Dongmei Chen

Examining the effect of spatial resolution and texture window size on classification accuracy: an urban environment case

The purpose of this paper is to evaluate spatial resolution effects on image classification. Classification maps were generated with a maximum likelihood (ML) classifier applied to three multi-spectral bands and variance texture images. A total of eight urban land use/cover classes were obtained at six spatial resolution levels based on a series of aggregated Colour Infrared Digital Orthophoto Quarter Quadrangle (DOQQ) subsets in urban and rural fringe areas of the San Diego metropolitan area. The classification results were compared using overall and individual classification accuracies. Classification accuracies were shown to be influenced by image spatial resolution, window size used in texture extraction and differences in spatial structure within and between categories. The more heterogeneous are the land use/cover units and the more fragmented are the landscapes, the finer the resolution required. Texture was more effective for improving the classification accuracy of land use classes at finer resolution levels. For spectrally homogeneous classes, a small window is preferable. But for spectrally heterogeneous classes, a large window size is required.

Optimization in multi-scale segmentation of high-resolution satellite images for artificial feature recognition

Multi‐resolution segmentation, as one of the most popular approaches in object‐oriented image segmentation, has been greatly enabled by the advent of the commercial software, eCognition. However, the application of multi‐resolution segmentation still poses problems, especially in its operational aspects. This paper addresses the issue of optimization of the algorithm‐associated parameters in multi‐resolution segmentation. A framework starting with the definition of meaningful objects is proposed to find optimal segmentations for a given feature type. The proposed framework was tested to segment three exemplary artificial feature types (sports fields, roads, and residential buildings) in IKONOS multi‐spectral images, based on a sampling scheme of all the parameters required by the algorithm. Results show that the feature‐type‐oriented segmentation evaluation provides an insight to the decision‐making process in choosing appropriate parameters towards a high‐quality segmentation. By adopting these feature‐type‐based optimal parameters, multi‐resolution segmentation is able to produce objects of desired form to represent artificial features.

Sensitivity of multitemporal NOAA AVHRR data of an urbanizing region to land-use/land-cover changes and misregistration

Our objectives were to: (1) investigate the sensitivity of multitemporal image data from the Advanced Very High Resolution Radiometer (AVHRR) satellite data for detecting land-use/land-cover changes primarily associated with urbanization and (2) test the effectiveness of a misregistration compensation model on the same data set. Empirical analyses were conducted using two near-anniversary, single-date NOAA AVHHR images of a rapidly urbanizing region of southern and Baja California. Analyses were facilitated by reference data from detailed GIS data layers of land-use/land-cover types for the 2 years corresponding to image acquisition dates (1990 and 1995). Almost all AVHRR pixels containing land-use/land-cover changes were mixed with nonchange areas, even when the extent of change features was greater than the nominal 1 km 2 ground sampling area. The strongest signals of image brightness change were detected by temporal differences of NDVI and Channel 4 surface temperature. ‘‘Undeveloped to urban’’ and ‘‘undeveloped to water’’ were the land-use/land-cover transition sequences with the most definitive AVHRR change signals. Mean magnitudes of misregistration errors were estimated to be around 0.2 pixel units in x and y directions. Mean values for misregistration noise equivalent in brightness change (MNEDB) were 0.02, 0.02, and 1.96 K for image differences of Channel 1 reflectance, NDVI, and Channel 4 surface temperature, respectively. The misregistration compensation model reduced false detection of change, but improvements in detection of land-use/land-cover changes were not conclusive. D 2002 Elsevier

Strategies for Integrating Information from Multiple Spatial Resolutions into Land-Use/Land-Cover Classification Routines

With the development of new remote sensing systems, veryhigh spatial and spectral resolution images now provide a source for detailed and continuous sampling of the Earth’s surface from local to regional scales. This paper presents three strategies for selecting and integrating information from different spatial resolutions into classification routines. One strategy is to combine layers of images of varying resolution. A second strategy involves comparing the a posteriori probabilities of each class at different resolutions. Another strategy is based on a top-down approach starting with the coarsest resolution. The multiresolution strategies are tested using simulated multiresolution images for a portion of the rural-urban fringe of the San Diego Metropolitan Area. The classification accuracy obtained from using three multiple strategies was greater when compared with that from a conventional single-resolution approach. Among the three strategies, the top-down approach resulted in the highest classification accuracy with a Kappa value of 0.648, compared to a Kappa of 0.566 for the conventional classifier.

