This data set group consists of observations of surface meteorological parameters. These include short and long time-scale operational data sets, as well as supplemental data sets collected specifically for the experiment. The operational data sets contain historical climatological data (daily and monthly sets of precipitation and temperature data, four-and-one-half years of NOAA-obtained regional surface meteorological data, and daily rain data collected by two research groups within the FIFE study area). The supplemental FIFE data sets include data from portable automatic meteorological stations and quantified cloud coverage data from two stationary cameras.

NOAA regional surface reports were extracted from the NOAA Operational System and transmitted to FIS. The data set includes standard meteorological values (pressure, temperature, wind, etc.), weather variables (visibility, present weather, etc.), and cloud parameters (type, height, etc.). The data were for the time period of July 2, 1985 through October 23, 1988.

These regional surface data were extracted from the NOAA Operational System, which were archived at the National Climatic Data Center (NCDC). The data set includes standard meteorological variables (temperature, wind, pressure, etc.), weather variables (weather and visibility), and cloud parameters (type, height, amount). This data set covers the time period from October 1, 1988 through October 31, 1989.

This data set contains daily observations of air temperature and precipitation from Kansas State University. The time period covered is from January, 1900. Daily minimum and maximum temperatures are included.

This data set contains monthly summed daily precipitation data that were recorded by observers at Kansas State University. The time period covered is from January, 1858 to December, 1987.

This data set contains daily precipitation values collected by two separate research groups from a total of 42 rain gauge stations. The long-term ecological research (LTER) group had up to 12 recording stations within the LTER portion of the FIFE site. This data spans the years of 1982 through 1987, with fewer stations in the earlier years. The Princeton group collected 30-minute precipitation data (summed to form daily amounts) especially for FIFE from 30 rain gauges within the Kings Creek Watershed. These data are only for the period of May, 1987 through October, 1987.

Tipping bucket rain gauges were placed at 30 locations in the 11.7-square-kilometer Kings Creek Basin, which was in the northwest corner of the FIFE study area. From these rain gauges, one-minute rainfall readings were automatically recorded and later aggregated into 30-minute totals. The time period is from May 1987 through October 1987.

Ten portable automatic meteorological stations were installed within the FIFE site before IFC 1. Two more stations were added during IFC 4. All stations recorded air temperature, humidity, wind speed, soil temperature, reflected solar radiation, net radiation, surface temperature, and precipitation. Two of the original and the two additional stations were also equipped with extra radiation sensors (global, direct, diffuse, PAR, and long-wave). Data were collected from May of 1987 through October of 1989.

This is a derived data set containing averages of the observations from all of the AMS stations reporting in each 30-minute interval from May 1 through December 31, 1987. The raw data were taken from the AMS 30-minute averages data set. Additional data describing the clouds were extracted from the Cloud Camera data set and obtained independently from the hourly cloud observations at Marshall Field, KS, approximately 12 km west of the FIFE site. Although the time series is mostly continuous, there are some data gaps. (This data set is not in standard FIFE format. It is provided as received from the investigator and is in the "Grab Bag" section of the CD-ROM).

The amount of clouds was quantitatively measured by two stationary cameras. One camera was located in the northwest quadrant of the FIFE study area and the other was on the roof of the Kansas State University physics building. Data are for the years of 1987 and 1989 during the IFCs.

Several of the data items, in particular the cloud parameters, are encoded. Extensive use of the data documentation will be needed to understand the meaning of these variables. These data were also obtained directly from the NOAA surface stations, therefore the QA (quality assurance) which is usually performed by NCDC before archiving has not been done.

This data set is also highly encoded and FIS documentation is needed to understand many of the parameters. This data set has been QAed by NCDC.

Collected precipitation is measured by a volumetric cylinder (an 8-inch non-recording rain gauge). Temperature is measured by a maximum-minimum thermometer.

Collected precipitation was measured in the same method as stated above (Historic Daily Met Data).

