Jianwen Ai

Quantitative Retrieval of Geophysical Parameters Using Satellite Data

The remote sensing information service grid node (RSIN) is a tool for dealing with climate change and quantitative environmental monitoring. Based on the high-throughput computing grid, RSIN enables a workflow management system for data placement. The accompanying unified data-and-computation-schedule algorithm helps load balancing between and within workflow steps.

Validation and analysis of aerosol optical thickness retrieval over land

Aerosol optical thickness (AOT) retrieval from Moderate Resolution Imaging Spectroradiometer (MODIS) data has been well established over oceans, but this is not the case over land. In this article, the AOT data sets retrieved by exploiting the synergy of TERRA and AQUA MODIS data (SYNTAM) over land are validated with ground-based measurements from Aerosol Robotic Network (AERONET) data, as well as from the National Aeronautics and Space Administration (NASA) AOT products, amended with a DeepBlue algorithm in Asian (15–60° N and 35–150° E) and American areas (30–40° N and 100–120° W). Overall, AOT retrieval errors of around 10–20% against AERONET data are found at both 1 and 10 km resolutions. The spectral and spatial sensitivities of the AOT correlation are explicitly addressed at both 1 and 10 km resolutions. Three window sizes, 1 × 1, 3 × 3 and 5 × 5, are tested for SYNTAM to evaluate the effect of window size on parameter statistics, and it is found that the accuracy of the SYNTAM method decreases with increasing window size. The validations at three spectral bands of 0.47, 0.55 and 0.66 μm show that the accuracies of different bands are 80–90% similar, and that the band at 0.47 μm has the highest accuracy most of the time. Comparisons between AOT data sets derived from the SYNTAM and AOT products from the NASA Dark Dense Vegetation (DDV) and the DeepBlue algorithms are also conducted using data from the USA. More pixels with AOT values for the area could be retrieved using the SYNTAM method with the NASA DeepBlue algorithm. The AOT values of more than 90% of pixels derived by both methods are very close. This clearly shows that AOT data from SYNTAM are very close to the AOT data set from the NASA DeepBlue algorithm in cloud-free areas. The synergic use of both the SYNTAM and DeepBlue algorithms could produce AOT values over much greater land areas.

Grid-enabled high-performance quantitative aerosol retrieval from remotely sensed data

As the quality and accuracy of remote-sensing instruments improve, the ability to quickly process remotely sensed data is in increasing demand. Quantitative remote-sensing retrieval is a complex computing process because of the terabytes or petabytes of data processed and the tight-coupling remote-sensing algorithms. In this paper, we intend to demonstrate the use of grid computing for quantitative remote-sensing retrieval applications with a workload estimation and task partition algorithm. Using a grid workflow for the quantitative remote-sensing retrieval service is an intuitive way to use the grid service for users without grid expertise. A case study showed that significant improvement in the system performance could be achieved with this implementation. The results of the case study also give a perspective on the potential of applying grid computing practices to remote-sensing problems.

Prior knowledge-supported aerosol optical depth retrieval over land surfaces at 500 m spatial resolution with MODIS data

Aerosol optical depth (AOD) values at a spatial resolution of 500 m were retrieved over terrain areas by applying a time series of Moderate resolution Imaging Spectroradiometer (MODIS) 500 m resolution data in the Heihe region (36–42° N, 97–104° E) of Gansu Province, China; in the Pearl River Delta (18–30° N, 108–122° E), China; and in Beijing (39–41° N, 115–118° E), China. A novel prior knowledge scheme was used in the algorithm that performs cloud screening, simultaneous AOD and surface reflectance retrieval from the MODIS 500 m Level 1B data. This prior knowledge scheme produced a new Ångström exponent α, utilizing a Terra pass time α and an Aqua pass time α to better satisfy the invariant α assumption. The retrieved AOD data were compared with AOD data observed with the ground-based, automatic Sun-tracking photometer CE318 at corresponding bands in the Heihe region and with Aerosol Robotic Network (AERONET) data in the Pearl River Delta and in Beijing. Validation experiments demonstrated the potential of applying the algorithm to MODIS 500 m AOD retrieval on land; validation showed the uncertainty of Δτ = ±0.1±0.2τ over various types of underlying land surface, including cities, where τ is the aerosol optical depth. The root mean square errors (RMSEs) were around 0.1 for inland regions and up to 0.24 for cities by the sea, such as Hong Kong and Zhongshan, China.

A high throughput geocomputing system for remote sensing quantitative retrieval and a case study

Abstract The quality and accuracy of remote sensing instruments have been improved significantly, however, rapid processing of large-scale remote sensing data becomes the bottleneck for remote sensing quantitative retrieval applications. The remote sensing quantitative retrieval is a data-intensive computation application, which is one of the research issues of high throughput computation. The remote sensing quantitative retrieval Grid workflow is a high-level core component of remote sensing Grid, which is used to support the modeling, reconstruction and implementation of large-scale complex applications of remote sensing science. In this paper, we intend to study middleware components of the remote sensing Grid – the dynamic Grid workflow based on the remote sensing quantitative retrieval application on Grid platform. We designed a novel architecture for the remote sensing Grid workflow. According to this architecture, we constructed the Remote Sensing Information Service Grid Node (RSSN) with Condor. We developed a graphic user interface (GUI) tools to compose remote sensing processing Grid workflows, and took the aerosol optical depth (AOD) retrieval as an example. The case study showed that significant improvement in the system performance could be achieved with this implementation. The results also give a perspective on the potential of applying Grid workflow practices to remote sensing quantitative retrieval problems using commodity class PCs.

