time-lapse analysis with earth resistance and electrical resistivity imaging
DESCRIPTION
A presentation by Rob Fry at the DART horizon Scanning workshop on the 17th September 2013TRANSCRIPT
Time-lapse analysis with earth resistance and electrical resistivity imaging
Robert Fry
Ph.D CandidateThe University of Bradford
DART heritage remote sensing horizon scanning workshopSeptember 2013
Current English Heritage geophysical guidelines state that it is preferable to conduct earth resistance surveys:
‘when the moisture contrasts are at their most accentuated’
(David et al. 2008, 27)
• An aspect we do not know how to predict, or how it will effect individual target features
Aim
• To attempt to better understand the earth resistance response over each study area, how and why it changes, and how to predict when archaeological features will best be detected using these techniques.
In this presentation:
• Introduce some of the data collected during the DART fieldwork period and demonstrate the problems with predicting resistance surveys
• Introduce a new methodology for the quantification of geophysical contrast, the key to successful detection
• Demonstrate how each study area interacts with weather, specifically the change in soil moisture
• How a novel analysis of the ERI profiles may help untangle the cause of each anomaly, and create a basis for future modelling
Earth Resistance analysisGeoscan Twin-Probe RM15 multiplexed earth resistance
Creating a robust solution to measuring a contrast factor. Detection * Magnitude = Contrast
Harnhill, Cirencester
Diddington, Cambridgeshire
Cherry Copse Quarry Field
Diddington Clay FieldPasture Field
Clay
soi
ls
Free
-dra
inin
g so
ils
Seasonal effect?June 2011 Vs. June 2012 Cherry Copse
Harnhill, Cirencester
Diddington, Cambridgeshire
Free
-dra
inin
g so
ils
Clay
soi
ls
Cherry Copse Quarry Field
Diddington Clay FieldPasture Field
Harnhill, Cirencester
Diddington, Cambridgeshire
Free
-dra
inin
g so
ils
Clay
soi
ls
Cherry Copse Quarry Field
Diddington Clay FieldPasture Field
How can we quantify the success of the ER response?
• A new 2-part methodology quantifies a contrast factor based on:– A detection test, based on non-parametric statistics which returns a score (from 0-1) determining
how different the data from the ditch is from the background. If the score is above 0.4, the ditch is determined to be detectable within the dataset.
Mann-Whitney Test Calculation of the Z statistic Calculation of Pearson’s r score Calculating the f(r) when r>0.4
– A magnitude test, where the average measurement from the centre of the ditch response, is compared against an average background measurement running parallel to the ditch. The percentage difference between the two measurements is the magnitude of the ditch anomaly.
Calculation of the Specific Population Magnitude Factor
– From these two calculations, the contrast factor of the response is calculated:
Contrast Factor = f(r) * SPMF
Harnhill, Cirencester
Diddington, Cambridgeshire
Free
-dra
inin
g so
ils
Clay
soi
ls
Cherry Copse Quarry Field
Diddington Clay FieldPasture Field
Weather data
Evapotranspiration - Rainfall = Cumulative Moisture Balance
Harnhill, Cirencester
Diddington, Cambridgeshire
Free
-dra
inin
g so
ils
Clay
soi
ls
Cherry Copse Quarry Field
Diddington Clay FieldPasture Field
Harnhill, Cirencester
Diddington, Cambridgeshire
Free
-dra
inin
g so
ils
Clay
soi
ls
Cherry Copse Quarry Field
Diddington Clay FieldPasture Field
Harnhill, Cirencester
Diddington, Cambridgeshire
Free
-dra
inin
g so
ils
Clay
soi
ls
Cherry Copse Quarry Field
Diddington Clay FieldPasture Field
Harnhill, CirencesterFr
ee-d
rain
ing
soils
Clay
soi
ls
Cherry Copse Quarry Field
Diddington Clay FieldPasture Field
Diddington, Cambridgeshire
Harnhill, Cirencester
Diddington, Cambridgeshire
Free
-dra
inin
g so
ils
Clay
soi
ls
Cherry Copse Quarry Field
Diddington Clay FieldPasture Field
At Cherry Copse – there is a negative correlation between contrast factor and weather, the drier the site, the better the earth resistance response – to a point.
Once the ground becomes too dry the contrast factor reduces substantially. Showing a steady decrease in moisture retention from the ditch fills. Prolonged drought would probably result in further decrease in contrast.
At Quarry Field – there is a negative correlation to the weather, with the driest conditions most likely to produce the best contrast factors to sufficiently detect the ditch.
During wet periods, the contrast is significantly reduced.
At Pasture Field – there is a positive correlation to weather, with wetter conditions providing a better ER contrast. The ditch can however be detected throughout the year.
