FIFE BASELINE AMS DATA DOCUMENTATION: FIFE Site averaged data files: 1987-1989. as data sets: Site Averaged AMS Data: 1987 (Betts) Site Averaged AMS Data: 1988 (Betts) Site Averaged AMS Data: 1989 (Betts) Site Averaged AMS Data: 1987-89 (Betts) ffo_ams.doc Date: June 10, 1996 Revised release KEY REFERENCE: Betts and Ball, 1997: FIFE surface climate and site-average data set 1987-1989, J. Atmos. Sci. (3rd FIFE special issue) In press. REVISION HISTORY: Preliminary release: March 14 1996 First revision: June 10, 1996 Major changes: 1) Redid (more editing of bad data) 30-min rainfall average for all 3 years 1987,88, and 89. Included std dev and count for all 3 years 2) Redid IR Down average count and std deviation for 1987 only, after finding more data ****We have no plans to revisit this data; but let us know of errors *** See also: FIFE Site Averaged Surface Flux Data as: Site Averaged Flux Data: 1987 (Betts) ffoflx87.dat Site Averaged Flux Data: 1988 (Betts) ffoflx88.dat Site Averaged Flux Data: 1989 (Betts) ffoflx89.dat Site Averaged Flux Data: 1987-89 (Betts) ffo_flx.dat FIFE Site Averaged Gravimetric Soil Moisture Data: 1987-1989 (Betts) as: Site Avg. Grav. Soil Mstr: 1987 (Betts) ffogrv87.dat Site Avg. Grav. Soil Mstr: 1988 (Betts) ffogrv88.dat Site Avg. Grav. Soil Mstr: 1989 (Betts) ffogrv89.dat Site Avg. Grav. Soil Mstr: 87-89 (Betts) ffo_grv.dat FIFE Site Averaged Neutron Probe Soil Moisture Data: 1987-1989 (Betts) as: Site Avg. Neut. Soil Mstr: 1987 (Betts) ffoneu87.dat Site Avg. Neut. Soil Mstr: 1988 (Betts) ffoneu88.dat Site Avg. Neut. Soil Mstr: 1989 (Betts) ffoneu89.dat Site Avg. Neut. Soil Mstr: 87-89 (Betts) ffo_neu.dat Source: Alan K. Betts and John H. Ball Atmospheric Research, RR#3, Box 3125, Pittsford VT 05763. akbetts@aol.com 802 483-2087 (voice) 802 483-6167 (FAX) JUNE 10 1996 release File names size (bytes) approximate Zipped size ffoams87.dat 2493717 780784 ffoams88.dat 3217602 1104980 ffoams89.dat 2796498 1076429 ffo_ams.dat 9414162 2895242 1.0 Origin of Data This data set is a site averaged product of the Portable Automatic Meteorological Station (AMS) data acquired during the 1987-1989 FIFE experiment. The raw data have been extensively cleaned and editted before the site average was generated. The center of the FIFE 15x15 km site is close to 39.05þN, 96.53þW. The 3 years of data were acquired and processed separately over a period of several years; and as a result there are a few differences in the 87 product and in the underlying methodology. The 1987 data was downloaded piecemeal (by 2400 baud modem!) from the FIFE information center, editted and merged. Subsequently the 1988 data was retrieved from the FIFE CD ROM Vol 1; and the 1989 data was retrieved directly from the Oak Ridge DAAC. As our computers have improved by more than an order of magnitude in speed and disc space during the processing period, the 1989 data received a more uniform processing than the 1987. All the originating data however should correspond to that in the group 8, "Surface Meteorological Measurements (SUR_MET)" or FIFE CD-ROM Volume 1, dated May 22, 1994, FIS Data Base Table "AMS_DATA_8x". 2.0 Description of data sets This data set is a time series of 30-min average variables for the periods: ffoams87.dat May 1, 1987 - Dec 31, 1987 (50 fields; 14 variables (mean , SD, and count; plus 4 cloud fields); date (2), UTC, year sequence #.) ffoams88.dat Jan 1, 1988 - Dec 31, 1988 (46 fields; 14 variables (mean ,SD, and count); date (2), UTC, year sequence #.) ffoams89.dat Jan 1, 1989 - Nov 10, 1989 ffo_ams.dat May 1, 1987 - Nov 10, 1989 (composite of all 3 years) 2.1 Time index The time interval of the year is broken down into 30-min intervals. There is one variable sample at each 30 minute time interval so in a LEAP (NOTE 1988) there are 366 x 48 = 17658 time intervals. This field in the data base in named "YSq#" (Yearly Sequence #) and makes a convenient linear time scale; in addition to day and UTC. 2.2 Stations Selected We believe we have included all the AMS statiion data. In 1988 some sites were moved during the year (see CD ROM). 