chlorophyll-α concentration demonstrates the efficacy of SIASS in bias correction and consistency improvement. In general, SIASS can be applied to remove cross-mission discrepancies among sensors to improve the overall consistency.

Retrieval of columnar water vapor using multispectral radiometer measurements over northern China

Water vapor is an important component in hydrological processes that basically involve all types of seasons, including dry (e.g., drought) or wet (e.g., hurricane or monsoon). This study retrieved columnar water vapor (CWV) with the 939.3 nm band of a multifilter rotating shadowband radiometer (MFRSR) using the modified Langley technique. Such an investigation was in concert with the use of the atmospheric transmission model MODTRAN for determining the instrument coefficients required for CWV estimation. Results of the retrieval of CWV by MFRSR from September 23, 2004 to June 20, 2005 at the XiangHe site are presented and analyzed in this paper. To improve the credibility, the MFRSR results were compared with those obtained from the AErosol RObotic NETwork CIMEL sun-photometer measurements, co-located at the XiangHe site, and the moderate resolution imaging spectroradiometer (MODIS) near-infrared total precipitable water product (MOD05), respectively. These comparisons show good agreement in terms of correlation coefficients, slopes, and offsets, revealing that the accuracy of CWV estimation using the MFRSR instrument is reliable and suitable for extended studies in northern China.

Comparison of Suomi-NPP OMPS total column ozone with Brewer and Dobson spectrophotometers measurements

The objective of this study is to evaluate the accuracy of the daily nadir total column ozone products derived from the nadir mapper instrument on the Ozone Mapping and Profiler Suite (OMPS) flying onboard the Suomi National Polar-orbiting Partnership satellite (S-NPP) launched as a part of the Joint Polar Satellite System (JPSS) program between NOAA and NASA. Since NOAA is already operationally processing OMPS nadir total ozone products, evaluations were made in this study on the total column ozone research products generated by NASA’s science team, utilizing the latest version of their Backscatter Ultraviolet (BUV) retrieval algorithms, to provide insight into the performance of the operation system. Comparisons were made with globally distributed ground-based Brewer and Dobson spectrophotometer total column ozone measurements. Linear regressions show fair agreement between OMPS and ground-based total column ozone measurements with a root-mean-square error (RMSE) of approximately 3% (10 DU). The comparison results indicate that the OMPS total column ozone data are 0.59% higher than the Brewer measurements with a standard deviation of 2.82% while 1.09% higher than the Dobson measurements with a standard deviation of 3.27%. Additionally, the variability of relative differences between OMPS and ground total column ozone were analyzed as a function of latitude, time, viewing geometry, and total column ozone value. Results show a 2% bias over most latitudes and viewing conditions when total column ozone value varies between 220 DU and 450 DU.

Diagnosis of the artificial intelligence-based predictions of flow regime in a constructed wetland for stormwater pollution control

Abstract Monitoring the velocity field and stage variations in heterogeneous aquatic environments, such as constructed wetlands, is critical for understanding hydrodynamic patterns, nutrient removal capacity, and hydrographic impact on the wetland ecosystem. Obtaining low velocity measurements representative of the entire wetland system may be challenging, expensive, and even infeasible in some cases. Data-driven modeling techniques in the computational intelligence regime may provide fast predictions of the velocity field based on a handful of local measurements. They can be a convenient tool to visualize the general spatial and temporal distribution of flow magnitude and direction with reasonable accurancy in case regular hydraulic models suffer from insufficient baseline information and longer run time. In this paper, a comparison between two types of bio-inspired computational intelligence models including genetic programming (GP) and artificial neural network (ANN) models was implemented to estimate the velocity field within a constructed wetland (i.e., the Stormwater Treatment Area in South Florida) in the Everglades, Florida. Two different ANN-based models, including back propagation algorithm and extreme learning machine, were used. Model calibration and validation were driven by data collected from a local sensor network of Acoustic Doppler Velocimeters (ADVs) and weather stations. In general, the two ANN-based models outperformed the GP model in terms of several indices. Findings may improve the design and operation strategies for similar wetland systems.