Long-Term Changes of Lake Level and Water Budget in the Nam Co Lake Basin, Central Tibetan Plateau

Long-term changes in the water budget of lakes in the Tibetan Plateau due to climate change are of great interest not only for the importance of water management, but also for the critical challenge due to the lack of observations. In this paper, the water budget of Nam Co Lake during 1980‐2010 is simulated using a dynamical monthly water balance model. The simulated lake level is in good agreement with field investigations and the remotely sensed lake level. The long-term hydrological simulation shows that from 1980 to 2010, lake level rose from 4718.34 to 4724.93m, accompanied by an increase of lake water storage volume from 77.33 3 10 9 to 83.66 3 10 9 m 3 . For the net lake level rise (5.93m) during the period 1980‐2010, the proportional contributions of rainfall‐runoff, glacier melt, precipitation on the lake, lake percolation, and evaporation are 104.7%, 56.6%, 41.7%,222.2%,and280.9%, respectively. Apositivebut diminishingannualwater surplusisfound inNamCo Lake, implying a continuous but slowing rise in lake level as a hydrological consequence of climate change.

Characteristics and drivers of global NDVI‐based FPAR from 1982 to 2006

Fraction of Absorbed Photosynthetically Active Radiation (FPAR) is a state parameter in most ecosystem productivity models and is also the key terrestrial product. In this study, Normalized Difference Vegetation Index (NDVI) from Advanced Very High Resolution Radiometer (AVHRR) Global Inventory Modeling and Mapping Studies (GIMMS) was used to estimate FPAR from 1982 to 2006, using an intermediate model. Our research focused on the analysis of long-term global FPAR interannual trend patterns and driving forces involving climate and land cover changes. Results showed that interannual trend and spatial distribution patterns of global FPAR were independent of the changes in AVHRR instruments, and differed by season dynamics and vegetation types. Compared with other seasons, the period during JJA (June-August) exhibited more areas with decreasing FPAR and greater reduction range. For FPAR interannual trend, a wholly different correlation pattern was observed between temperature and precipitation, especially for arid and semi-arid regions. A significant influence of extreme droughts such as those associated with El Nino/Southern Oscillation (ENSO) on FPAR variability was found. The result also revealed the increasing and decreasing interannual trend of FPAR corresponding to the afforestation in the Three North Shelterbelts Program in China and deforestation in tropical forests in Southeast Asia. Driving factor analysis indicated that the climate and land cover changes had an interactive effect on the FPAR annual anomalous variation.

A maximum noise fraction transform with improved noise estimation for hyperspectral images

Feature extraction is often performed to reduce spectral dimension of hyperspectral images before image classification. The maximum noise fraction (MNF) transform is one of the most commonly used spectral feature extraction methods. The spectral features in several bands of hyperspectral images are submerged by the noise. The MNF transform is advantageous over the principle component (PC) transform because it takes the noise information in the spatial domain into consideration. However, the experiments described in this paper demonstrate that classification accuracy is greatly influenced by the MNF transform when the ground objects are mixed together. The underlying mechanism of it is revealed and analyzed by mathematical theory. In order to improve the performance of classification after feature extraction when ground objects are mixed in hyperspectral images, a new MNF transform, with an improved method of estimating hyperspectral image noise covariance matrix (NCM), is presented. This improved MNF transform is applied to both the simulated data and real data. The results show that compared with the classical MNF transform, this new method enhanced the ability of feature extraction and increased classification accuracy.

