Remote Sensing and Geographic Information System Data Integration: Error Sources and Research Issues Ross S. Lunetta U. S. Environmental Protection Agency, Environmental Monitoring Systems Laboratory, P.O. Box 93478, Las Vegas, NV 89193-3478 Russell G. Congalton Department of Forestry and Resource Management, University of California, 204 Mulford Hall, Berkeley, CA 94720 Lynn K. Fenstermaker Desert Research Institute, C/O U.S.EPA, EMSL-LV (AMS), P.O. Box 93478, Las Vegas, NV 89193-3478 John R. lensen Department of Geography, University of South Carolina, Columbia, SC 29208 Kenneth C. McGwire Department of Geography, University of California, Santa Barbara, CA 93106 Larry R. Tinny EG&G Energy Measurements, Remote Sensing Laboratory, P.O. Box 1912, M/S P-02, Las Vegas, NV 89125 ABSTRACT:Data derived from remote sensors are increasingly being utilized as a data source in geographic information systems (GIs). Error associated with the remote sensing and GIs data acquisition, processing, analysis, conversion, and final product presentation can have a significant impact on the confidence of decisions made using the data. The goal of this paper is to provide a broad ove~ewof spatial data error sources, and to identify priority research topics which will reduce impediments and enhance the quality of integrated remote sensing and GIs data. Potential sources of error will be identified at each data integration process step, impacts of error propagation on the decision making and implementation processes will be assessed, and priority error quantification research topics will be recommended. Suggested priorities for error quantification research topics include the development of standardized and more cost- effective remote sensing accuracy assessment procedures, development of field verification data collection guidelines, procedures for vector-to-raster and raster-to-vector conversions, assessment of scaling issues for the incorporation of elevation data in georeferencing, and development of standardized geometric and thematic reliability legend diagrams. INTRODUCTION assessment, and final product presentation. Error may be trans- ITH THE PROLIFERATION OF GEOGRAPHIC INFO~TION ferred from one data process step to the next unknown to the WSYSTEMS (GIs) in both industry and government for nu- analysts until it manifests in the final product, error may ac- merous applications, there has been a tremendous increase in cumulate throughout the process in an additive or multiplica- demand for remote sensing as a data input source to spatial tive fashion, and individual process error(s) can be overshadowed database development. Products derived from remote sensing by other errors of greater magnitude. The potential sources of are particularly attractive for GIS database development because error which may enter a remote sensing data processing flow they can provide cost-effective, large area coverage in a digital are illustrated in Figure 1. Although the typical processing flow format that can be input directly into a GIs. Because remote is displayed in a clockwise direction, bidirectional and cross- sensing data are typically collected in a raster data format, the element processing flow patterns are possible. For example, data can be cost-effectively converted to a vector or quadtree data conversion usually occurs after data analysis. However, in format for subsequent analysis or modeling applications (Lee, some instances conversion may occur in the data processing 1991). step. Usually these conversions are in the form of raster-to- hhough the use of remote sensing data for spatial database raster (e.g., resampling pixel size) or vector-to-raster. development is increasing rapidly, our understanding of asso- In theory, the amount of error entering the system at each ciated data processing errors, especially for integrating multiple step can be estimated. In practice, however, error is typically spatial data sets, lags far behind. Performing spatial data analy- only assessed at the conclusion of data analysis (i-e., the final sis operations with data of unknown accuracy, or with incom- product), if it is assessed at all. Usually, the decision maker is patible error types, will produce a product with low confidence provided graphic final products, statistical data, or modeling limits and restricted use in the decision making process. Al- results with little or no information concerning the confidence though some research has addressed spatial error (Veregin, that can be placed in the information. This limits the confidence 1989a), we need to clearly identify the types of error that may in the implemented decision(s). It is imperative that we improve enter into the process, understand how the error propagates our ability to quantify the error associated with the data, and throughout the processing flow, and develop procedures to bet- monitor the error as it propagates through a GIs application. ter quantify and report the error using standardized techniques, The following sections review the nature of the error that may i.e., techniques for all spatial data users. be introduced and identify significant improvements that must The process of integrating remote sensing data into a GIs be addressed. usually includes the following analytical procedures: data ac- The objectives of this paper are first, to identify the potential quisition, data processing, data analysis, data conversion, error sources of error in the data processing flow for the integration PHOTOGRAMMETRIC ENGINEERING& REMOTESENSING, 0099-1112/91/5706-677$03.00/0 Vol. 57, No. 6, June 1991, pp. 677-687. 01991 American Society for Photogrammetry and Remote Sensing PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1991 .Geometric Rectification Radiometric Rect~fication DATA ANALYSIS I Quantitative Analysis I cdAFINAL PRODUCT PRESENTATION Raster to Vector .Spatial Error .Vector to Raster .Thematic Error .Sampling .Spatial Autocorrelation .Locational Accuracy \ .Error Matrix 1 Discrete Multivariate Statistics .Reporting Standards FIG.1. The accumulation of error in a "typical" remote sensing information processing flow. of remote sensing data into a GIs; second, to discuss and illus- broad implications for remote sensing and GIS integration. They trate the consequences of error in the decision making and im- provide a sound and necessary mapping basis applicable to plementation processes; and, finally, to recommend important remote sensing imagery. The following discussion identifies some research and development issues to overcome error-related im- of the primary issues involved, such as basic geometric aspects pediments for the incorporation of remote sensing data prod- of imaging, scene environmental considerations, platforms, and ucts into GIs data analysis applications. ground control (Richards, 1986). Illumination geometry can affect image quality and subse- DATA ACQUISITION ERROR quent analyses. Ideally, illumination geometry is constant or Environmental and cultural data may be acquired by either nearly constant throughout an image. In practice, however, ac- in situ or remote measurement. Some data acquisition errors are quisition needs dictate a relatively wide total field-of-view (TFOV), common to any form of data collection and may be introduced resulting in a range of illumination measurement geometries. from a number of sources. Some of these sources, such as at- Passive systems are dependent upon solar illumination. Solar mospheric conditions and the natural variability of the land- elevation and azimuth conditions for aircraft acquisitions can scape, cannot be controlled. Conversely, other types of data significantly limit the duration of suitable acquisition windows collection error, such as geometric or radiometric error, may be (Brew and Neyland, 1980). controlled. One of the most difficult sources of error to quantify Maintaining constant image scale would facilitate image entry is human subjectivity during data analysis and interpretation. into a GIs. Scale variations are introduced by numerous factors, Nevertheless, it is important to have an understanding of the such as off-nadir viewing (tilt for aerial cameras) and terrain type and amount of error possible from all data acquisition sources relief displacement. The instantaneous field of view (IFOV)of and to control it whenever possible. Extensive information may an imaging system also introduces scale variations, which are be found in the literature on many of the data acquisition error most pronounced in wide TFOV systems. Imaging geometry var- sources, e-g., Desachy et al. (1985), Duggin et al. (1985), and ies by sensor type and effects. A brief comparison of sensors Salsig (1990). Data acquisition errors, excluding those errors as- such as aerial cameras, multispectral scanners, and side-looking sociated with natural and human variability, will be briefly dis- airborne radars illustrates this issue. cussed in the following paragraphs. The design of conventional aerial camera systems provides a central perspective geometry and produces radial geometric ef- The processing of multiple data layers in a GIS database is fects, i.e., effects due to relief displacement. Most mapping sys- predicated upon accurate spatial registration between data lay- tems have high quality lenses, filters, and image motion ers. Therefore, it is critical that all remotely sensed data be geo- compensation to achieve film geometric stability during expo- metrically accurate with the same cartographic projection as the sure. Camera systems