Hankui K. Zhang

Mapping crop phenology using NDVI time-series derived from HJ-1 A/B data

With the availability of high frequent satellite data, crop phenology could be accurately mapped using time-series remote sensing data. Vegetation index time-series data derived from AVHRR, MODIS, and SPOT-VEGETATION images usually have coarse spatial resolution. Mapping crop phenology parameters using higher spatial resolution images (e.g., Landsat TM-like) is unprecedented. Recently launched HJ-1 A/B CCD sensors boarded on China Environment Satellite provided a feasible and ideal data source for the construction of high spatio-temporal resolution vegetation index time-series. This paper presented a comprehensive method to construct NDVI time-series dataset derived from HJ-1 A/B CCD and demonstrated its application in cropland areas. The procedures of time-series data construction included image preprocessing, signal filtering, and interpolation for daily NDVI images then the NDVI time-series could present a smooth and complete phenological cycle. To demonstrate its application, TIMESAT program was employed to extract phenology parameters of crop lands located in Guanzhong Plain, China. The small-scale test showed that the crop season start/end derived from HJ-1 A/B NDVI time-series was comparable with local agro-metrological observation. The methodology for reconstructing time-series remote sensing data had been proved feasible, though forgoing researches will improve this a lot in mapping crop phenology. Last but not least, further studies should be focused on field-data collection, smoothing method and phenology definitions using time-series remote sensing data.

An Automated Approach for Sub-Pixel Registration of Landsat-8 Operational Land Imager (OLI) and Sentinel-2 Multi Spectral Instrument (MSI) Imagery

Moderate spatial resolution satellite data from the Landsat-8 OLI and Sentinel-2A MSI sensors together offer 10 m to 30 m multi-spectral reflective wavelength global coverage, providing the opportunity for improved combined sensor mapping and monitoring of the Earth’s surface. However, the standard geolocated Landsat-8 OLI L1T and Sentinel-2A MSI L1C data products are currently found to be misaligned. An approach for automated registration of Landsat-8 OLI L1T and Sentinel-2A MSI L1C data is presented and demonstrated using contemporaneous sensor data. The approach is computationally efficient because it implements feature point detection across four image pyramid levels to identify a sparse set of tie-points. Area-based least squares matching around the feature points with mismatch detection across the image pyramid levels is undertaken to provide reliable tie-points. The approach was assessed by examination of extracted tie-point spatial distributions and tie-point mapping transformations (translation, affine and second order polynomial), dense-matching prediction-error assessment, and by visual registration assessment. Two test sites over Cape Town and Limpopo province in South Africa that contained cloud and shadows were selected. A Landsat-8 L1T image and two Sentinel-2A L1C images sensed 16 and 26 days later were registered (Cape Town) to examine the robustness of the algorithm to surface, atmosphere and cloud changes, in addition to the registration of a Landsat-8 L1T and Sentinel-2A L1C image pair sensed 4 days apart (Limpopo province). The automatically extracted tie-points revealed sensor misregistration greater than one 30 m Landsat-8 pixel dimension for the two Cape Town image pairs, and greater than one 10 m Sentinel-2A pixel dimension for the Limpopo image pair. Transformation fitting assessments showed that the misregistration can be effectively characterized by an affine transformation. Hundreds of automatically located tie-points were extracted and had affine-transformation root-mean-square error fits of approximately 0.3 pixels at 10 m resolution and dense-matching prediction errors of similar magnitude. These results and visual assessment of the affine transformed data indicate that the methodology provides sub-pixel registration performance required for meaningful Landsat-8 OLI and Sentinel-2A MSI data comparison and combined data applications.

Unified fusion of remote-sensing imagery: generating simultaneously high-resolution synthetic spatial–temporal–spectral earth observations

Current satellite remote-sensing systems compromise between spatial resolution and spectral and/or temporal resolution, which potentially limits the use of remotely sensed data in various applications. Image fusion processes, including spatial and spectral fusion (SSF) and spatial and temporal fusion (STF), provide powerful tools for addressing these technological limitations. Although SSF and STF have been extensively studied separately, they have not yet been integrated into a unified framework to generate synthetic satellite images with high spatial, temporal and spectral resolution. By formulating these two types of fusion into one general problem, i.e. super resolving a low spatial resolution image with a high spatial resolution image acquired under different conditions (e.g. at different times and/or in different acquisition bands), this letter proposes a notion of unified fusion that can accomplish both SSF and STF in one process. A Bayesian framework is subsequently developed to implement SSF, STF and unified fusion to generate ‘virtual sensor’ data, characterized by high spatial, temporal and spectral resolution simultaneously. The proposed method was then applied to the fusion of Moderate Resolution Imaging Spectroradiometer (MODIS) and Landsat Enhanced Thematic Mapper Plus (ETM+) images of the Hong Kong area, with the average spatial correlation coefficient exceeding 0.9 for near infrared–red–green bands between the fused result and the input Landsat image and with good preservation of the MODIS spectral properties.