2.3 List of Variables Variable Name Long Name (FIS Name) SAS Type* Variable Description/Units 1 Ysq_no 8 "Time interval of the year broken down into 30-minute intervals" 2 date OBS_DATE 8 "Observation date (mm/dd/yyyy)" 3 Jul_Date N/A 8 "Julian date (day of year 1-366)" 4 UTC N/A 8 "Decimal time" 5 P ATMOSPHERIC_PRESSURE 8 "Average atmospheric pressure (mbar)" 6 P_k 8 "Number of sites included in atmospheric pressure average" 7 P_s 8 "Atmospheric pressure standard deviation (mbar)" 8 T DRY_BULB_TEMPERATURE 8 "Average dry bulb temperature (degrees C)" 9 T_k 8 "Number of sites included in average dry bulb temperature (T)" 10 T_s 8 "Dry bulb temperature standard deviation" 11 Tw WET_BULB_TEMPERATURE 8 "Average wet bulb temperature (degrees C)" 12 Tw_k 8 "Number of sites included in average wet bulb temperature (T)" 13 Tw_s 8 "Wet bulb temperature standard deviation" 14 Rain30 RAINFALL_RATE(**) 8 "Average rainfall rate for 30 minutes (Rain30)" 15 Rain30_k 8 "Number of sites included in average 30-minute rainfall rate (Rain30)" 16 Rain30_s 8 "Standard deviation of average 30-minute rainfall rate (Rain30)" 17 U U_COMPNT_WIND_VELOC 8 "Average U component of wind velocity (m/s)" 18 U_k 8 "Number of sites included in average U component wind velocity (U)" 19 U_s 8 "Standard deviation of U component wind velocity (U)" 20 V V_COMPNT_WIND_VELOC 8 "Average V component of wind velocity (m/s)" 21 V_k 8 "Number of sites included in average V component wind velocity (V)" 22 V_s 8 "Standard deviation of V component wind velocity (V)" 23 Tsurf SURF_TEMP 8 "Average surface temperature (degrees C)" 24 Tsurf_k 8 "Number of sites included in average surface temperature (Tsurf) " 25 Tsurf_s 8 "Standard deviation of surface temperature (Tsurf)" 26 Tsoil10 SOIL_TEMP_10CM 8 "Average soil temperature at 10cm (degrees C)" 27 Tsoil10k 8 "Number of sites included in average soil temperature at 10cm (Tsoil10)" 28 Tsoil10s 8 "Standard deviation of soil temperature at 10cm (Tsoil10)" 29 Tsoil50 SOIL_TEMP_50CM 8 "Average soil temperature at 50cm (degrees C)" 30 Tsoil50k 8 "Number of sites included in average soil temperature at 50cm (Tsoil50)" 31 Tsoil50s 8 "Standard deviation of soil temperature at 50cm (Tsoil50)" 32 SolDn TOTAL_INCIDENT_RADTN 8 "Average total incident radiation (W/m2)" 33 SolDn_k 8 "Number of sites included in total incident radiation average (SolDn)" 34 SolDn_s 8 "Standard deviation of average total incident radiation (SolDn)" 35 SolRef SHORTWAVE_SOLAR_REFL 8 "Average shortwave solar reflectance (W/m2)" 36 SolRef_k 8 "Number of sites included in average shortwave solar reflectance (SolRef)" 37 SolRef_s 8 "Standard deviation of average shortwave solar reflectance (SolRef)" 38 Rnet NET_RADTN 8 "Average net radiation (W/m2)" 39 Rnet_k 8 "Number of sites included in average net radiation (Rnet)" 40 Rnet_s 8 "Standard deviation of average net radiation (Rnet)" 41 IRDn INCIDENT_LONGWAVE_RADTN 8 "Average incident longwave radiation (W/m2)" 42 IRDn_k 8 "Number of sites included in average incident longwave radiation (IRDn)" 43 IRDn_s 8 "Standard deviation of average incident longwave radiation (IRDn)" 44 Q N/A (calculated) 8 "Average mixing ratio (g Kg-1) (see formula below)" 45 Q_k 8 "Number of sites included in average mixing ratio (Q)" 46 Q_s 8 "Standard deviation of average mixing ratio (Q)" 47 ToCldoct 8 "Average total cloud cover (octas)" 48 HiCldoct 8 "Average high cloud cover (octas)" 49 TimePhot 8 "Time of day sky photographs were taken" 50 MF_cld10 8 "Reported cloud cover from Marshall Field (tenths)" 51 obs_time 8 "Observation time (hhmm)" 52 year 8 "Year" [ffo_ams.dat file only] * SAS Type: 8 = Numeric Variable SAS Type: $ n = Character Variable of Length n (**) The FIS rainfall rate is in mm/5mins, and was multiplied by 6 to give mm in 30mins. In 1987 there are also four cloud variables; ToCldoct cloud sky cover in eights. HiCldoct