Multisensor Satellite Image Fusion and Networking for All-Weather Environmental Monitoring

Given the advancements of remote sensing technology, large volumes of remotely sensed images with different spatial, temporal, and spectral resolutions are available. To better monitor and understand the changing Earths environment, fusion of remotely sensed images with different spatial, temporal, and spectral resolutions is critical for distinctive feature retrieval, interpretation, mapping, and decision analysis. A suite of methods have been developed to fuse multisensor satellite images for different purposes in the past few decades. This paper provides a thorough review of contemporary and classic image fusion methods and presents a summary of their phenomenological applications, with challenges and perspectives, for environmental systems analysis. Cross-mission satellite image fusion, networking, and missing value pixel reconstruction for environmental monitoring are described, and their complex integration is illustrated with a case study of Lake Nicaragua that elucidates the state-of-the-art remote sensing technologies for advancing water quality management.

An intercomparison of multidecadal observational and reanalysis data sets for global total ozone trends and variability analysis

A four-step adaptive ozone trend estimation scheme is proposed by integrating multivariate linear regression (MLR) and ensemble empirical mode decomposition (EEMD) to analyze the long-term variability of total column ozone from a set of four observational and reanalysis total ozone data sets, including the rarely explored ERA-Interim total ozone reanalysis, from 1979 to 2009. Consistency among the four data sets was first assessed, indicating a mean relative difference of 1% and root-mean-square error around 2% on average, with respect to collocated ground-based total ozone observations. Nevertheless, large drifts with significant spatiotemporal inhomogeneity were diagnosed in ERA-Interim after 1995. To emphasize long-term trends, natural ozone variations associated with the solar cycle, quasi-biennial oscillation, volcanic aerosols, and El Nino–Southern Oscillation were modeled with MLR and then removed from each total ozone record, respectively, before performing EEMD analyses. The resulting rates of change estimated from the proposed scheme captured the long-term ozone variability well, with an inflection time of 2000 clearly detected. The positive rates of change after 2000 suggest that the ozone layer seems to be on a healing path, but the results are still inadequate to conclude an actual recovery of the ozone layer, and more observational evidence is needed. Further investigations suggest that biases embedded in total ozone records may significantly impact ozone trend estimations by resulting in large uncertainty or even negative rates of change after 2000.

Preliminary study of a dust event over Beijing by using satellite data and ground-based measurements

This paper discusses the analysis of the severe dust storm that occurred over Beijing from 26th April to 3rd May in 2012 with the use of combined satellite observations and ground-based measurements. In this study, we analyze the pollution characteristics of particulate matters near ground, with the main focus on spatio-temporal and vertical distributions of aerosol during this event by using ground-based Aerosol Robotic Network (AERONET), MODerate resolution Imaging Spectroradiometer (MODIS) and Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) data. Results show that the Aerosol Optical Depth (AOD) measured at 550 nm from the AERONET Beijing station has an ascending trend with a peak value of 2.5 on 1st May. Moreover, the AOD variation from the MODIS data agrees well with AERONET observations during the same time period. In addition, the vertical distribution of total attenuated backscatter coefficient (TABC), volume depolarization ratio (VDR) and color ratio (CR) of CALIPSO data are comprehensively analyzed. Results from these analyses show that the dust mainly accumulates in the layer at altitudes of 1.5 to 4.5 km on 1st May. In this dust layer, the values of TABC are generally around 0.002~0.0045 km-1sr-1 and VDR and CR are typically around 0.1~0.5 and 0.6~1.4 respectively. Thus, the combined satellite and ground-based observations are of great use for monitoring and analyzing air quality with high accuracy.

Evaluation of the consistency of OMI-TOMS total ozone with collocated ground-based measurements

As Ozone Monitoring Instrument (OMI) onboard the Aura satellite has provided global scale ozone measurements on a daily basis since 2004, the long-term stability and consistency of ozone retrievals is thus of critical importance, especially for the ozone recovery assessment. This study aims to evaluate the long-term stability of total ozone derived from the OMI Total Ozone Mapping Spectrometer (OMI-TOMS) algorithm, by comparing with collocated ground-based total ozone measurements recorded from 42Dobson spectrophotometers during the period 2004-2015. It is indicative that the OMI-TOMS total ozone is in good agreement with collocated ground-based measurements, with a R2 of 0.96 and root mean square error (RMSE) of 3.3%. Further investigations show that the OMI-TOMS total ozone is of quality, as no significant latitude dependence is observed. In the past 12 years, the OMI-TOMS total ozone is highly consistent with the ground-based Dobson total ozone, with a variation of mean relative difference less than 1%. In general, the OMI-TOMS total ozone performs well and can be used with confidence.