UND Microwave Radiometer Profiler Data README

Author: 
 Laura S. Leo
ORCID: 
 http://orcid.org/0000-0003-4103-6862
email: 
 lleo1@nd.edu
Institution: 
 University of Notre Dame 

1.  Dataset Overview
 - Preliminary Microwave Radiometer (MWR) data collected at two locations, as detailed below. 
 - Time period covered: 2017/04/28 - 2017/06/19
 - Physical locations (also included in each data file):
	(a)NEsiteMWR: east of the double ridge
					  PT-TM06/ETRS89: 34651.071, 6170.539,               
					  WGS84 Lon/Lat:7°43'44.25"W 39°43'23.27"N
					  Elevation:  259.831 m
				  
	(d)VALsiteOrange: orange grove site in the center of the Vale do Cobrao, in proximity of OU CLAMPS microwave radiometer.
					  PT-TM06/ETRS89: 33948.340, 5004.683                         
					  WGS84 Lon/Lat:7°44'13.97"W 39°42'45.57"N
					  Elevation: 296.140 m	

2.  Instrument  
The University of Notre Dame (ND) Microwave Radiometer Profiler is a Radiometrics MP-3000A:
• 21 K band (22-30 GHz) plus 14 V band (51-59 GHz) factory-calibrated channels
• Measures brightness temperatures in both water vapor and oxygen bands
• Measures cloud base temperature and height with optional IRT sensor
• Includes surface pressure, temperature, and humidity sensors at instrument height 
• Standard Neural Net retrievals provide atmospheric profiles of temperature (K), water vapor (gm-3), relative humidity (%), and liquid water (gm-3) up to 10 km temperature.

3. Data description:
All output files are named automatically using the following format:
yyyy-mm-dd_hh-mm-ss_xxx.csv,
where yyyy is the year when the file was started, mm is the month of the year, dd is the day of the month, hh is the hour of the day, mm is the minute of the hour, ss is the second of the minute, and xxx defines the output file type as follows:
•	xxx=lv0 level0 file
•	xxx=lv1 level1 file
•	xxx=lv2 level2 file
All output files contain a sequential record number in the first field, starting with the number 1. All output files contain a date/time stamp in the second field of all records that contain time-dependent data.
lv0 files contain raw, unprocessed data in engineering units. lv0 files contain 100 percent of the information needed to reprocess the raw data with alternative calibration information or algorithms.
lv1 files contain real-time brightness temperatures (TB) for each channel specified in the configuration file. 
Real-time level1 files are produced from contemporaneous level0 data and calibration information in the configuration file.
lv2 files contain records of real-time retrievals of temperature (K), water vapor (gm-3), relative humidity (%), and liquid water (gm-3) profiles. The retrievals are produced using the contemporaneous level1 data and the neural network files specified in the configuration file. 
Observations were collected at zenith and “off-zenith angles” (20º and 160º) and used for additional retrievals of temperature, vapor density, liquid density, and relative humidity profiles. These additional profiles are labelled N,S and A in the lv2 output file (north, south and average).

4.Record Type Conventions
All output files contain a record type number in the third field of all records. The record type number defines the header or data type in that record. Record types for each file type are grouped in blocks and numbered sequentially beginning with the number assigned to the header for that block. Record headers define all the fields in each block.
Data is logged sequentially in the order of the observations. For some types of analysis, it is more convenient to sort the data based on different parameters. Sorting a file by record type is often a useful first step to analysis. When a file is sorted by record type (third column in a spreadsheet, for example), the data automatically sorts into logical blocks with the appropriate header for each block appearing at the top of each block.

5. IMPORTANT NOTES:
5a. This dataset is preliminary and a future version of lv2 data will be released by July 2018. The future release will have a flag policy to identify suspicious/bad data associated to rain events and instrument start-up events[First 5 hours of data after power on are nominally NOT reliable].
5b.Therefore, the user of this current preliminary release should be cautious in using data associated with rain events (information included in this dataset) and instrument start-up events (daily files broken down in two or more sub-daily files indicate a shutdown and restart of the instrument due to several possible reasons, such as malfunctioning, troubleshooting, re-calibration or loss of power).
5c. LN2 calibrations were performed during May 7 and Jun 15. The current dataset results from raw data reprocessed using the LN2 calibration conducted on June 15. 5d. Due to a bug in the radiometric reprocessing software, the first and second profiles of each lv2 file should be discarded (these data will be indeed flagged as bad in the future dataset release). 
5d. The reliability of lv2 data strictly depends on the Neural networks (NNs), which are specific to the site where the radiometer is operated. Due to the lack of historical radiosonde data (needed to train the NNs) representative of Perdigao site in terms of both climatology and elevation, the NNs used for this dataset are not optimal. As results, Lv2 data presents biased values when compared to co-located radiosonde profiles collected during Perdigao experiment. The disagreemnt is generally larger at atitudes greater than 2-3 km above ground level and for off-zenith profiles.