cloud sky cover in eights. TimePhot the time of day of sky photograph. Mf_cld10 reported cloud cover from Marshall Field, in tenths. The data definition file (*.def) accompanying each data set provides the exact list of variables included in the data set. The data files are comma delimited ASCII files and contain a header line listing the data fields included in the file. The comprehensive list of variables above is provided for general information only. 3. Data Quality Most variables are continuous for all but a few short periods. There is no Radiation data for an extended period of one month from April 11 through May 10, 1988; because the instruments were being calibrated. However, other than this period, Rnet is available about 99% of the time. IRDn is missing for three extended time intervals and may be of questionable value for much of the time it is available. 3.1 Data Filters and Cleaning. A set of filters was applied to reject bad data points and help to identify lines of bad data (such as produced by electrical noise). Table 2 summarizes these filters. Further data editing was effected by scanning the data and nulling out obvious bad data manually. Some bad values will, however, have slipped through, these can often be identified by a large value of the standard deviation. Table 2. Data filters Variable Max/Min (1988) Filter Range P 999.94/938.29 930 to 1000 T 37.94/-23.17 -30 to 45 Tw 36.01/-22.43 -30 to 40 & (Tw-T)>0.05 Rain30 7.75/0 U 31.20/-27.56 V 22.29/-17.86 Tsurf 54.76/-39.34 -50 to 60 Tsoil10 39/-2.85 -10 to 50 Tsoil50 26/1.38 -5 to 40 SolDn 1034.71/-4.88 -5 to 1200 SolRef 543,45/-4.93 -5 ot 600 Rnet 8.39.52/-87.36 -98 to 1000 IRDn 498.52/143.60 100 to 600 Q N/A (calculated) Each variable (for all sites), together with the site-average and standard deviation were then examined graphically and manually (and iteratively) edited to eliminate bad data; and generate a cleaned-up site average for each variable. Some such as temperature and wet-bulb temperature were examined in pairs. Some bad data has escaped this process; often this is indicated by a solitary extreme standard deviation. 3.2 Interpolation We did not interpolate variables at a site before averaging; but we have interpolated most site-average variables to fill gaps (where these are small <6 hours) to reduce the burden on modellers. Gaps of more than a day have generally not been interpolated; although we have filled a few missing days temperature profiles with a simple and obvious sawtooth patterns (in 1988 and 1989). Some missing Net radiation values at night have been set to -30 Wm-2. In 1988 and 1989 SolDn and SolRef have been set to 0 (zero) between sun set and sun rise. INTERPOLATED DATA IS EASILY IDENTIFIED. If the count on a variable is zero the data has been interpolated. 4. Original 1987 AMSDOC.txt This is the documentation from our original May 26-Oct 16 1987 site average data set from the FIFE CD ROM. We include it for its remarks on the cloud data fields; and formula used for mixing ratio q, as we have not reprocessed this period, except for rain and IRDN. FIFE-87 compacted surface data set. (Betts and Ball, 1992) Rev. Date: December 11, 1992 This data set contains FIFE-site averages derived from (ten) AMS stations (data every 30 minutes), together with cloud data from two sources, from May 26 - Oct 16, 1987. The raw data for these files came from the FIS Data Base tables AMS_DATA_87 and CLOUD_CAMERA_DATA. The MF_Cld field was obtained independently from the hourly cloud observations of Marshall AAF, KS, approximately 12 km west of the FIFE site. Data Processing 1. Surface Portable Automated Meteorological Stations. These were PAM-II stations supplied by the National Center for Atmospheric Research (NCAR), with pressure (PS), temperature (T)and wet-bulb (TW) temperature sensors mounted at 2 m and an anemometer at 5.4 m, as well as ground temperature sensors at 10 cm (T1soil) and 50 cm (T2soil) below the surface. Each station measured