The magnitude of the anomaly appears to be less effected by the weather variables, due to the deeper cut of the ditch.
At Diddington Clay Field – there is a positive correlation to the weather, however, the weak ER response is significant to detect a ditch anomaly only at the wettest points of survey. The rest of the survey the ditch is undetected within the data.
Earth resistance time-lapse and weather data
• Common misconceptions: • There is no ‘seasonality’ of data. The geophysical response does not change as a factor of a change in
seasons. The data will not necessarily produce the same response at similar times of the year. • Wet / damp weather does not always make for the best detection of ditches using ER. • All archaeological features will produce different anomalies, at different times of the year, dependant on the
soils, environmental constraints, and the percolation of moisture within those soils.
• At the sites, the relationship between the cumulative moisture balance and the earth resistance contrast is linear. This linear trend is however not always the same between sites, and does have caveats (such as the drought period at Cherry Copse)
• Understanding why the different sites produce a sufficient detectable contrast with earth resistance techniques will improve the detection of these features for future work and enable modelling of the sites to better ascertain the changing moisture dynamics through the soil profile. • Pursuing an overall model for the ‘best’ survey time for earth resistance survey is extremely difficult, especially for clay soils.
‘When moisture contrasts are at their most accentuated’
• What causes the contrast? – Is it the same for all sites? - Is the rule universal?
Incorporating ERI analysis
FlashRes 64 Electrical Imaging
Using the ERI data to tell more about what is influencing the change in response
Extraction of resistivity data from ERI into separate contexts
Cherry Copse
• The resisitivities between Contexts 3 (silty clay) & 7 (sandy clay) are extremely similar throughout. Suggesting that these contexts do not produce the ditch anomaly
• Main contrast between is seen within the data around Context 9 (sandy loam) and the limestone geology abutting it (Contexts 4&5)
Cherry Copse
Jun-11Jul-1
1
Aug-11
Sep-11
Oct-11
Nov-11
Dec-11Jan
-12
Feb-12
Mar-12
Apr-12
May-12
Jun-12Jul-1
2
Aug-12
Sep-12
0
20
40
60
80
100
120
140
Changing difference between the resistivity of Contexts 4 & 5 (averaged) to Context 9
DIffe
renc
e in
Res
istivi
ty v
alue
s (O
hm.m
)
ER contrast factor
Correlation Contexts
0.25m 0.586*0.5m 0.608*
0.75m 0.7**1m 0.658*
Difference in resistivity between Contexts 4&5 and 9
Earth resistance contrast factor Correlation
Strong correlations:
•The difference in resistivity between the anomaly producing contexts correlate well with the ER contrast factors
• This is best seen at twin-probe spacing of 0.75m and 1m (known depth of Context 9 = 0.68-0.92m)
• The time-lapse correlation shows there to be a 4 week delay between the surface increase in CMB to the effect on the resistivity values of the contexts at this depth
Quarry Field
• Only significant difference between contexts occurs within Context 1, between the topsoil directly above the ditch cut and the topsoil around it.
• The higher resistivity of the soil above the ditch creates the resistance anomaly identified as the ditch (not the ditch itself)
• A properly relating to the sump created by the field drain to which the ditch was used for
Quarry Field
• The ER contrast factor results correlate strongly to the calculated difference between the samples from context 1. (0.772 sig. 0.01 level)
• The effect is best seen at the smaller probe separation linking to the topsoils.
• Sump creates a downward force on the moisture above the ditch, which is most apparent during dry periods, becoming drier and a higher resistivity.
Conclusions and future workA context specific extraction methodology can help disentangle the results of the ER data, and provide greater clarity to the specific problems of detection within each study area.
Each survey area is different(!) – this makes life more interesting, however means any prediction to a ‘best survey time’ is limited.
At Cherry Copse, the ditch anomaly is identified manly due to the differences in resisitivities between the archaeological Context 9 and geological Contexts 4 and 5. This change is most effective after a 4 week delay from weather variables.
The anomaly at Quarry Field appears to have been created from a difference in soil resistivity directly above the ditch cut, rather than directly from the ditch fills themselves. A sump has been created which forces moisture from the topsoil downwards, creating a high resistance feature.
The free draining soils appear to produce a clear anomaly throughout the fieldwork period, although the magnitude of the anomaly created changes differently to weather variables.
The clay soils have indeed been ‘difficult’ for detection, however the ditches situated on these soils can be detected, provided the survey is timely, and after consideration of the weather variables and history of the sites.
Importance of both the soils and weather variables – can this be spread to a wider area?
Future Work:The research (and the PhD) is unfinished and still at a relatively early stage Modelling the data to create prediction and percolation routines for the sites (David Jordan has already started looking into this)Lots of scope for future research, different soils, scenarios, better temporal resolution etc.
Thank you for listening