Workload and task management of Grid-enabled quantitative aerosol retrieval from remotely sensed data

As the quality and accuracy of remote sensing instruments improve, the ability to quickly process remotely sensed data is in increasing demand. Quantitative retrieval of aerosol properties from remotely sensed data is a data-intensive scientific application, where the complexities of processing, modeling and analyzing large volumes of remotely sensed data sets have significantly increased computation and data demands. While Grid computing has been a prominent technique to tackle computational issues, little work has been done on making Grid computing adapted to remote sensing applications. In this paper, we intended to demonstrate the usage of Grid computing for quantitative remote sensing retrieval applications. A workload estimation and task partition algorithm was developed, and it executes a generic remote sensing algorithm in parallel over partitioned datasets, which is embedded in a middleware framework for remote sensing retrieval named the Remote Sensing Information Service Grid Node (RSIN). A case study shows that significant improvement of system performance can be achieved with this implementation. It also gives a perspective on the potential of applying Grid computing practices to remote sensing problems.

Aerosol optical depth retrieval over land using MODIS data and its application in monitoring air quality

Atmospheric remote sensing offers us a view to estimate air quality in describing the aerosol distribution either for a local or global coverage because aerosol parameters, such as aerosol optical depth (AOD) are significant indicators of the air quality. However, AOD retrieval over land still remains a difficult task because the measured signal is a composite of reflectance of sunlight by the variable surface covers and back scattering by the semitransparent aerosol layer. In this paper, an approach using bi-angle with Moderate Resolution Imaging Spectroradiometer (MODIS) data was presented. The derived AOD is compared to AERONET observations in the Asia area and a retrieval error within 16% is found. Moreover, a biomass burning episode in North China between June 7, 2007 was presented, it is demonstrated that AOD increased up to 2.0 during the burning phase and then returned to normal values (0.2–0.5), which fully in line with the observation result.

Synthetic retrieval of aerosol optical depth and surface reflectance using Terra and Aqua platforms in semi-arid regions

Aerosol quantitative retrieval from remote sensing over land surface is still a challenging task, especially for bright land areas such as desert, urban, coast, arid and semi-arid regions. A new aerosol optical depth (AOD) and surface reflectance remote sensing retrieval model is developed by exploiting a kernel-driven BRDF (Bidirectional Reflectance Distribution Function) model and the SYNTAM (Synergy of TERRA and AQUA MODIS) model, which considered the surface BRDF effect while retrieving AOD. After applying this new model to Terra and Aqua MODIS data in the Heihe River Basin of China, AOD and surface reflectance of this region are retrieved. Results show that the multiple correlation coefficient (R2) between retrieved AOD from MODIS and in situ measurements of CIMEL CE318 Sun-photometers is 0.92 at 0.55/zm. Using ASD Field Spec spectral radiometer measurements to validate retrieved surface reflectance, the RMSE values for band 1~3 are lower than 0.06.

Grid Enabled Simultaneous Retrieval of Aerosol and Ground Surface Reflectance from Integration of AERONET and Satellite Data

In this paper, a method using multi-resource remotely sensed data and ground base data for quantitative determination of aerosol optical properties was demonstrated. The model exploits the synergy of TERRA and AQUA MODerate resolution Imaging Spectrometer (MODIS) data, to simultaneously retrieve both Aerosol Optical Thickness (AOT) and surface reflectance. The ground station data routinely coming from AERosol Robotic NETwork (AERONET) of ground-based sun- and sky-scanning radiometers were assimilated as variables describing model initial states. To meet the increased computational needs caused by the complex and computational intensity, the algorithm was migrated to run as parallel processing on a Grid platform. Experimental results were presented with a realistic application, using data collected by MODIS over China mainland. The results show that Grid-enabled model allowed on-demand large volume of ground-based data assimilation with parameters, and achieved improvement both in retrieval accuracy and computing performance.

Monitoring the heavy fog using AOD derived from MODIS data

On Oct. 28th, 2009, a heavy fog hit East China. From the MODIS RGB composite image, it can be seen clearly that, much of land of the area (110°–123°E, 30°–42°N) was covered by the heavy fog or cloud. Using MODIS multi-satellite algorithm for aerosol optical depth (AOD) retrieval, we get the AOD Map at 470, 550 and 660 nm from MODIS data at 1km × 1km resolution. By Validating with the AERONET data, the results have good precision. The average relative error is about 10% and the correlation coefficient is as high as 0.88. Comparing with 10km MODIS aerosol products, our results can show more details because of the high resolution. From the AOD maps we can see the fog scope and the distribution as well as the relative thickness. Therefore, it is an effective method to monitor the fog.