Xiaolin Zhu

A Modified Neighborhood Similar Pixel Interpolator Approach for Removing Thick Clouds in Landsat Images

Thick-cloud contamination is a common problem in Landsat images, which limits their utilities in various land surface studies. This letter presents a new method for removing thick clouds based on a modified neighborhood similar pixel interpolator (NSPI) approach that was originally developed for filling gaps due to the Landsat ETM+ Scan Line Corrector (SLC)-off problem. The performance of the proposed method was evaluated with both simulated and real cloudy images and compared with that of a contextual multiple linear prediction (CMLP) method. The results show that the modified NSPI approach can greatly reduce the edge effects by CMLP. The reflectance restored by the modified NSPI approach is more accurate than that by CMLP, especially when the cloud-free auxiliary and cloudy images are acquired from different seasons and have different spectral characteristics.

Blending MODIS and Landsat images for urban flood mapping

Satellite images provide important data sources for monitoring flood disasters. However, the trade-off between spatial and temporal resolutions of current satellite sensors limits their uses in urban flooding studies. This study applied and compared two data fusion models, the Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM) and Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM), in generating synthetic flooding images with improved temporal and spatial resolution for flood mapping. The synthetic images are produced in two scenarios: (1) for real-time prediction based on Landsat and MODIS images acquired before the investigated flooding; and (2) for post-disaster prediction based on images acquired after the flooding. The 2005 Hurricane Katrina in New Orleans was selected as a case study. The result shows that the Landsat-like images generated can be successfully applied in flood mapping. Particularly, ESTARFM surpasses STARFM in predicting surface reflectance in both real-time and post-flooding predictions. However, the flood mapping results from the Landsat-like images produced by both STARFM and ESTARFM are similar with overall accuracy around 0.9. Only for the flooding maps of real-time predictions does ESTARFM get a slightly higher overall accuracy than STARFM, indicating that the lower quality of the Landsat-like image generated by STARFM may not affect flood mapping accuracy, due to the marked contrast between land and water. This study suggests great potential of both STARFM and ESTARFM in urban flooding research. Blending multi-sources images could also support other disaster studies that require remotely sensed data with both high spatial and temporal resolution.

Estimating Tree-Root Biomass in Different Depths Using Ground-Penetrating Radar: Evidence from a Controlled Experiment

Roots have important functions in the ecosystem. Therefore, establishing root-related parameters such as root size, biomass, and 3-D architecture is necessary. Traditional methods for measuring tree roots are labor intensive and destructive to nature, limiting quantitative and repeated assessments in long-term research. Ground-penetrating radar (GPR) provides a nondestructive method for measuring tree roots. This study investigates the feasibility of a GPR system with 500-MHz, 900-MHz, and 2-GHz measurement frequencies for detecting tree roots and estimating root biomass under controlled experimental conditions in a sandy area. After energy attenuation correction and velocity analysis, not only the individual root in subsurface is able to be located but also the parameters that correlate well with root biomass can be extracted from the processed GPR data. The major findings were as follows. First, both the amplitude and amplitude-area indices were confirmed to be more effective for estimating root biomass after attenuation-effect compensation. This result suggests that the calibration of GPR wave-attenuation effects and velocity changes with depth are helpful in estimating root biomass from GPR parameters. Second, the selection of GPR system frequency was mainly dependent on field conditions, particularly soil water content. Lower frequency was recommended for developing root biomass estimation model under varied soil conditions. Third, the new method based on the metal reflector experiment was effective and easy to perform in situ for attenuation-effect correction.

Disaggregation of Remotely Sensed Land Surface Temperature: A Generalized Paradigm

The environmental monitoring of earth surfaces requires land surface temperatures (LSTs) with high temporal and spatial resolutions. The disaggregation of LST (DLST) is an effective technique to obtain high-quality LSTs by incorporating two subbranches, including thermal sharpening (TSP) and temperature unmixing (TUM). Although great progress has been made on DLST, the further practice requires an in-depth theoretical paradigm designed to generalize DLST and then to guide future research before proceeding further. We thus proposed a generalized paradigm for DLST through a conceptual framework (C-Frame) and a theoretical framework (T-Frame). This was accomplished through a Euclidean paradigm starting from three basic laws summarized from previous DLST methods: the Bayesian theorem, Toblers first law of geography, and surface energy balance. The C-Frame included a physical explanation of DLST, and the T-Frame was created by construing a series of assumptions from the three basic laws. Two concrete examples were provided to show the advantage of this generalization. We further derived the linear instance of this paradigm based on which two classical DLST methods were analyzed. This study finally discussed the implications of this paradigm to closely related topics in remote sensing. This paradigm develops processes to improve an understanding of DLST, and it could be used for guiding the design of future DLST methods.

