Joseph M. Piwowar

Time-Series Analysis of Medium-Resolution, Multisensor Satellite Data for Identifying Landscape Change

The overall goal of this study is to use medium-resolution satellite imagery to determine recent changes in the landscape of the coastal zone near Sanya in the Province of Hainan, China. A search for suitable satellite imagery revealed that the only way to identify the changes was to use data from three different sensors acquired over a 12-year time period: a 1987 Landsat 5 Thematic Mapper (TM) image, a 1999 Landsat 7 Enhanced Thematic Mapper Plus (ETM� ) image, and two SPOT 2 High Resolution Visible (HRV) images acquired in 1991 and 1997. Given that the Landsat and SPOT images have different spatial resolutions and that the spectral bands cover somewhat different spectral ranges, the challenge was how to combine the images in digital format to be able to detect subtle changes in the landscape. Measures of brightness, greenness, and the normalized difference vegetation index (NDVI) were explored using standardized principal components analysis (PCA). Approximately 38 percent of the scene was occupied by water, so tests were performed with the water included and also with the water masked out to remove these low-variance pixels. Factor loadings and input-band contributions were used to interpret component images. Results show that PCA of the visible bands, representing brightness, is the superior approach for identifying new urban features in the landscape. For identification of changes to vegetation, the near-infrared (NIR) bands outperformed NDVI. Selected standardized PCA images with visible and NIR bands are recommended for identifying general changes to an urban landscape using a time-series of imagery acquired by different satellite sensors. Benefits of using a mask are believed to be dependent upon study-site characteristics.

TEMPORAL MIXTURE ANALYSIS OF ARCTIC SEA ICE IMAGERY: A NEW APPROACH FOR MONITORING ENVIRONMENTAL CHANGE

Abstract In this paper, we introduce the idea of temporal mixture analysis (TMA) for analyzing long sequences of hypertemporal remote sensing imagery. The basis of this approach is spectral mixture analysis, which we adapt from the spectral domain to the time domain. The TMA procedure is demonstrated by applying it to a 9-year record of scanning multichannel microwave radiometer sea ice concentrations in the Northern Hemisphere. We find that end-member fraction images provide a unique summary of spatial arrangements and temporal characteristics of the mapped phenomenon during a specific period and can be used to characterize climatic normals. A key distinction that differentiates temporal mixture imagery from similar images derived through more traditional means is that the data presented are derived from the temporal characteristics of the analyzed phenomenon and not the type of feature present.

Integration of spatial data in vector and raster formats in a geographic information system environment

This paper examines the common methods for converting spatial data sets between vector and raster formats and presents the results of extensive benchmark testing of these procedures. The tests performed are unique in this field since: (1) they used both synthetic and real test data sets; (2) they measured conversion quality, accuracy and efficiency, not just how fast the procedure operated; and (3) they were conducted in a generic geographic information system (GIS) environment without the aid of specialized computer hardware. The results show that the best overall techniques are the ones which take advantage of spatial relationships inherent in the data sets. These were the Scan Line algorithm for vector to raster conversions and the Boundary Linking algorithm for raster to vector conversions.

Integration of Remote Sensing and GIS to Detect Pockets of Urban Poverty: The Case of Rosario, Argentina

The advent of high spatial resolution, multispectral satellite imagery has allowed analysis of remotely sensed images of urban land cover to become more useful to urban planning and decision making than in the past. The addition of radar imagery at relatively high spatial resolution (6 metres at best), with the advantages that it is not affected by cloud and diurnal light conditions and that it is sensitive to the targets geometric shape, surface roughness and moisture content offers additional capability in this regard. This paper incorporates analysis of Canadian RADARSAT-1 and American Landsat TM satellite imagery and ground-based GIS data to identify known pockets of urban poverty. Poverty is defined, based on a limited number of census variables related to dwelling construction materials and per household overcrowding. The objective is to provide a proof of concept that remote sensing data, especially from synthetic aperture radar, and ground-based GIS data can be successfully integrated for urban planning purposes. The results suggest that the approach used is reasonable and that, with future refinement, it offers planners and decision makers a timely and cost effective means to locate and monitor poverty pockets in urban areas. This is especially important in large, rapidly urbanising areas in the developing world.

HYPERTEMPORAL ANALYSIS OF REMOTELY SENSED SEA ICE DATA FOR CLIMATE CHANGE STUDIES

Climatologists have speculated that a spatially coherent pattern of high-latitude temperature trends could be an early indicator of climatic change. The sensitivity of sea ice to the temperature of the overlying air suggests the possibility that trends in Arctic ice conditions may be useful proxy indicators of general climatic changes. Aspects of the north-polar ice pack which have been identified as key parameters to be monitored include ice extent, concentration, type, thickness and motion dynamics. In spite of the considerable interannual, regional and seasonal variations exhibited by these data, there may be some evidence of an emerging trend towards decreasing ice extent and concentration. Collecting data in such a remote and harsh environment to support these analyses is only possible through satellite remote sensing. Remote sensing in the microwave portion of the electromagnetic spectrum is particularly relevant for polar applications because microwaves are capable of penetrating the atmosphere under virtually all conditions and are not dependent on the sun as a source of illumination. In particular, analyses of passive microwave imagery can provide us with daily information on sea-ice extent, type, concentration, dynamics and melt onset. A historical record of Arctic imagery from orbiting passive microwave sensors starting from 1973 provides us with an excellent data source for climate change studies. The development of analysis tools to support large area monitoring is integral to advancing global change research. The critical need is to create techniques which highlight the space-time relationships in the data rather than simply displaying voluminous quantities of data. In particular, hypertemporal image analysis techniques are required to help find anticipated trends and to discover unexpected or anomalous temporal relationships. Direct hypertemporal classification, principal components analysis and spatial time-series analysis are identified as three primary techniques for enhancing change in temporal image sequences. There is still a need for the development of new tools for spatial- temporal modelling.