A New Look at Image Fusion Methods from a Bayesian Perspective

Component substitution (CS) and multi-resolution analysis (MRA) are the two basic categories in the extended general image fusion (EGIF) framework for fusing panchromatic (Pan) and multispectral (MS) images. Despite of the method diversity, there are some unaddressed questions and contradictory conclusions about fusion. For example, is the spatial enhancement of CS methods better than MRA methods? Is spatial enhancement and spectral preservation competitive? How to achieve spectral consistency defined by Wald et al. in 1997? In their definition any synthetic image should be as identical as possible to the original image once degraded to its original resolution. To answer these questions, this research first finds out that all the CS and MRA methods can be derived from the Bayesian fusion method by adjusting a weight parameter to balance contributions from the spatial injection and spectral preservation models. The spectral preservation model assumes a Gaussian distribution of the desired high-resolution MS images, with the up-sampled low-resolution MS images comprising the mean value. The spatial injection model assumes a linear correlation between Pan and MS images. Thus the spatial enhancement depends on the weight parameter but is irrelevant of which category (i.e., MRA or CS) the method belongs to. This paper then adds a spectral consistency model in the Bayesian fusion framework to guarantee Wald’s spectral consistency with regard to arbitrary sensor point spread function. Although the spectral preservation in the EGIF methods is competitive to spatial enhancement, the Wald’s spectral consistency property is complementary with spatial enhancement. We conducted experiments on satellite images acquired by the QuickBird and WorldView-2 satellites to confirm our analysis, and found that the performance of the traditional EGIF methods improved significantly after adding the spectral consistency model.

Spatio-temporal reflectance fusion via unmixing: accounting for both phenological and land-cover changes

Owing to technical limitations the acquisition of fine spatial resolution images (e.g. Landsat data) with frequent (e.g. daily) coverage remains a challenge. One approach is to generate frequent Landsat surface reflectances through blending with coarse spatial resolution images (e.g. Moderate Resolution Imaging Spectroradiometer, MODIS). Existing implementations for data blending, such as the Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM) and Enhanced STARFM (ESTARFM), have their shortcomings, particularly in predicting the surface reflectance characterized by land-cover-type changes. This article proposes a novel blending model, namely the Unmixing-based Spatio-Temporal Reflectance Fusion Model (U-STFM), to estimate the reflectance change trend without reference to the change type, i.e. phenological change (e.g. seasonal change in vegetation) or land-cover change (e.g. conversion of a vegetated area to a built-up area). It is based on homogeneous change regions (HCRs) that are delineated by segmenting the Landsat reflectance difference images. The proposed model was tested on both simulated and actual data sets featuring phenological and land-cover changes. It proved more capable of capturing both types of change compared to STARFM and ESTARFM. The improvement was particularly observed on those areas characterized by land-cover-type changes. This improved fusion algorithm will thereby open new avenues for the application of spatio-temporal reflectance fusion.

Improving Landsat ETM+ Urban Area Mapping via Spatial and Angular Fusion With MISR Multi-Angle Observations

Urban landscapes are a complex combination of buildings, roads, vegetation, soil, and water, each of which exhibits unique radiative and thermal properties. To understand the dynamics of patterns and processes and their interactions in heterogeneous landscapes such as urban areas, more precise urban mapping techniques are of essential importance. Several investigations have demonstrated that Bidirectional Reflectance Distribution Function (BRDF) information can be utilized to complement spectral information to improve land cover (especially vegetation) classification accuracies on the local, regional and global scales. However, the potential benefits of adding remotely sensed angular information to improve urban mapping have rarely been explored. This paper uses Multi-angle Imaging SpectroRadiometer (MISR) data to investigate the view angle effects on spectral response and discrimination of urban land cover types in Shenzhen, China. For this purpose, a spatial and angular fusion (SAF) model was developed for blending MISR and Enhanced Thematic Mapper Plus (ETM+) images. A classification of the fused data with twenty channels using support vector machines (SVM) and a post-classification probability relaxation were then performed after channel selection through principal-component analysis (PCA). The results showed that the contribution of MISR to improving ETM+ urban mapping accuracy was 2.86% in our experiments and its statistical significance was validated by McNemars test.

