|
<br />APRIL 1984
<br />
<br />SMITH ET AL.
<br />
<br />507
<br />
<br />TABLE 5. Sources of data for HIPLEX-I primary
<br />response variables.
<br />
<br />CIC2, CIC5:
<br />
<br />Particle Measuring Systems (PMS) 2D-C
<br />probe on cloud physics aircraft
<br />(depolarization channel)
<br />Decelerator slides exposed by cloud
<br />physics aircraft during pass 5 min after
<br />treatment time
<br />PMS 2D-P probe on cloud physics
<br />aircraft
<br />Johnson-Williams cloud liquid water
<br />concentration sensor on cloud physics
<br />aircraft
<br />SWR-75 5-cm radar (10 beamwidth)
<br />
<br />CCR5:
<br />
<br />PIC8, MVD8, TFPI,
<br />TIPA, AER:
<br />AWC8:
<br />
<br />TFE, T1PR, RERC:
<br />
<br />these procedures completely in the experimental de-
<br />sign, but in several instances it was found that the .
<br />specifications still left some ambiguity. These. ambi-
<br />guities were resolved in several discussion meetings
<br />involving HIPLEX-I participants, and the results are
<br />reflected in the Appendix. Such a process can be tol-
<br />erated in an exploratory experiment, but it surely rein-
<br />forces the need to perform trial TUns of all aspects of
<br />an experiment, including analysis of the data, prior to
<br />launching any confirmatory experiment.
<br />Procedures for calculating the radar response vari-
<br />ables were similarly described in the design document.
<br />An important task in determining the radar variables
<br />was to identify which echoes were considered to be
<br />part of each test case. This "boxing" of the echoes was
<br />done by an analyst who was not allowed to visit the
<br />field site during the actual experiment and had no
<br />knowledge of the treatment decisions for each case.
<br />(In fact, she was furnished with an excess of data with
<br />coded dates, so that she did not even know which data
<br />concerned actual HIPLEX-I test cases.) She was sup-
<br />plied with the radar data, the aircraft flight tracks, the
<br />reported positions and times of treatment, and all other
<br />information available up to the time when the decision
<br />envelope was opened. That information plus knowl-
<br />edge about the expected dispersion of the seeding plume
<br />was used to delineate which echoes were considered
<br />part of the test case as distinct from other neighboring
<br />echoes. The analyst also had to screen out echoes due
<br />to "skin paint" reflections from the project aircraft.
<br />The response variables were then calculated by
<br />computer programs, except for CCR5 which involved
<br />manual counting of crystals on the. decelerator slides.
<br />In many cases, at least two different investigators made
<br />independent determinations of the variables (CIC2,
<br />CIC5, PIC8, MVD8, A WC8, TFPI, TFE, TIPA, TIPR,
<br />RERC). For many of the variables, additional manual
<br />checks were made of the computed values for a. few
<br />selected cases in order to verify the proper operation
<br />of the computer programs (CIC2, CIC5, PIC8, A WC8,
<br />TFPI, TFE, TIP A, TIPR, RERC, AER). This cross
<br />checking was found valuable because several instances
<br />of disagreement or outright errors were uncovered and
<br />
<br />I,
<br />
<br />the procedures corrected. The checking revealed con-
<br />ceptual, as well as computational, errors in the deter-
<br />minations of the response variables, and pointed out
<br />some of the previously mentioned ambiguities in the
<br />original design. Thus, considerable effort was devoted
<br />to insuring the accurate determination of the HIPLEX-
<br />I response variables.
<br />Tables 6 and 7 present the resulting values of the
<br />response variables. Table 7 contains two secondary
<br />variables which were not specified in the design doc-
<br />ument:
<br />
<br />RERB: Radar-estimated rainfall determined at cloud
<br />base level.
<br />RERL: Radar-estimated rainfall determined for the
<br />lowest antenna tilt.
<br />
<br />These variables were introduced because the restriction
<br />on the radar data usable for estimating precipitation
<br />amounts to the + I OOC levelled to an excessive number
<br />of default or zero values in the defined response vari-
<br />ables TIPR and RERe. This restriction was introduced
<br />in the hope of assuring predominantly liquid particles
<br />in the radar beam, but the 2D-P image data from the
<br />cloud physics aircraft showed some occurrences of
<br />millimeter-sized ice particles even at + Woe. In ret-
<br />rospect, it might have been better to use the best avail-
<br />able radar estimates as response variables, even though
<br />they involved some uncertainties related to the particle
<br />composition.
<br />
<br />8. Conclusions
<br />
<br />Table 6 summarizes the primary data set for the
<br />statistical evaluation of HIP LEX-I. Companion papers
<br />present the results of the planned statistical evaluation
<br />(Mielke et al., 1984) and the physical evaluation of
<br />the experiment (Cooper and Lawson, 1984). Further
<br />exploratory analyses will probably continue for some
<br />time. Computer simulations of both seeded and non-
<br />seeded versions of each HIPLEX-I test case are being
<br />carried out using a two-dimensional, slab-symmetric
<br />numerical cloud model (Hsie et aI., 1980; Kopp et aI.,
<br />1983), and results of these studies will appear elsewhere.
<br />We believe that HIPLEX-I represents a significant
<br />advance in our understanding of how to design and
<br />carry out a randomized seeding experiment on con-
<br />vective clouds. Among the important concepts that
<br />were employed, some for the first time in weather
<br />modification, in designing and carrying out HIPLEX-
<br />I were the following:
<br />
<br />I) The investigation was carried out in successive
<br />stages, with: a) an initial exploratory stage to develop
<br />an understanding of the clouds of the area, upon which
<br />the design of a physically meaningful seeding experi-
<br />ment could be based; b) an exploratory randomized
<br />seeding experiment (HIPLEX-I) on individual, small
<br />clouds; and c) intended follow-on experiments that
<br />would move up the scale to larger cloud systems, as
<br />
|