Archaeological applications of multi/hyper-spectral data: challenges and potential

Archaeological applications of multi/hyper-spectral data challenges and potential Anthony Beck School of Computing, University of Leeds, UK DART Project Champion

Overview Brief history of aerial remote sensing The Electromagnetic Spectrum and Multi/Hyper spectral sensors Sensor resolution Archaeological site detection and contrast The use of non-visual remote sensing for heritage detection DART What wont be discussed but is in the paper Relationship of multi/hyper spectral sensors (and multi-sensor approaches) to landscape survey Use of multi-hyper spectral techniques for landscape theme generation DEMs, Land use and cover (Communication networks, Hydrology networks, Settlements, Field Systems, etc.) and Soil/Geology maps

Aerial remote sensing Traditional archaeological remote sensing technique Low level aerial platform Handheld SLR and digital cameras Reliance on oblique photography Optical and Near Infrared wavelengths Used since early 1900s Early pioneers established many of the techniques still used today Crawford, UK Keiller, UK Poidebard, Middle East

Map of GPS tracks Slide courtesy of Stefano Campana

Aerial archaeology: Pros and Cons Most successful archaeological detection technique Reliant on specific seasonal and environmental conditions Increasingly extreme conditions are required for the detection of ‘new’ sites Low understanding of the physical processes at play outside the visual wavelengths Significant bias in its application in the environmental areas where it is productive (for example clay environments tend not to be responsive) Surveys don’t tend to be systematic Interpretation tends to be more art than science

EM spectrum and Aerial Photography (Log scale) Traditional aerial photography Tiny Tiny Tiny

Multi and Hyperspectral Sensors Dimension and number of recordable wavelengths. There is NO archaeological spectral signature. Allows one to select the portion of the spectrum where there is the most contrast. Hence, an improvement in archaeological detection. Poorly understood outside the visual

Mono-spectral (panchromatic) remote sensing 1 Dimension 1 wavelength Limited definition Archaeology is poorly understood outside the visual There is NO archaeological spectral signature.

Multi-spectral remote sensing Many dimensions Many wavelengths There is NO archaeological spectral signature. Allows one to select the portion of the spectrum where there is the most contrast. Hence, an improvement in archaeological detection.

Hyper-spectral remote sensing Hyper (lots) of dimensions Hyper (lots) of wavelengths. Allows more freedom in spectrum/contrast identification. Hence, an improvement in archaeological detection. Spectral signature allows identification of soil/vegetation There is NO archaeological spectral signature.

There is NO archaeological spectral signature

Sensor Resolution Multi/Hyper spectral refers to the SPECTRAL resolution There are 3 other axes of resolution Spatial resolution Temporal resolution Radiometric resolution

Spatial Resolution Sensor resolving power Distance from object Object size Ground resolution Implications well understood

Temporal resolution How often a sensor records a particular area Seasonal Diurnal Variable understanding

Radiometric Resolution Sensitivity of sensor to discriminate signal strength Landsat 8bit (256 value bins), Ikonos 11bit (2048 values) Much archaeological subtlety can be lost in single band imagery with high radiometric depth (geophysics) 20-30 shades of grey can be discriminated by the eye Implications poorly understood

Multi/Hyper spectral summary Non-visual remote sensing has huge potential for the detection of archaeological features However, traditional aerial photographic techniques are not a good starting point Requires a thorough understanding of how archaeological contrast is produced so that the correct sensor can be applied at the correct time: (De) Formation processes Local (contrasting) matrix Ambient conditions Sensor characteristics

First Principals - Archaeological Site Detection Discovery requires the detection of one or more site constituents which are sufficient to suggest that a site might be present. The important points for archaeological site detection are that: Archaeological sites are physical and chemical phenomena. There are different kinds of site constituents. The abundance and spatial distribution of different constituents vary both between sites and within individual sites. These attributes may be masked or accentuated by a variety of other phenomena. Importantly from a remote sensing perspective archaeological site do not exhibit consistent spectral signatures

First Principals – Archaeological Sites Archaeological sites show up as: Structures Shadow marks Soil marks Crop marks Thermal anomalies Influenced by effects of: Weather Season Soil type and soil moisture Crop type

First Principals – Archaeological Site Examples Micro-Topographic variations Soil Marks variation in mineralogy and moisture properties Differential Crop Marks constraint on root depth and moisture availability changing crop stress/vigour Proxy Thaw Marks Exploitation of different thermal capacities of objects expressed in the visual component as thaw marks Now you see me Now you dont

Why Non-Visual Remote Sensing? Many archaeological contrasts are easier to identify in non-visual wavelengths: Crop stress and vigour Soil mineralogy Moisture Temperature Use of non-visual wavelengths has a number of benefits: Can extend the window of opportunity for archaeological identification May not require extreme environmental conditions May be applicable in ‘non-responsive environments’

Contrast and Archaeological Detection The nature of archaeological residues and their relationship with the immediate matrix determines how easily residues can be detected. Detection requires the following: A physical, chemical or biological contrast between an archaeological residue at its immediate matrix A sensor that can ‘detect’ this contrast Sensor utilised during favourable conditions i.e. you’re unlikely to detect thaw marks in summer using photography! Although you could detect the underlying thermal anomalies using a different sensor at this time. Here the underlying process remains the same (a thermal variation) and the detecting sensor is in part determined by the environmental conditions. It is this contrast between an archaeological feature and its matrix that one is wanting to observe.