The derivation of an Arctic sea ice normal through temporal mixture analysis of satellite imagery

Abstract The use of temporal mixture analysis (TMA) for creating a long-term baseline, or environmental “normal” is described. TMA is a promising analysis method derived from the hyperspectral image processing technique of spectral mixture analysis (SMA). TMA is algebraically identical to SMA, except that it is applied to temporal spectra rather than to electromagnetic spectra. TMA has particular potential to extract climate signals from long image sequences. To demonstrate the utility of TMA, this paper documents its use to isolate nine fundamental temporal signals (“endmembers”) from a 20-year Northern Hemisphere sea ice concentration image time series. The temporal endmembers establish a baseline of temporal variability that can be treated as an environmental normal. The “fraction images” produced by the analysis highlight the regions where the temporal signals are strongest and provide new insights into the dynamics of the Arctic sea ice cover.

An Environmental Normal of Vegetation Vigour for the Northern Great Plains

Normalized difference vegetation index (NDVI) data from the Advanced Very High Resolution Radiometer (AVHRR) sensor on board the National Oceanic and Atmospheric Administration (NOAA) satellites were used to create a spatially detailed baseline of vegetation conditions in the northern Great Plains of North America. An environmental normal of vegetation vigour was created from NDVI means and standard deviations calculated over 22 years for each 10-day period during the growing season. Significant vegetation vigour anomalies - differences from the normal - were subsequently identified and associated with concurrent temperature and precipitation data. Growing season vegetation vigour anomalies were found to be most dependent on weather patterns from the previous spring, and in some cases, from the preceding summer. Regions with the densest and most diverse vegetation covers were impacted the most by temperature and precipitation. Statistically significant increases in vegetation vigour over the 22-year period were measured across the entire study area, with the exception of the vegetation communities with the sparsest ground covers. This increase was matched by a similarly significant rise in annual NDVI variability for all of the phenologies. The changes in vegetative cover leading to the increase in NDVI values may be related to warmer winter temperatures.

Multitemporal change analysis of multispectral imagery using principal components analysis

Early change analysis studies established the fundamental basis for applying the principal components analysis (PCA) transformation to remote sensing images acquired on two dates. There are an increasing number of studies, however, which extend this basis to longer image time series with little concern for its appropriateness. In particular, when multispectral and multitemporal data are used in the same analysis, the components may be difficult to interpret since they would contain not only temporal variation, but spectral changes as well. In this paper we seek to establish an appropriate ordination technique to condense the multispectral information from each date prior to multitemporal PCA. We find that the Normalized Difference Vegetation Index (NDVI) provides superior results because it produces annual composites with a strong physical basis.

Image Analysis On A Macintosh II.

Current research in remote sensing is directed towards understanding the relationship between image data and the physical features they represent. The Macintosh II provides a framework by which these relationships can be explored, implemented and tested quickly. Enhancements, filtering, image transformations and other image display functions allow for visual exploration of the image data. Numerical exploration is facilitated through statistical measures and graphs. Implementation of new and existing algorithms using numerical analysis programs allows the researcher to minimize the time between the idea and result. Rigorous quantitative analysis can be supplemented with visual presentation either as images or graphs. This combined visual and numerical approach and ease of movement between visual and numerical presentation, promotes the willingness to explore. In this presentation we illustrate image processing and analysis capabilities of the Macintosh II with a specific illustration of development and testing of texture algorithms for SAR sea ice classification We conclude with suggestions for stand-alone and networking

Integrating remote sensing, GIS and dynamic models: Cellular automata approach for the simulation of urban growth for the city of Montreal

This paper presents a cellular automata approach for modeling the dynamics of urban growth. Cities are among the most complex spatial systems and modeling their dynamics of growth using traditional modeling techniques is a challenging task. Cellular automata (CA) have been widely used for modeling urban growth because of their computational simplicity and explicit representation of time and space. CA models are able to generate complex patterns from the interaction of simple components of the system using simple rules. This makes CA suitable for modeling the complexity of urban dynamics of the cities. Integrating GIS tools and remote sensing data with CA has the potential to provide realistic simulation of the future urban growth of cities. The proposed approach is applied to model the growth of the City of Montreal, Quebec over the last three decades. Land use/land cover maps derived from Landsat data acquired in 1975 and 1990 were used to train a CA model which was then used to project the land use in 2005. A comparison of the projected and actual land uses for 2005 (map also derived from Landsat imagery) is presented and discussed.