Alberta Hail Project Archive Project

Cloud Physics Aircraft 2D Data Description

2-D record definition

A 2-D record contains 1024 slices. Each slice is 32 elements wide and each element is in a shadowed or non-shadowed state. A slice is sampled once per True Air Speed (TAS) clock pulse. Each record contains image, time and reference slices.

An image consists of all the slices between a reference slice and the following time slice. These slices are referred to as image slices. The first image slice contains at least one element in the shadowed state. Time and reference slices appear in pairs.

2-D records are composed by a probe which is under limited control of the data acquistion system. The computer based system of the Alberta Research Council differs from the DAS.

TAS clock pulse

A TAS clock pulse comes from the data system and not from the probe. The TAS clock is a distance clock. The ideal frequency of the TAS clock occurs when the aircraft travels the same distance as 1 array element width in order to make the element square.

An array element represents a square cross sectional area. There are 32 elements across the aircraft path. The size of the area is dependent upon 2 factors:

the optics of the probe which determines the distance across the path (the Y-direction), and
the TAS clock frequency which determines the distance along the path (the X-direction).

The time_slice

The time slice consisting of the 32 elements contains two pieces of information. The first 8 elements can describe either the probe, or be an extension of the time elements, or give specialized information. (For example; in the 2-D C probe these elements describe the phase discrimination code.) The next 24 elements are a binary representation of the number of completely unshadowed image slices omitted between the previous reference slice and the first slice of the image terminated by the time slice.

picture
               descriptor elements
             1      8
             --------------------------------
            !       *                        !
             --------------------------------
                     9                      32  
                           time elements

An example of the layout of a time slice. The first 8 elements describe the probe in some way, or can be an extension of the time elements. The next 24 elements describe the timing information.

reference slice

A reference slice follows a time slice, and has the 24 time elements in the shadowed state.

Example Picture:

                11111111111111111111111111111111
                11111111111111111111111111111111
                11111111111111111111111111111111
                11111111000000000000000000000000 time slice 
                11111111000000000000000000000000 reference slice
                11111111000000000111111111111111
 Image A        11111100000000000001111111111111 image slice
                11110000000000000000001111111111
                11100000000000000000000011111111
                11100000000000000000011111111111
                11110000000000000000011111111111
                11111111000000000001111111111111
                11111111111000001111111111111111
                11111111111111111111111111111111
                11111111111111111111111111111111
                11111111111111111111111111111111
                11111111111111111111111111111111
                11111111000000000000000000000000 time slice A
                11111111000000000000000000000000 reference slice B
 Image B        11111111111111100011111111111111
                11111111111111000011111111111111
                11111111111111100111111111111111
                11111111111111110111111111111111
                11111111111111111111111111111111
                11111111111111111111111111111111
                11111111000000000000000000000000 time slice B
                11111111000000000000000000000000 reference slice C
 Image C        11111111111111111111110000001111
                11111111111111111111111000001111
                11111111111111111111111000111111
                11111111111111111111111000011111
                11111111111111111111111100001111
                11111111111111111111111110011111
                11111111111111111111111111111111
                11111111111111111111111111111111

How do you determine sample volume?

The sample volume is probe dependent, and defined as the product of the probe's aperture, depth of field, the true air speed and the time interval.

True-airspeed clock pulses between images and image slices are accumulated in order to accurately determine the absolute time and sample volume of images within a record. The number of particles larger than 32 elements in size is reduced proportionally to their size to account for an effective sampling volume.

How do you handle overloads?

We use a Science Engineering Assoc. (SEA) interface rather than a PMS DAS to record our 2D data. Consequently, in our data there are no overloads. The following section briefly describes our data and the acquisition system.

COMPOSING A RECORD

As in DAS based systems, 2D records are composed by a probe which is under limited control of the data acquisition system. In the INTERA/ARC computer based system (CBS), the TAS clock is started when the computer signals the interface to collect a record (for example one record collected every half second). This begins acquisition of slices in a buffer of the accumulation of clock pulses. There are two buffers in a probe, each capable of holding a record. When the last slice has been stored in one buffer, the probe automatically begins filling the other buffer, and notifies the data system. Then the SEA interface asks the probe to transfer data from the full buffer. At this time the interface turns the TAS clock off, which suspends acquisition of all slices in the probe's active buffer. The computer is notified when 1024 slices (1 record, 4096 bytes) have been transferred to memory. The computer may delay the start of the clock to control the number of records acquired of may restart the clock as soon as computer memory is available.