Detection and (De-)Formation Processes Unfortunately archaeological sites do not produce distinct Spectral Signatures Rather: produce localised disruptions to a matrix The nature of these disruptions vary and include: Changes to the soil structure Changes to moisture retention capacity Changes in geochemistry Changes in magnetic or acoustic properties Changes to topography At least one of these disruptions will produce a contrast which is detectable

Environmental and ambient conditions Local conditions structure how any contrast difference is exhibited: Soil type Crop type Moisture type Diurnal temperature variations Expressed contrast differences change over time Seasonal variations impact on the above (crop, moisture, temperature in particular) Diurnal variations: sun angle (topographic features), temperature variations Exacerbated by anthropogenic actions Cropping Irrigation Harrowing

DART aims to understand What are the factors that produce archaeological contrasts?

DART aims to understand How these contrast processes vary over space and time?

DART aims to understand What causes these variations?

DART aims to understand How can we best detect these dynamic contrasts (sensors and conditions)?

DART Research Methodology Collect data from and around residues at different times under different conditions Develop soil and physical models to determine: under what environmental conditions contrast is strongest where this contrast is expressed in the sensor spectrum how to calibrate a sensor to improve residue detection Develop tools to: detect currently undetectable residues (those in ‘difficult’ soils) improve residue detection capacity in well-studied areas improve the search options for archival resources Evaluate the results: Using the decision tools to programme hyperspectral and geophysical surveys

DART Outcomes If successful will allow the determination of: What archaeological components can be detected Using which sensors Under what conditions Production of a suite of tools which will allow Cultural Resource Managers to: Have a thorough understanding of the physical, chemical and biological phenomena involved in detection To further understand how this varies over time and under different ambient conditions Systematically detect as complete a range of residues as possible Programme surveys and manage costs more effectively It is an Open Science project Wherever practicable all data will be in the public domain as soon as possible

DART Overview 3 year Science and Heritage project (AHRC and EPSRC funded) £800k FEC application 40 months of researcher time 3 PhD Studentships Soil dynamics and geophysical prospection Knowledge-based approaches to archaeological remote sensing Modelling sensor responses from physical measurements to enhance electromagnetic archaeological detection Consortium consists of 25 key academic, heritage and industry organisations Computer vision Geophysics and remote sensing Knowledge engineering Policy Practitioners Researchers Soil science

Summary: Detection Discovery requires the detection of one or more site constituents which are sufficient to suggest that a site might be present. The important points for archaeological residue detection are that: Archaeological sites are physical and chemical phenomena. There are different kinds of site constituents. The abundance and spatial distribution of different constituents vary both between sites and within individual sites. These attributes may be masked or accentuated by a variety of other phenomena. Importantly from a remote sensing perspective archaeological site do not exhibit consistent spectral signatures

Summary: Multi/Hyper Spectral Many archaeological contrasts are easier to identify in non-visual wavelengths: Crop stress and vigour Soil mineralogy Moisture Temperature Use of non-visual wavelengths has a number of benefits: Can extend the window of opportunity for archaeological identification May not require extreme environmental conditions May be applicable in ‘non-responsive environments’

Conclusions Multi/Hyper spectral technologies have the potential to improve archaeological detection However requires an understanding of: 1. Nature of the archaeological residues Nature of archaeological material (soil, stone etc.) Formation/Deformation processes Nature of the surrounding material with which it contrasts 2. Sensor characteristics Spatial, spectral, radiometric and temporal How these can be applied to detect 1. 3. Environmental characteristics Complex natural and cultural variables that can change rapidly over time Try to understand the periodicity of change Without an understanding of the above expensive techniques may be utilised in inappropriate conditions DART is the first step in providing this understanding

Archaeological applications of multi/hyper-spectral data challenges and potential Anthony Beck School of Computing, University of Leeds, UK DART Project Champion Follow DART and its outputs using the following: Website: www.comp.leeds.ac.uk/dart Blog: dartheritage.wordpress.com Twitter: follow DART_Project SlideShare presentations: www.slideshare.net/DARTProject Scribd documents: www.scribd.com/dart_project

Archaeological applications of multi/hyper-spectral data: challenges and potential

More Related Content

What's hot (20)

Viewers also liked (18)

Similar to Archaeological applications of multi/hyper-spectral data: challenges and potential (20)

More from DART Project (20)

Recently uploaded (20)

Archaeological applications of multi/hyper-spectral data: challenges and potential