Surveying Prospection for Archaeology Environmental Science Spatial sampling

Surveying & Prospection for Archaeology & Environmental Science Spatial sampling & soil properties Phil Buckland

Contents • Soil chemistry & physical properties as proxy data sources • What are (spatial) samples - why are they taken? - how are they taken? - what can they tell us? • Sample data - examining, - manipulating, - interpreting.

Soil chemistry & properties. . . Proxy indicator: a measurable variable that tells us about conditions or changes in the past that we cannot directly measure. Commonly used in environmental archaeology, quaternary geology, environmental change analysis (& monitoring)

Soil chemistry & properties. . . Biological proxies - fossil insects, plant macrofossils, molluscs, tree rings. . . Chemical proxies - phosphates, oxygen isotopes, carbon isotopes (14 C), other isotopes & ratios Physical proxies - organic content, magnetic susceptibility, colour (full spectrum), dust, particle size, sedimentation, raised beaches

Soil chemistry & properties. . . phosphates (P) - element (phosphorus) - organic and inorganic - measure amount & ratios in sediments (citric acid extraction) using spectrophotometer organic content = Loss On Ignition (LOI) - ratio of organic: inorganic matter in sediment - measure by burning and calculating weight loss magnetic susceptibility (MS) - ability of material to sustain an applied magnetic fields. - measure induced magnetic field in sample compared to applied field Clark, A (1990/2000) ‘Seeing beneath the soil’

Soil chemistry & properties. . . phosphates (P) - Phosphate degrees P° - increased amounts often indicate human activity - linked to decay of organic materials (organisms) - Decay leads to: 1) release of phosphate ions (PO 4) 2) ions bind to soil particles phosphates Archaeological site? distance Background level e. g. waste, manuring, food storage - past & present (pollution)

Soil chemistry & properties. . . organic content = Loss On Ignition (LOI) - % - increased amounts often indicate human activity - linked to decay of organic materials (organisms) - accumulations of organic matter lead to increase Bog (mire)? LOI Archaeological site? Confirm with macrofossil & insect analyses distance e. g. waste, manuring, food storage

Soil chemistry & properties. . . magnetic susceptibility (MS) - SI (no units) - dependent on iron content of soil - heating increases MS due to oxidation of iron - erosion, ploughing etc. can expose different materials Archaeological site? MS Road (modern) distance e. g. fire, industry, pollution - past & present

Samples Sample: ’A sample is that part of a population which is actually observed. ’ www. wikipedia. org Population: ’. . . a set of potential measurements or values, including not only cases actually observed but those that are potentially observable’ www. wikipedia. org More simply: A sample is that part of reality that we actually measure

Samples Examples: Population Sample All people in Sweden Every 100 th person in Sweden A 10 hectare meadow 100 randomly placed 1 m squares in the meadow Phosphate levels in an area 1 km around an archaeological site Soil samples taken at 20 m intervals throughout the area An infinite number of rolls of two dice 100 rolls of two dice Fluctuations in heavy metal levels in the water of the Bay of Bothnia Weekly heavy metal test samples from water 5 km East of Holmsund How well the samples reflect the population requires careful consideration - and can result from good project design.

Samples Examples: Population Sample All people in Sweden Every 100 th person in Sweden A 10 hectare meadow 100 randomly placed 1 m squares in the meadow Phosphate levels in an area 1 km around an archaeological site Soil samples taken at 20 m intervals throughout the area An infinite number of rolls of two dice 100 rolls of two dice Fluctuations in heavy metal levels in the water of the Bay of Bothnia Weekly heavy metal test samples from water 5 km East of Holmsund How well the samples reflect the population requires careful consideration - and can result from good project design.

Samples Examples: Population Sample All people in Sweden Every 100 th person in Sweden A 10 hectare meadow 100 randomly placed 1 m squares in the meadow Phosphate levels in an area 1 km around an archaeological site Soil samples taken at 20 m intervals throughout the area An infinite number of rolls of two dice 100 rolls of two dice Fluctuations in heavy metal levels in the water of the Bay of Bothnia Weekly heavy metal test samples from water 5 km East of Holmsund How well the samples reflect the population requires careful consideration - and can result from good project design.

Samples Examples: Population Sample All people in Sweden Every 100 th person in Sweden A 10 hectare meadow 100 randomly placed 1 m squares in the meadow Phosphate levels in an area 1 km around an archaeological site Soil samples taken at 20 m intervals throughout the area An infinite number of rolls of two dice 100 rolls of two dice Fluctuations in heavy metal levels in the water of the Bay of Bothnia Weekly heavy metal test samples from water 5 km East of Holmsund How well the samples reflect the population requires careful consideration - and can result from good project design.

Samples Examples: Population Sample All people in Sweden Every 100 th person in Sweden A 10 hectare meadow 100 randomly placed 1 m squares in the meadow Phosphate levels in an area 1 km around an archaeological site Soil samples taken at 20 m intervals throughout the area An infinite number of rolls of two dice 100 rolls of two dice Fluctuations in heavy metal levels in the water of the Bay of Bothnia Weekly heavy metal test samples from water 5 km East of Holmsund How well the samples reflect the population requires careful consideration - and can result from good project design.

