ARCHAEOLOGICAL LAND USE CHARACTERIZATION USING MULTISPECTRAL REMOTE SENSING DATA Ivan E. Villalon-Turrubiates 1 , Member, IEEE, and Maria J. Llovera-Torres 2 1 Universidad de Guadalajara, Centro Universitario de los Valles, México, villalon@ieee.org 2 Universidad Autónoma de San Luis Potosí, México, llovera_torres@alumnos.uaslp.edu.mx ABSTRACT Much of human history can be traced through the impacts of human actions upon the environment. The use of remote sensing technology offers the archeologist the opportunity to detect these impacts which are often invisible to the naked eye. The extraction of remote sensing signatures from a particular geographical region allows the generation of geophysical signature maps; this can be achieved using an accurate and recently developed multispectral image classification approach based on pixel statistics for the class description, which is referred to as the Weighted Pixel Statistics method. This paper presents the prospective study for archaeological land use characterization using multispectral remote sensing data provided by SPOT-5 imagery. The results obtained with this study probe the efficiency of the classification technique. 1. INTRODUCTION Generations of archaeologists have longed for some way of learning from archaeological sites without actually having to dig them. The prominent archaeologist Lewis Roberts Binford, Ph.D., once said "ideally, we should have an X-ray machine which would allow us to locate and formally evaluate the range of variation manifest in cultural features" [1]. Recently, Binford's elusive X-ray machine has been actualized in a series of increasingly and highly sophisticated remote sensing contrivances. These new techniques can (prior to excavation) provide information of where the archeological sites are and what they contain. Surface studies are necessary to obtain data that excavations cannot provide. Some examples are the applications of aerial and satellite photography, which shows a wide panorama of the archaeological site, allows examining ground marks, find walls from former occupations, etc. [2]. In some cases is possible to identify in the ground the necessary elements to study the materials and establish a chronology (even a tentative one), shows the environmental relationship with the site, important aspects as space dimension and distances to another related sites [3]. The extension and complexity of the site are necessary elements for planning and managing the archaeological research site. The geophysical techniques employed in prospecting studies are important because provides information to the researcher to help solving difficulties, to act within large terrain extensions, to detect archaeological contexts and maximize the excavation efficiency [4]. Considerable progress has been made generally in the application of remote sensing (RS) techniques to both research and operational problems for urban assessment, urban planning and natural resources management. Modern applied theory of signal and image processing for land cover and land use analysis is now a mature and well developed research field, presented and detailed in many works ([5] thru [8] and the references therein are only some indicative examples). Although the existing theory offers a manifold of statistical techniques to tackle with the particular geophysical monitoring problems, in many applications areas there still remain some unresolved crucial theoretical and data processing problems. One of the most important problems to be solved is particularly related to the extraction of physical characteristics for applications in archaeological land use characterization. Modern digital signal and image processing techniques are currently used by archaeologist to detect the impacts of human actions upon the environment. This information can be used to address issues in human settlement, environmental interaction, and climate change [9]. Archeologists want to know how ancient people successfully adapted to their environment and what factors may have led to their collapse or disappearance. Remote sensing can be used as a methodological procedure for detecting, inventorying, and prioritizing surface and shallow-depth archeological information in a rapid, accurate, and quantified manner [10]. The application of an accurate tool recently developed in [11] for supervised segmentation, classification and quantification of archaeological land use (ALU) using multispectral remote sensing (MRS) imagery is based on the analysis of pixel statistics, and is referred to as the weighted pixel statistics (WPS) method.    ,((( ,*$566 