Data Acquisition Chapter 2

Data Acquisition 1 st step: get data 1 st step: get data – Usually data gathered by some geophysical device – Most surveys are comprised of linear traverses or transects Typically constant data spacing Typically constant data spacing Perpendicular to target Perpendicular to target Resolution based on target Resolution based on target Best for elongated targets Best for elongated targets – When the data is plotted (after various calculations have been made): Profile

Grids When transects are combined a grid can be formed. When transects are combined a grid can be formed. – Good for round or blob-shaped targets Or if target geometry is unknown Or if target geometry is unknown – Useful for making contour maps – Allows transects to be created in multiple directions

Data Reduction Often the raw data collected is not useful. Often the raw data collected is not useful. – Data must be converted to a useful form Removing the unwanted signals in data: Reduction Removing the unwanted signals in data: Reduction Targets are often recognized by an “anomaly” in the data Targets are often recognized by an “anomaly” in the data – Values are above or below the surrounding data averages. Not all geophysical targets produce spatial anomalies. Not all geophysical targets produce spatial anomalies. – E.g. seismic refraction produces travel time curves  depth to interfaces Also a type of reduction. Also a type of reduction.

Signal and Noise Even after data is reduced, a profile may not reveal a clear anomaly due to noise. Even after data is reduced, a profile may not reveal a clear anomaly due to noise. – Noise: Unwanted fluctuations in measured data. May be spatial or temporal May be spatial or temporal What causes noise? What causes noise? – Signal: The data you want, i.e. no noise. Noise can be removed using mathematical techniques Noise can be removed using mathematical techniques – Stacking – Fourier Analysis – Signal Processing Magnetic or Gravity profile

Stacking Stacking is useful when: Stacking is useful when: – Noise is random – Signal is weak – Instrument is not sensitive If noise is random If noise is random – Take multiple readings – Sum the readings – Noise cancels out Destructive Interference Destructive Interference – Signal should add Constructive Interference Constructive Interference Stacking improves signal to noise ratio Stacking improves signal to noise ratio – Commonly used with numerous techniques.

Resolution Even if you have a good signal to noise ratio, detection of your target depends on your resolution. Even if you have a good signal to noise ratio, detection of your target depends on your resolution. – Know what you are looking for before you begin – Know the limits of your data resolution

Modeling Most geophysical data is twice removed from actual geological information Most geophysical data is twice removed from actual geological information – Reduced data is modeled Models Models – Aim to describe a specific behavior or process – Are only as complex as the data allows Occam’s Razor: “Entities should not be multiplied unnecessarily” Occam’s Razor: “Entities should not be multiplied unnecessarily”

Model Types In the most basic sense models come in two flavors: In the most basic sense models come in two flavors: – Forward model Given some set of variables, what is the result. Given some set of variables, what is the result. I.e. you input the “cause” and some “effect” is produced I.e. you input the “cause” and some “effect” is produced – Inverse model Given some measurements, what caused them Given some measurements, what caused them You know the “effect”, try to determine the “cause” You know the “effect”, try to determine the “cause” Often involves mathematical versions of “guess and check” Often involves mathematical versions of “guess and check” Depth = D Fault Slip GPS Station Motions

Model Types Models also come in several flavors based on technique Models also come in several flavors based on technique – Conceptual Model Models an idea…no math/physical parts Models an idea…no math/physical parts – Analog Model A tangible model “scaled” to reproduce geologic phenomena A tangible model “scaled” to reproduce geologic phenomena – Empirical Model Based on trends in data Based on trends in data – Analytical Model Solves an equation Solves an equation Usually deals with simple systems Usually deals with simple systems – Numerical Model Computer-based approximations to an equation. Computer-based approximations to an equation. – Thousands, millions, or billions of calculations Can handle complex systems. Can handle complex systems. Analog Model Empirical Model From Wells & Coppersmith 1994

Non-Uniqueness of Models Typically, multiple models can fit data Typically, multiple models can fit data – So any given model is non- unique – Distinguish between models based on Match with geologic data Match with geologic data Model with least parameters (most simple) Model with least parameters (most simple) Data has limited resolution Data has limited resolution – Surveys must be finite – “Blurs the picture” – Omission of detail emphasizes key features

Geologic Interpretation After data is collected and modeling is complete the results must be interpreted into the geological context. After data is collected and modeling is complete the results must be interpreted into the geological context. Use all available data. Use all available data. – Don’t only look, when you can hear and touch! Interpretations are also typically non-unique Interpretations are also typically non-unique – Many geologic materials have similar properties. – Best interpretations use all available data, geologic, geophysical, chemical, etc…