DESCRIPTION psd is a background noise acceleration Power Spectral Den- sity (PSD) estimation algorithm designed specifically to extract all necessary information and data for calculating the PSD estimates from a SEED data volume. It is an export- able version of the acceleration PSD algorithm that has been in routine use at the Berkeley Seismological Laboratory (BSL) for approximately the past decade. The psd algorithm is a script that calls a series of three algorithms that extract the requisite station coordinate, sensor transfer function, and data files. The essential components of the psd script are: rdseed -s -f $sdv > sdv_temp which uses "rdseed" to retrieve all station and channel header information from the SEED data volume "$sdv" and write it to a temporary file "sdv_temp"; grep B050F sdv_temp | sdv_coord > instr.sdv.coord which does a pattern grep on the station coordinate entries in the sdv_temp file and uses the supplied "sdv_coord" algo- rithm to generate "instr.sdv.coord" containing the station coordinate information; grep B05 sdv_temp | sdv_resp > instr.sdv.resp which does a pattern grep on the station sensor entries in the sdv_temp file and uses the supplied "sdv_resp" algorithm to generate "instr.sdv.resp" containing the sensor transfer functions and related information; rdseed -d -o 1 -f $sdv which uses "rdseed" to dump the data records in SAC binary format; and finally a foreach loop: foreach sbdf (`ls | grep SAC`) sdv_psdest $sbdf set x = $status if ( $x == 0 ) then \rm $sbdf echo File $sbdf processed. else set np = `sachdr $sbdf | grep npts | awk '{ print $3 }'` echo File $sbdf contained only $np samples. endif end to process each of the SAC binary data files, that contain at least 65536 data samples without any time gaps, and cal- culate the acceleration PSD estimate using the "sdv_psdest" algorithm. Note that the SAC algorithm sachdr, supplied in the PASSCAL Software Suite, is called in psd to determine the number of samples in the file. The sdv_psdest algorithm was intentionally designed to operated in an automated fashion without any optional com- mand line arguments. This minimizes the possibility of inappropriate use of the algorithm that could produce biased PSD estimates. The original in-house version of the sdv_psdest algorithm used at the Berkeley Seismological Laboratory was written in Fortran 77. The f2c translator was used to convert the Fortran 77 source code to C source files. The Fortran 77 source code was cleaned up and neces- sary subrouutines, not included in the f2c libraries, were added to expedite the conversion process. The sdv_coord and sdv_resp algorithms were converted in a similar fashion. The only f2c option used in the conversion was "-c" which causes the original Fortran 77 source code to be embedded as comments in the C source files. The sdv_psdest algorithm uses a statistical approach to robustly estimate the background noise PSD. The input time series is parsed into eight (possibly overlapping) time series and each of the resulting time series are appropri- ately windowed prior to calculating their PSD estimates. For short time series, less than 1.5 hours in length, the time series are detrended and sine tapered while for longer time series the dominant semi-diurnal gravitational tide signal is also removed to avoid biasing the long-period PSD estimates. A caveat: since the transfer function represen- tation in the SEED data volumes for typical broadband seismometers does not include the static component of the response, the background noise PSD estimates for periods longer than an approximately an hour will be biased high and hence they will be unreliable.