Spatio-Temporal Data Comparisons for Global Highly Pathogenic Avian Influenza (HPAI) H5N1 Outbreaks

Highly pathogenic avian influenza subtype H5N1 is a zoonotic disease and control of the disease is one of the highest priority in global health. Disease surveillance systems are valuable data sources for various researches and management projects, but the data quality has not been paid much attention in previous studies. Based on data from two commonly used databases (Office International des Epizooties (OIE) and Food and Agriculture Organization of the United Nations (FAO)) of global HPAI H5N1 outbreaks during the period of 2003–2009, we examined and compared their patterns of temporal, spatial and spatio-temporal distributions for the first time. OIE and FAO data showed similar trends in temporal and spatial distributions if they were considered separately. However, more advanced approaches detected a significant difference in joint spatio-temporal distribution. Because of incompleteness for both OIE and FAO data, an integrated dataset would provide a more complete picture of global HPAI H5N1 outbreaks. We also displayed a mismatching profile of global HPAI H5N1 outbreaks and found that the degree of mismatching was related to the epidemic severity. The ideas and approaches used here to assess spatio-temporal data on the same disease from different sources are useful for other similar studies.

Commuting behaviors and exposure to air pollution in Montreal, Canada

BACKGROUND Vehicular traffic is a major source of outdoor air pollution in urban areas, and studies have shown that air pollution is worse during hours of commuting to and from work and school. However, it is unclear to what extent different commuting behaviors are a source of air pollution compared to non-commuters, and if air pollution exposure actually differs by the mode of commuting. This study aimed to examine the relationships between commuting behaviors and air pollution exposure levels measured by urinary 1-OHP (1-hydroxypyrene), a biomarker of exposure to polycyclic aromatic hydrocarbons (PAHs). METHODS A cross-sectional study of 174 volunteers living in Montreal, 92 females and 82 males, aged 20 to 53 years was conducted in 2011. Each participant completed a questionnaire regarding demographic factors, commuting behaviors, home and workplace addresses, and potential sources of PAH exposure, and provided a complete first morning void urine sample for 1-OHP analysis. Multivariable general linear regression models were used to examine the relationships between different types of commuting and urinary 1-OHP levels. RESULTS Compared to non-commuters, commuters traveling by foot or bicycle and by car or truck had a significantly higher urinary 1-OHP concentration in urine (p=0.01 for foot or bicycle vs. non-commuters; p=0.02 for car or truck vs. non-commuters); those traveling with public transportation and combinations of two or more types of modes tended to have an increased 1-OHP level in urine (p=0.06 for public transportation vs. non-commuters; p=0.05 for commuters with combinations of two or more types of modes vs. non-commuters). No significant difference in urinary 1-OHP variation was found by mode of commuting. CONCLUSION This preliminary study suggests that despite the mode of commuting, all types of commuting during rush hours increase exposure to air pollution as measured by a sensitive PAH metabolite biomarker, and mode of commuting did not explain exposure variation.

Nonparametric Evaluation of Dynamic Disease Risk: A Spatio-Temporal Kernel Approach

Quantifying the distributions of disease risk in space and time jointly is a key element for understanding spatio-temporal phenomena while also having the potential to enhance our understanding of epidemiologic trajectories. However, most studies to date have neglected time dimension and focus instead on the “average” spatial pattern of disease risk, thereby masking time trajectories of disease risk. In this study we propose a new idea titled “spatio-temporal kernel density estimation (stKDE)” that employs hybrid kernel (i.e., weight) functions to evaluate the spatio-temporal disease risks. This approach not only can make full use of sample data but also “borrows” information in a particular manner from neighboring points both in space and time via appropriate choice of kernel functions. Monte Carlo simulations show that the proposed method performs substantially better than the traditional (i.e., frequency-based) kernel density estimation (trKDE) which has been used in applied settings while two illustrative examples demonstrate that the proposed approach can yield superior results compared to the popular trKDE approach. In addition, there exist various possibilities for improving and extending this method.