Spatio-spectral fusion of satellite images based on dictionary-pair learning

This paper proposes a novel spatial and spectral fusion method for satellite multispectral and hyperspectral (or high-spectral) images based on dictionary-pair learning. By combining the spectral information from sensors with low spatial resolution but high spectral resolution (LSHS) and the spatial information from sensors with high spatial resolution but low spectral resolution (HSLS), this method aims to generate fused data with both high spatial and spectral resolution. Based on the sparse non-negative matrix factorization technique, this method first extracts spectral bases of LSHS and HSLS images by making full use of the rich spectral information in LSHS data. The spectral bases of these two categories data then formulate a dictionary-pair due to their correspondence in representing each pixel spectra of LSHS data and HSLS data, respectively. Subsequently, the LSHS image is spatial unmixed by representing the HSLS image with respect to the corresponding learned dictionary to derive its representation coefficients. Combining the spectral bases of LSHS data and the representation coefficients of HSLS data, fused data are finally derived which are characterized by the spectral resolution of LSHS data and the spatial resolution of HSLS data. The experiments are carried out by comparing the proposed method with two representative methods on both simulation data and actual satellite images, including the fusion of Landsat/ETM+ and Aqua/MODIS data and the fusion of EO-1/Hyperion and SPOT5/HRG multispectral images. By visually comparing the fusion results and quantitatively evaluating them in term of several measurement indices, it can be concluded that the proposed method is effective in preserving both the spectral information and spatial details and performs better than the comparison approaches.

A generalization of spatial and temporal fusion methods for remotely sensed surface parameters

Remotely sensed surface parameters, such as vegetation index, leaf area index, surface temperature, and evapotranspiration, show diverse spatial scales and temporal dynamics. Generally the spatial and temporal resolutions of remote-sensing data should match the characteristics of surface parameters under observation. These requirements sometimes cannot be provided by a single sensor due to the trade-off between spatial and temporal resolutions. Many spatial and temporal fusion (STF) methods have been proposed to derive the required data. However, the methodology suffers from disorderly development. To better inform future research, this study generalizes the existing methods from around 100 studies as spatial or temporal categories based on their physical assumptions related to spatial scales and temporal dynamics. To be specific, the assumptions are related to the scale invariance of the temporal information and temporal constancy of the spatial information. The spatial information can be contexture or spatial details. Experiments are conducted using Landsat data acquired on 13 dates in two study areas and simulated Moderate Resolution Imaging Spectroradiometer (MODIS) data. The results are presented to demonstrate the typical methods from each category. This study concludes the following. (1) Contexture methods depend heavily on how components maps (contexture) are defined. They are not recommended except when components maps can be estimated properly from observed images. (2) The spatial and temporal adaptive reflectance fusion model (STARFM) and enhanced STARFM (ESTARFM) methods belong to the temporal and spatial categories, respectively. Thus, STARFM and ESTARFM should be better applied to temporal variance – dominated and spatial variance – -dominated areas, respectively. (3) Non-linear methods, such as the sparse representation-based spatio-temporal reflectance fusion model, can successfully address land-cover changes in addition to phonological changes, thereby providing a promising option for STF problems in the future.

Support Vector Regression-Based Downscaling for Intercalibration of Multiresolution Satellite Images

This paper introduces a nonlinear super-resolution method for converting low spatial resolution data into high spatial resolution data to calibrate multiple sensors with a moderate spatial resolution difference, e.g., the Landsat 7 Enhanced Thematic Mapper Plus (ETM+) (30 m) and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) visible and near infrared (NIR) sensors (15 m). A preliminary linear calibration was first applied to reduce the radiometric difference. The remaining nonlinear part of the radiometric and spatial resolution differences were then calibrated by downscaling the ETM+ data to ASTER data using a support vector regression (SVR)-based super-resolution method. Experiments were conducted on two subsets (representing rural and urban areas) of the ETM+ and ASTER scenes located in the central United States on top of atmospheric reflectance observed on August 13, 2001. It was found that the radiometric difference between the two sensors caused by their spectral band difference could be largely reduced by a linear transfer equation, and the reduction could be more than 60% for the green and NIR bands. The SVR-calibrated data showed improvement over the linearly calibrated data in terms of quantitative measures and visual analysis. Furthermore, SVR calibration improved the spatial resolution of the ETM+ data toward resembling the 15-m cell size of the ASTER pixel. Consequently, the proposed method has the potential to extend an ASTER scenes swath width to match that of an ETM+ scene.

Reconstructing Seasonal Variation of Landsat Vegetation Index Related to Leaf Area Index by Fusing with MODIS Data

In the development of an empirical relationship between the leaf area index (LAI) and the vegetation index (VI), the infrequency of the medium resolution VI often makes it difficult, sometimes impossible, to find VI observations acquired close to the LAI measurement date. To overcome this dilemma, this paper presents a method, named reduced simple ratio (RSR), to reconstruct seasonal time series of a VI at the Landsat resolution. Each RSR time series is represented by a double logistic (D-L) curve with seven unknown parameters. The methodology solves these parameters using a multi-objective optimization method by blending frequent MODIS observations with Landsat observations acquired at a few dates (usually fewer than seven) in a year. We tested the reconstructing approach in a boreal forest in Canada and a cropland area in Australia. The reconstructed Landsat RSR compared well with the observed RSR even when only two Landsat images were used for reconstruction, and better accuracy was achieved when more Landsat images were used. Ground LAI measurements were taken at a date not coincident with any of the Landsat dates in the Canada study area. Results of LAI retrieval showed that the measured LAI had a higher correlation with the reconstructed RSR at the measurement date than with the observed Landsat RSR at the three acquisition dates.