IMPLICATIONS

With the CBS, data acquisition by the probe is disabled between records, so that an image will never span 2 records. The time when the TAS clock is started is recorded. The end time of the record occurs when the transfer to memory is complete. This transfer takes approximately 73 milliseconds. Since the interface cannot turn the TAS clock off instantly, some image slices are collected in the active buffer. Therefore the time slice for the first image in each record is not well defined.

DIFFERENCES WITH DAS

In a PMS DAS-based system, the probe notifies the DAS when the buffer being filled is half-full. If the previous buffer is still being transferred to tape the DAS stops the TAS clock, starts an overflow counter and sets an overflow flag. When tape transfer is complete, the TAS clock is restarted and the overflow timer stopped. (The flag/timer are recorded in the next slow record.)

With the DAS, at low particle concentrations the TAS clock is never disabled. However, at concentrations so high that the data rate exceeds that ot the tape, the TAS clock will be disabled at the midpoint of a record, so that incomplete particle image and time information may be recorded. There is no indication in the 2D record that thie has occured. Furthermore, if more than one 2D record is recorded before a slow data record, the overflow information cannot be resolved.

The absolute time of an image is not recorded with the image.
The image is in a record for which there are two times.
The time the TAS clock was started to collect a record is recorded.
The end time recorded is when the transfer to memory of the 2-D record buffer is complete.
The transfer time is estimated at 73 milliseconds.
The first few slices of a record are not well defined.
Possible events that can occur after these first few slices:
1. the probe sees no image and counts the number of omitted non-shadowed slices
2. the probe sees no image and therefore writes a time slice
3. the probe sees image and records the image slices

Approach

The approach taken to determine the absolute time of an image is likely correct to within milliseconds.

The slices up to and including the first time slice of a record are ignored by the image statistic processing. The start time of a record is assumed to represent the absolute time of the first reference slice. The absolute time of an image will be the time of the first image slice of the image.

  
  t0 - start time of record, time of first reference slice
  ti - time of beginning of ith image

  t1 = t0 + (elapsed time for all the omitted slices 
                      before first image slice)

  t2 = t1 + (elapsed time for first image slices 
             + elapsed time for time slice 
             + elapsed time for reference slice 
             + time slice )

  t(i+1) = ti + (elapsed time for ti images
                 + elapsed time for time slice 
                 + elapsed time for reference slice 
                 + time slice)

  t(i+1) = ti + (N * deltat)

  N = number of slices in previous image 
      + 1-time slice 
      + 1-reference slice 
      + time slice

  deltat = (element width) / tas
   [s]        [m]           [m/s]

  element width =  25 10e-6 [m] for 2-D C probe
                  200 10e-6 [m] for 2-D P probe

  tas = true air speed obtained from the last slow record before
        the 2-D record was recorded

Exceptions

A time slice with the value zero is called a zero time slice. The value used for this time slice is 0.
Time slices which represent a time greater than the elapsed time for the 2-D record (less the transfer time of 73 ms) are replaced with a zero time slice.
The sum of the times for all the images may exceed the elapsed time for the record. What is to be done in this case is not defined.

Processing of 2D imagery data

Our processing software has been written in Whitesmith's PASCAL, and C and uses some of the algorithms from the University of Wyoming as a guide. The programming language PASCAL utilizes the concept of structured programming. C is a general-purpose programming language which features economy of expression, modern control flow and data structures, and a powerful set of operators.

If you remove artifacts, what type of artifacts do you eliminate and what technique do you use?

During 2D data processing, artifacts were removed according to the rejection criteria of Cooper (1978). All remaining images were accepted as real particles. Adjustments to the concentrations were made to include depth-of-field and sample volume versus particle size effects, as discussed in Heymsfield and Baumgardner (1985).