surface skin temperature (TSfc) with a downward looking radiometer, net radiation (RNet) with a net radiometer, measured reflected solar radiation (SolRef) and recorded rainrate and accumulation from a tipping bucket raingauge. A subset of stations (2 before August, 10 1987 and then 4) measured incoming solar radiation (SolDn), and longwave radiation (LWDN) (2 before October, 1987, and then 4). The archive data uses NCAR's calibrations of the instruments. Data Filters A set of filters was applied to reject bad data points and identify lines of bad data (such as produced by electrical noise). These are shown in Table 87-1. Table 87-1 Variable May 26-Jun 24 Jun 25-Aug21 Aug 22-Oct 16 SolDn -5 to 1200 -5 to 1200 -5 to 1200 SolRef -5 to 250 -5 to 250 -5 to 250 RNet -98 to 1000 -98 to 1000 -98 to 800 TSfc 0 to 55 10 to 55 -10 to 35 TSoil10 9 to 30 19 to 30 9 to 21 TSoil50 11 to 30 19 to 30 11 to 19 T and TW were filtered to eliminate any data where T was less than TW. The data was scanned for low values of surface pressure. For both the radiation and thermodynamic data, further data editing was done by scanning the data and nulling out obvious bad data manually. Some bad values will however have slipped through. The mixing ratio (Q) was calculated from TW and PS (surface pressure) as follows: QW = 622/(0.1631*PS*exp(-17.67*TW/(TW+243.5))-1) Q = QW-(1006/2501)*(T-TW) Data Averaging All the station data that passed this editing were averaged for each date and time to give a site average, a standard deviation, and a count of stations in the average. The center of the FIFE area is close to 39.05þN, 96.53þW. The standard deviation is between sites, which are located at different altitudes, on different types and slopes of terrain, different soils etc. Isolated high standard deviations probably indicate an erroneous value that has not been filtered, although in some variables such as SolDn or RNet, they may be caused by more extensive cloud over a few sites. We have not included a standard deviation for LWDN, since there are only 2 sites with this measurement for much of the time. 2. Cloud data Two fields relating to daytime cloud cover for total cloud and high cloud (in octas) are from the archive of digitized photographic data (from Ann Henderson-Sellers). They are not 30 min. averages. The instantaneous digitized mean cloud cover from each single frame was simply assigned to the 30 min. time block in which it was taken. Since photos were typically every 40 min., every fourth time had no photo. A further field indicates the exact time of a photo. The MF_Cld field was obtained independently from the hourly cloud observations of Marshall AAF, KS, approximately 12 km west of the FIFE site. This data set is in tenths cloud cover, estimated by observer. Acknowledgment. You may use these files freely. Please send me (Alan Betts) an e-mail if you do, so I can notify you of any updates. Please notify me of any errors you find. This data processing was supported by NASA under Contract NAS5-32356 to Alan Betts, and an acknowledgment of the years of work that went into the production of these files would be appreciated in publications based substantially on them. The FIFE CD-ROM can be cited as Strebel et al (1994). 5.References. Betts, A.K., J.H. Ball, and A.C.M. Beljaars, 1993: Comparison between the land surface response of the European Centre model and the FIFE-1987 data. Q.J.R.M.S., 119, 975-1001. Betts, A.K. and J.H. Ball, 1994: Budget analysis of FIFE-1987 sonde data. J.G.R., 99, 3655-3666. Betts, A.K. and J.H. Ball, 1995: The FIFE surface diurnal cycle climate. J.G.R. 100, 25679-25693. Betts A. K. And J. H. Ball, 1997: FIFE surface climate and site-average dataset: 1987-1989. (FIFE special issue 3 J.Atmos. Sci.i in press) Strebel, D. E., D. R. Landis, K. F. Huemmrich, and B. W. Meeson, 1994: Collected data of the First ISLSCP Field Experiment, in Surface Observations and non-image data sets., Vol 1, CD-ROM, NASA GSFC, Greenbelt, MD 20771