This data set contains precipitation obtained from two sets of rain gauges. The LTER data was collected using a weighting-type gauge. Here a weighing mechanism converted the weight of the rainfall caught by the rain gauge into a curvilinear movement of a recording pen, which made an inked trace on a chart. The height of the pen marks on the chart were calibrated to rainfall amount. The Princeton University group used a tipping bucket rain gauge with a 0.2-mm tip connected to a data logger that recorded the date and time (to within one minute) of the tip. The one-minute rainfall readings were aggregated into 30-minute intervals and then summed for the day.

In addition to the description above (Rain Daily Data) for the Princeton University group, it has been estimated that due to hardware and software problems, generally only 20 out of the 30 locations would actually be actively recording at any one time.

The AMS stations recorded micrometeorological observations on a fine spatial scale at 5-minute intervals. The 5-minute data were averaged to 30-minute intervals to produce this data set.

This data set provides a nearly continuous record of average meteorological conditions over the FIFE study area thoughout the main experiment and monitoring period.

A standard (SLR) 35-mm camera with an attached fish-eye lens was used to record pictures of the sky. An image of the whole sky is optically projected onto a plane surface via a refraction system.

Many of these data sets contain the same parameters measured at approximately the same time and at different locations.

The NOAA and NCDC data sets may be used to provide a large-scale view of meteorological conditions (meso-scale), while the AMS data could be utilized to describe the smaller scale processes (micro-scale).

The AMS data set provides a nearly continuous high temporal and spatial resolution record, over a period of 2 1/2 years, of meteorological conditions over a prairie grassland. The AMS FIFE-Area Averages data set is one attempt to derive from these data an integrated and quality filtered data set that can be compared to or used to initialize atmospheric circulation models.

The historical data sets could be used as a baseline to compare the collected meteorological data sets (NOAA, NCDC, and AMS). The question of how "normal" the years of 1987 and 1989 were compared to the last 100 years could be explored.

The cloud camera data may be (and has been) compared directly with the NOAA surface data. Dr. Ann Henderson-Sellers has concluded from this analysis that there are several problems in the recording and/or observing of reported cloud amounts within the NOAA operational data set. Certain biases are noted and Dr. Henderson-Sellers questions the validity of the NOAA cloud parameter (amount) data. A description of this analysis is included in the scanned document section (SCAN_DOC directory) of this CD-ROM.

The 30-minute and daily rain data sets could be summed to hourly rain amounts, which could be directly compared or used in conjunction with the precipitation contained in the NOAA, NCDC, and AMS data sets. Likewise, the rain daily data could be compared with daily sums of precipitation values recorded at the NOAA and AMS meteorological stations. Regional precipitation differences and biases could be examined.

Several discrepancies were noted in the NOAA surface meteorological data by FIS staff and an investigator. The visibilities for Manhattan and Fort Riley are constant under all conditions. The cloud-cover estimates for the above two stations also show a statistical bias, according to an analysis by Dr. Henderson-Sellers. There may be other problems which have not been discovered. In addition, a number of the stations contained in this data set are third-level facilities dependent on volunteers or lacking equipment. Furthermore, as noted before, this data set did not pass through the usual National Climatic Data Centers' (NCDC) QA program. In summary, the quality of some of this data is suspect and no general quality assurance has been attempted.

The NCDC surface data were received from NCDC and therefore have been subjected to their Q/A program. No known errors are apparent in this set. FIS staff inserted 9999 numbers where "missing" values were encountered.

Within the historic daily met and monthly met data sets, no problems are known. They will suffer from changing equipment, techniques, and personnel, of course, considering the length of the data record.

