References

Sensit data - Processing Sensit Field Data

   Data Processing and Calibration document - View DPC.pdf

The "Data Processing and Calibration" document contains theory of operation, catcher information and technical information about the eroding mass sensor. Last update Feb 10, 2005.

  Movie of Owen's Dry Lake data. This movie is the first animation of actual girded erosion data covering a large area. Data was taken by the California Great Basin Unified Air Pollution Control District (GBAPCD). High speed internet connection preferable.

The large grid (lake bed is approximately 12mi x 23mi) of meteorological towers collecting Sensit data (one kilometer gird locations) is shown in the adjacent image.  This monumental effort undertaken by GBAPCD is the largest erosion monitoring effort ever undertaken. The project is currently on-going with remarkable success controlling saltation,  migration and decreasing airborne particulates.

Understanding Sensit raw data

Understanding Sensit data is the key to success.

1) The dynamic range (a glass half empty or half full?) - Consider the following; The sensor's primary output is linear response to kinetic energy. Considering the simple equation for kinetic energy as 1/2mv^2, it can be seen that particle velocity is a powerful influence. Also, mass (simplified as a sphere) varies as the cubed of particle diameter shown by the equation for the volume of a sphere as 4/3 pi r^3.

Now consider the characteristics of typical saltation. Particle diameters of 100 microns to 1 millimeter. Particle velocities ranging from 5 m/s to 50 m/s. This represents a kinetic energy dynamic range of 10^5. It is extremely difficult for any linear electronic measurement device to operate over a dynamic range of 10^5 without changing ranges.

Sensit accommodates this dynamic range by attempting to reduce the minimum detectable signal to near-zero and increasing the maximum signal resolution through the use of time integration. Both techniques are common for very sensitive measurements of a signal which is comprised of individual pulses such as photon counting or radiation counting.

Near-zero data resolution is accomplished by taking a measurement so sensitive that background noise become a substantial part of the signal. To measure a signal that is smaller than background noise, one must measure the background and then subtract it from the total signal. To make sure this background signal occurs often enough for the data logger to count it, we add a small amount of artificial background. The actual value of this background signal subtracted from the final signal is of no consequence. Remember that the final Sensit output signal is in the form of pulses, so we want at least 2 to 10 pulses per data logger sampling period.   The background signal varies between sensors of different sensitivities and is sometimes misunderstood by those who see it vary between sensors and do not understand it's derivation.

Maximum signal resolution is extended by the inherent nature of integration. The output format is in the form of pulses.  The largest number you record (for a typical event) is determined by length of time (i.e., integration) you count these pulses. This is set by the sampling period of the data logger. A longer integration period will also improve the minimum signal resolution by allowing a more accurate value for the average background signal.

This graph of Sensit response vs. mass data from a prototype real-time weighing sampler made by Charles Yates (USDA technical engineer) demonstrated the sensor's response to true mass was better than expected (r^2=0.997).

The scale-type weighing sampler used as a saltation sensor was abandon due to un-resolvable scatter caused by wind buffeting of the scale mechanism, long term drift, corrosion problems and limited sampler volume. But this rare data set was the first demonstration of the sensors ability to measure saltation.

Relationship between Sensit response vs. BSNE Catcher