The Importance of Cross Validation In Climate Studies: Selected Case Studies of Radar/Disdrometer Reflectivity Comparisons
Wallace L. Clark, CIRES/Univ. of Colorado and NOAA/AL, Boulder, CO; and K. S. Gage, C. R. Williams, and A. Tokay
Although it is counterintuitive to many researchers because of the strong variability within precipitation processes, reflectivity calibration of vertically pointing precipitation radars by comparison with collocated disdrometers can be very precise, and is absolute within the accuracy of the disdrometer. At clutter free sites where clean echoes are available within two or three hundred meters of the surface, routine ongoing calibration and monitoring of a radar to well within a dBZ is feasible. This is possible because the process of taking high-frequency differences (we took differences once a minute) between simultaneously observed reflectivities acts like a high-pass filter, and essentially eliminates meteorological variation from all but the tiniest space and time scales. Especially during stratiform, low wind conditions the remaining fluctuations are mainly due to instrumental precision, sampling error because of the disdrometer's small sample volume, effects due to altitude separation, and differences in size of the sample volumes. These remaining fluctuations generally have Gaussian like characteristics, and can be effectively minimized by the simple process of time averaging.
Over the course of the NASA TRMM Ground Validation Campaigns, The NOAA Aeronomy Laboratory collected several large data sets of profiler-disdrometer reflectivity comparisons, utilizing a Joss-Waldvogel disdrometer (JWD) paired with a 915 and/or 2835 MHz precipitation profiler radar. Examples of successful calibrations and reflectivity comparisons have been presented elsewhere. Here we examine subtle features of these observations made visible by reflectivity binning and time averaging. This procedure allowed us to objectively characterize the degree of contamination by clear air echo at low reflectivities and the saturation thresholds for our higher-powered wind profiling radars, as well as verify the lack of saturation of our lower powered precipitation profilers. In addition, we were able to characterize a reflectivity dependent bias in the particular Joss-Waldvogel disdrometer (JWD) we used. With respect to this last observation, an autopsy of the JWD after the TRMM KWAJEX campaign revealed that its moving parts were highly weathered. A recent profiler/JWD comparison at the NASA Wallops Island disdrometer test bed with a newly reconditioned JWD showed no such reflectivity dependent bias. Without cross-comparison, biases such as those found above can be difficult to detect, let alone characterize. Consequently, these studies reinforce the precept that continual instrument cross-validation should be a key process in the formation of climatological databases.
Extended Abstract (1.7M)
Supplementary URL: http://ams.confex.com/ams/15AppClimate/presentationreview.cgi?username=93839&password=755678
Session 5, Remote Sensing II
Wednesday, 22 June 2005, 8:30 AM-9:45 AM, South Ballroom
Previous paper Next paper
Browse or search entire meeting
AMS Home Page