The following statistics are collected for each image:

        time - absolute time of image
      length - number of image slices (shadowed and non-shadowed)
       shade - number of image slices with shadow elements
     noshade - number of image slices with no shadow elements
           x - maximum length of shadow along path
           y - maximum width of shadow across path
          my - length of longest shadow across path within a slice
         nmy - slice postition of my
        area - number of shades elements
       iedge - number of elements of image which touch left side
       jedge - number of elements of image which touch right side
       mleft - maximum number of contiguous slices parallel to left side
               but not at the edge
      mright - maximum number of contiguous slices parallel to right side
               but not at the edge
   spherical - true if image monotonically increases then decreases in width
         gap - number of gaps (non-shadow slices) bounded by shadow slices.
       depol - depolarization flag
         tas - True airspeed measured at the probe location which controls the
               strobe sampling rate of the probe.

Example Picture

                       1------ y direction --------->32

                       11111111111111111111111111111111
                       11111111111111111111111111111111
                       010101010 time slice 10101010101
                       0000000 reference slice 00000000
         1             11111111111000000001111111111111  !  first image slice
         2             11111111000000000000000111111111  !
         3             11111111000000000000000001111111  !
         4         1   11111100000000000000000000011111  !
         5         2   11111100000000000000000000000000  !  nmy = 5
         6         3   11111100000000000000000000000000  !  <--- jedge = 3
         7         4   1111110000000 area = # 0's 00000  !  my = 26
         8         5   11111100000000000000000000000011  x
         9         6   11111100000000000000000000000111  !
         10        7   11111100000000000000000000000111  !  
         11        8   11111100000000000000000000000111  !  
         12 mleft= 9   11111100000000000000000000000111  !  <-- mright = 4
         13 iedge = 0  11111111100000000000000000011111  !
         14            11111111111000000000000011111111  !
         15            11111111111110000000011111111111  V  shade = 15
         16            11111111111111111111111111111111
         17            111111<------- y -------------->
         18            11111111111111111111111111111111
         19            11111111111111111111111111111111
         20            11111111111111111111111111111111      nodshade = 6
         21 length=21  11111111111111111111111111111111   last image slice
                       010101010 time slice 10101010101
                       0000000 reference slice 00000000
                       11111111111111000000000111111111
                       11111111110000000000000000001111
                       ...

The calculated statistics characterize each image. The user can specify which type of image is to be rejected based on test criteria applied to each image. All images with zero time bars are rejected. The following additional test criteria are used:

SPHERE

Only images which satisfy the sphercial statistic are tested further. Images with areas smaller than a user specified number of shadowed elements are not tested for sphericity, since it is difficult to tell for small images whether they are round. The area of a spherical image must also exceed .75 xy. The x and y of the image must also be nearly equal. Some image elongation is allowed: the x can be up to 5 times y if the image is at the edge and the x can be up to 2 times y if the image is fully within view.

STREAKER

Streakers are images which are elongated in the direction of flight: their likely origin is water shedding from the probe tips. Images with y>24 are not tested. There are two tests: 1) Images with a (x > 6y) are considered to be water streakers. 2) If an image is completely within the field of view, smaller images are considered to be streakers if (x > 3y)

SPLASH

If the distance between two separate images is less than a user specified amount along the direction of flight, the second image is considered to be a result of a splash.

The default value for the C-probe is UGAPC = 2.5 cm. The default value for the P-probe is UGAPP = 0.12 cm. With these defaults, real ice concentrations greater than about 800 per litre are not accurately detected.

ZERO_AREA

Occasionally, something (small particle or electronic noise) will trigger the probe and no elements are shadowed. The result is zero-area images.

SMALL_AREA_RATIO

If the total shadowed area of an image is less than (0.4xy), the image is classified as several small parallel splash images. This test attempts to detect images with gaps parallet to the direction of flight.

GAP

Images which contain nonshadowed slices bounded by slices with shadow are considered to have a gap.

How is the particle size determined?

We define particle size as x+ 1 ( the length of the image along the direction of flight plus the first shadowed slice which starts image collection but is not recorded).

How is the particle habit determined?

We do not determine particle habit.