An Enhanced Physical Method for Downscaling Thermal Infrared Radiance

Thermal infrared (TIR) imagery plays a critical role in characterizing land surface processes and modeling energy balances. However, due to the low TIR radiance emitted from the Earths surface, TIR imagery acquired from satellite thermal sensors is often with limited spatial resolutions, which presents a serious obstacle to its applications in heterogeneous landscapes (e.g., the studies of urban heat island). In this letter, we developed a new method for downscaling TIR radiance by addressing the limitations of a previously developed physical downscaling method by Liu and Pu (2008). To validate our method, a 990-m TIR image was generated by upscaling a 90-m TIR image from the Advanced Spaceborne Thermal Emission and Reflection Radiometer and downscaled back to the 90-m resolution using the proposed method. The results show that the enhanced physical method not only greatly reduced the block effects and smooth effects found in the original physical method but also improved the downscaling accuracy over the original method.

MAP-MRF Approach to Landsat ETM+ SLC-Off Image Classification

Land cover classification is an important application of Landsat images. Unfortunately, the scan-line corrector (SLC) failure in 2003 causes about 22% pixels to remain unscanned in each Landsat 7 ETM+ image. This problem seriously limits the application of Landsat 7 ETM+ images for land cover classification. A common strategy for addressing this problem is filling the unscanned gaps before classification work. However, the simple and high-speed methods for gap-filling cannot yield satisfactory results, especially for heterogeneous landscapes, while the gap-filling methods with high accuracy are often complicated and inefficient in the use of time. This paper develops a new method based on the maximum a posteriori decision rule and Markov random field theory (the MAP-MRF classification framework) for classifying SLC-off ETM+ images without filling unscanned gaps beforehand. The proposed method efficiently avoids the complicated process for gap-filling. The performance of the proposed method was validated by simulated SLC-off images. The results show that the classification accuracy of the proposed method is even higher than that of classification from an image filled by the precise gap-filling algorithm neighborhood similar pixel interpolator, which indicates that an accurate land cover map can be generated without spending time and effort to fill gaps in SLC-off images prior to the land cover classification.

Predictions of Tropical Forest Biomass and Biomass Growth Based on Stand Height or Canopy Area Are Improved by Landsat-Scale Phenology across Puerto Rico and the U.S. Virgin Islands

Remotely-sensed estimates of forest biomass are usually based on various measurements of canopy height, area, volume or texture, as derived from LiDAR, radar or fine spatial resolution imagery. These measurements are then calibrated to estimates of stand biomass that are primarily based on tree stem diameters. Although humid tropical forest seasonality can have low amplitudes compared with temperate regions, seasonal variations in growth-related factors like temperature, humidity, rainfall, wind speed and day length affect both tropical forest deciduousness and tree height-diameter relationships. Consequently, seasonal patterns in spectral measures of canopy greenness derived from satellite imagery should be related to tree height-diameter relationships and hence to estimates of forest biomass or biomass growth that are based on stand height or canopy area. In this study, we tested whether satellite image-based measures of tropical forest phenology, as estimated by indices of seasonal patterns in canopy greenness constructed from Landsat satellite images, can explain the variability in forest deciduousness, forest biomass and net biomass growth after already accounting for stand height or canopy area. Models of forest biomass that added phenology variables to structural variables similar to those that can be estimated by LiDAR or very high-spatial resolution imagery, like canopy height and crown area, explained up to 12% more variation in biomass. Adding phenology to structural variables explained up to 25% more variation in Net Biomass Growth (NBG). In all of the models, phenology contributed more as interaction terms than as single-effect terms. In addition, models run on only fully-forested plots performed better than models that included partially-forested plots. For forest NBG, the models produced better results when only those plots with a positive growth, from Inventory Cycle 1 to Inventory Cycle 2, were analyzed, as compared to models that included all plots