As discussed, precipitation values in the rain daily data set were obtained by two different groups, the LTER and Princeton University. The LTER mechanical chart recorder would occasionally fail to chart an entire week, usually because the clock was not wound sufficiently. The LTER data points were routinely compared with that recorded by rain gauges located on the Kansas State University Campus. In general, the data from the LTER staff is of good quality, with a measurement error of + 1 mm. The rain gauge units used by Princeton University group were noted to be well constructed, but not totally suited for the Kansas environment. As discussed earlier, as many as ten of the 30 gauges might have problems at a given time. The loss of data in the logging unit was a problem in that the units did not withstand the Kansas temperature and humidity ranges well. In addition, rainfall amounts less than the size of the bucket (0.2 mm) may have been lost due to evaporation. Overall though the collected data were generally of high quality.

A number of problems and discrepancies were noted in working with the AMS data sets. The FIFE-installed instruments in the PAMS (i.e., the "extra" radiation sensors) did not have calibration factors applied when the preliminary data was sent. The values reported by NCAR were raw counts instead of physical units. This was corrected by FIS. Some automated quality control and checking was performed by NCAR, and quality flags are indicated for each variable in each reporting interval.

The AMS FIFE-area Averages data set has been cleaned by applying a set of filters to reject bad data points and identify lines of bad data (such as that produced by electrical noise). In addition to the automated filtering, for some dates an interactive process based on viewing graphical displays was used to exclude extreme points.

Within the cloud camera data, there was a problem reconstructing the 1989 date-and-time sequence. The times associated with August 11, 1989 are approximate (+/- 59 minutes). Otherwise, times associated with other days are assumed to be correct. There is no discernible bias in the cloud amount. The investigators are confident about the total cloud amounts. High cloud amount estimates are less reliable since there is greater uncertainty in identification of the exact type of cloud than in its amount. The RMS error on the total cloud amount has been computed to be 0.7 units.

Dr. Wood has discussed in a symposium article the use of the Rain 30-Minute Data for validating a water balance model for Kings Creek. Specifically, spatial variability within this area was examined as well as the parameterization of these models using remotely sensed data.

Dr. Grossman used data obtained by the portable automatic meteorological network in his evaluation of convective boundary layer budgets of sensible heat and moisture over an unstressed vegetation region.

Anonymous. 1993. Methods Manual for Konza Prairie Research Natural Area. Konza LTER publication.

Brock, F.V., G.H. Saum, and S.R. Semmer. 1986. Portable automated mesonet II. J. Atmos. Oceanic Tech. 3(4).

Famiglietti, J.S., E.F. Wood, M. Sivapalan, and D.J. Thongs. 1992. A catchment scale water balance model for FIFE. J. of Geophys. Res. 97(17):18,997-19,007.

Federal Meteorological Handbook No. 1. Surface Observations. Third Edition - 1982. U.S. Department of Commerce, National Oceanic and Atmospheric Administration, Washington, D.C.

Federal Meteorological Handbook No. 2. Surface Synoptic Codes. FCM-H2-1988. U.S. Department of Commerce, National Oceanic and Atmospheric Administration, Washington, D.C.

Henderson-Sellers, A., G. Seze, F. Drake, and M. Desbois. 1987. Surface-observed and satellite-retrieved cloudiness compared for the 1983 ISLSCP special study area for Europe. J. Geophys. Res. 92:4019-4034.

McGuffie, K., and A. Henderson-Sellers. 1989. Almost a century of "imaging" clouds over the whole-sky dome. Bull. Amer. Meteor. Soc. 70:1243-1253.

Militzer, J.W. 1987. User's Manual 001-44PAMII-001, Version 1.0, Portable Automated Mesonet II (PAMII). In: Nicolaidis, C. (ed.), Field Observing Facility, Atmospheric Technology Division, National Center for Atmospheric Research, Boulder, Colorado 80307-3000.

National Weather Service Observing Handbook No. 2 - Cooperative Station Observations. July, 1989.

WBAN Hourly Surface Obserations - 1440. Reference Manual. Data Processing Division, ETAC, USAF. National Climatic Center, NOAA.