3.1 Data Logger
NRG data loggers (9300CL) were used with 12 input channels. Data were sampled at 1 Hz and the following were stored: 15 minute averages and standard deviation plus daily maximum/minimum and time of occurrence. Sensors were Maximum #40 three cup anemometer, Maximum #200P wind vane, the Li-Cor 200S solar pyranometer, and the NRG 100 or 110S temp probe. Wire runs were two or three conductor cables strapped to the tower at regular intervals to minimize wind flapping. NRG supplied rubber boots were used to protect the connections at the base of each sensor from exposure to the elements. These boots were usually successful in controlling wire/contact deterioration, but at the same time if moisture were to come into the upper edge of the boot it was difficult to drain from the bottom and could cover the contacts and result in bad data. Raymondville had this happen and all wind vanes were replace at one visit due to the water ruining the electrical connections.
Loggers were checked out before being placed in the field by testing the inputs, phone programming, and proper initialization of data cards inserted in the logger. Sensors were all tested for proper signal output and wiring runs checked for continuity before and after installation. Wire connections were all similar so a process of starting /stopping each sensor in turn and recording the exact channel it was connected to by ground personnel was done to ensure that the channel number matched the expected sensor level input. See Appendix 1 for specific information on levels and the boom orientations.
Data are stored every 15 minutes on a removable data card. These data are transmitted to the receive PC at AEI on a weekly call-in schedule. The BASE program monitors the phone lines, answers the call as it comes in, determines which site is calling and what is the status of the data card and call-in schedule (card unread, 1st call of 6 tries, or card partially read, 4th call of 6, etc.). The BASE program then assigns a DOS filename to the data and stores it to the hard disk as it comes over the phone line. If any error occurs during transmission the logger will stop the call, and retry at the point where the data transmission broke off. The BASE program will sense that the call is a continuation and assign a file name with a different extension to show that the data is form the same period but a different portion of the data from the storage card. The screen display of the call in is printed.
The print out of the screen display the data is examined for bad sensor values (example, 5 wind speeds with high values and 1 with zero speed), proper voltage for system battery, etc. If any unusual or obviously bad values were noted, the printout would be marked with a highlighter, and the information recorded on an marker board in the data room. The printout was then filed in the weekly record folder for the site. When sensor or logger failure were noted, generally the system would be repaired within a week. Two anemometers at the upper levels helped in maintaining a high percentage for data recovery.
Raw data files on the receive computer were transferred to appropriate sub directories on a weekly basis. Each site had its own directory with sub directories based on the year of collection. Raw data files were also stored to floppy disks and transferred to an archive computer. This archive computer has a directory structure that duplicates the data receiving computer, as well as a magneto-optical disks that stored all the raw data files in the same directory system.
The raw data files were imported into the NRG Microsite program. Microsite takes the raw data files, applies a scale and offset to create a file with engineering units. This file is averaged by each hour of the day to create the hourly average files. A separate file contains the minimum/maximum values for each day (1 s values).
The 15 minute data in graphical format by week were inspected visually for obvious problems and typical trends. Suspect data were marked and were not used in the Microsite analysis program. The second check used routines within Microsite, where data were checked for range errors, trend errors, and relational errors. If they did not meet the criteria selected, they were marked. These criteria were developed using the U-WRAP guidelines  and modified per region. All marked data were then reviewed graphically to determine if the data remain eliminated, extended on either end, or unmarked for the data analysis. Marked data not used is considered flagged and that is noted on the month summary report (Percent of flagged). Percent data used on the month summary report is an indication of data logger operation.
After the quality control procedures, the monthly reports (Appendix 2) were produced using Microsite. Monthly reports were printed and stored by site. Monthly averages for windspeed, solar insolation, calculated wind power, percent of flagging (marking due to error or data exclusion), total data received, and wind shears were recorded in spreadsheets. (Appendix 3).
Data are backed up to CD-ROM quarterly and stored off-site in the Computer Center. A copy of Microsite and the hourly average data are updated monthly on another separate computer in AEI as quality assurance is finished for all sites.
Data recovery was over 95% for most sites. The sites with the most sensor problems, flagged data => 10%, were Guadalupe Pass and Quay Co, NM. At northern sites, a small part of the flagged data was due to icing. When a data logger failed, generally, there was only a loss of 1 week of data because data were stored on a data card, which could be retrieved, and there were regular weekly call-ins. The Amarillo Site was down from July until mid-December, 1998, because the tower buckled in the middle due to guy wire failure. Even though there was an electric fence around the site, cows got into the area and rubbed on the guy wires. Electric fence was replaced with barb wire fence around each guy anchor.