Samples & variation The things we measure vary in different ways. . . Continuous variables: - Vary continuously - Often MEASURABLE 0 Discrete variables: - Stepwise, or noncontinuous variation - Often COUNTABLE 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 14

Samples & variation The things we measure vary in different ways. . . Discrete sampling of continuous variables. low resolution high resolution

Samples & variation & interpolation The things we measure vary in different ways. . . Interpolation allows us to simulate/approximate the original variation low resolution high resolution . . . by assuming things about the real distribution.

Sampling Strategies Must consider: • Project aims and how they can be achieved • Variables to be measured and how they behave in reality • Scientific theory (& statistical ground rules) • Avoid bias • Encompass areas outside of the immediate area of investigation (background/reference samples)

Sampling Strategies Method e. g. pro’s con’s Random • statistically robust (no intentional bias) • stratified sampling is scalable • can (randomly) miss areas of interest • difficult to implement* Grid • practical in field • uniform coverage • good statistics • may miss higher resolution detail Strategic • can support other proxies from samples • easy for archaeologists • biased (stat’s unsound) • tends to prove nothing Strategic grids • can target interest • can be interpolation • good compromise problems (worst case = between detail and undetected) statistical robustness • some bias possible • easy to cover back -ground & features *without total station or good GPS

Sampling strategies Good sampling strategy can allow: • a good level of realism in models (reconstructions/interpretations) • measure and control of errors • valid use of summary and advanced statistics • results that stand up to rigorous interrogation • useful models for interpretation

Interpolation ‘interpolation is a method of constructing new data points from a discrete set of known data points’ www. wikipedia. org Translating sample point data into continuous surfaces

Interpolation A surface is a 3 dimensional representation of the values of any variable in two dimensional space (at an instance in time). e. g. • ground temperatures at a specific time • phosphate levels in soil • the ground surface = topography • the sea surface … although the two dimensional space does not have to be geographical… e. g. climate space

Interpolation A surface is a 3 dimensional representation of the values of any variable in two dimensional space (at an instance in time). … although the two dimensional space does not have to be geographical… Summer temperature Climate space map showing % of beetle species (in a sample) that tolerate different temperatures. Temperature range

Interpolation methods Numerous methods exist. Deterministic methods: ‘assign values to locations based on the surrounding measured values and on specified mathematical formulas that determine the smoothness of the resulting surface’ ESRI. E. g. • Spline • Inverse distance weighted Geostatistical methods: ‘are based on statistical models that include autocorrelation (the statistical relationship among the measured points)’ ESRI. E. g. • Kriging Most methods can be tuned to application

Interpolation methods - example Northing Prospection area in Skåne - ca. 600 x 200 m Easting Sample grid - semi-regular (sub-regular)

Interpolation methods - example Topography - Interpolation by Ordinary Kriging

Interpolation methods - example Topography - Interpolation by Inverse distance weighting

Interpolation methods - example Kriging Appears smoother Inverse distance weighting Appears blotchy, unrealistic? Ridge vs. mound

Interpolation methods - example Kriging Inverse distance weighting Kriging identifies a gradient W-E and applies it to the missing values IDW missing values fall off with distance from known points Kriging - uses relationship between data values of each point to every other point to construct values for missing points. Inverse distance weighting - missing values are a simple mathematical function of the value of the nearest point. More info: Arc. GIS help files; Internet; Recommended literature

Interpolation methods - example Implications? • omit (mask) unsampled area • probably use Kriging (but may have to adjust parameters) Kriging Inverse distance weighting

Interpolation methods - example Phospates (total phosphates - Ptot) Minor anomalies Major anomalies Human occupation sites?

Interpolation methods - example Loss On Ignition (% - weight loss after burning) High organic content: peat bog (mire)? Low organic content: erosion? mineral soil?

Interpolation methods - example Compare proxies. . . identify similarities in patterns. . . Some similarities in lows & highs: • variables support each other? • or autocorrelation - variables influence each other?

Interpolation methods - example Magnetic Susceptibility (MS) Low values due to bog? (waterlogged - reduced iron) Prehistoric fireplaces?

Interpolation methods - example Probable area of past human occupation

Considerations when interpreting Sediments move with time - so signals may be displaced Colluviation - translocation of sediments by gravity Farm n o i s ro E Deposition Bog

Considerations when interpreting Sediments move with time - so signals may be displaced Must be considered when interpreting proxy indicators Farm n o i os Er Deposition Bog Occupation phase Present day Phosphates

Considerations when interpreting Other considerations: • Ploughing, digging & erosion may expose or mix subsoils with different properties • Water & wind erosions & associated deposition may cover or destroy evidence - leaving an incomplete record Water deposited sediments -> Alluvial deposits Wind deposited sediments -> Aeolian deposits Gravitationally deposited sediments -> Colluvial deposits • Proxies may interact - i. e. values may be related by physical & chemical processes - ‘autocorrelation’ in statistics • Rates of decay, weathering & transportation will vary depending on climate and sediments/bedrocks • Geostatistics may give false positives if not used properly!

Integration of maps

Integration of maps

Integration of maps Rockart