National Center for Atmospheric Research
ATD...


Table of Contents

Introduction

This document is a standard product of NCAR/ATD/RTF which gives an overview of the measurements taken by PAM and ASTER and conditions during the CASES97 field experiment. This document can be obtained either in hard copy from RTF or in electronic form from the NCAR/ATD WWW site.

Dataset Status

The first 10 days or so of data are often missing and have errors as the stations were being set up. Please ignore these values unless you specifically need data before 20 April.

***** NEW: The data values were reprocessed 5 June 2000 to fix krypton hygrometer values (see below). Note that a lot of sensible heat flux values (H) are now missing, since the conversion from sonic temperature to actual temperature depends on humidity. For approximate values of H during these periods, use: rho*Cp*w'tc', where rho~1.16 g/m3 and Cp~(1006 + 1857*Q*0.001) J/kgK. *****

The 5-minute and 30-minute datasets have been corrected for:

A lower resolution dataset contains statistics averaged over 30-min periods for the entire project. These have been generated with all corrections applied to the data, though outlier values may still be present. In addition, some quantities have been derived, such as the sensible heat flux (H), latent heat flux (LE.dry), and soil heat flux evaluted at the surface (Gsfc).

In addition to the corrections mentioned above, the 30-minute dataset contains:

Table of Variables

A complete table of variables is available for the 5-min. statistics and the 30-min. statistics.

Data Access

Data for CASES97 are available in several forms: Also available is a computer-readable logbook of comments noted by RTF personnel.

Location

There are 6 PAM sites and 2 ASTER sites throughout the lower Walnut River watershed south and east of Wichita, Kansas. The sites were selected to have terrain and land cover typical of the watershed, with good exposure to the predominate winds (SE-S-SW and NW-N). The fetch for these winds generally is at least 200m. The positions of various sites were:
  1. PAM (map): rangeland, 37deg 35.44' N, 96deg 39.10' W
  2. PAM (map): rangeland, 37deg 33.11' N, 96deg 33.31' W
  3. PAM (map): new corn/beans, 37deg 29' 40" N, 97deg 01.90' W
  4. PAM (map): milo stubble/rangeland, 37deg 35.59' N, 96deg 57.12' W
  5. PAM (map): new milo, 37deg 30.21' N, 97deg 08.38' W
  6. PAM (map): winter wheat (floodplain), 37deg 16.30' N, 97deg 00.56' W
  7. ASTER (map): winter wheat, 37deg 26.65' N, 96deg 59.97' W
  8. ASTER (map): rangeland, 37deg 24.04' N, 96deg 56.80' W

Sensors

Most of this section will be completed later.

The PAM sites were instrumented with standard sensors and the ASTER sites were instrumented to function similar to the PAM stations. In addition, other sensors were added to the ASTER sites (see below). Generally, chemistry and spectral-resolving radiation sensors were added at site #7 and site #8 was used for testing sensors.

Since NCAR/ATD did not have enough flux sensors to instrument 8 stations, it was necessary to obtain sensors from several sources, and thus a mix of sensors was used for CASES. These were:

  1. Gill sonic; krypton hygrometer
  2. Gill sonic; krypton hygrometer
  3. ATI sonic; krypton hygrometer
  4. ATI sonic; krypton hygrometer
  5. ATI sonic; krypton hygrometer
  6. Gill sonic; krypton hygrometer
  7. ATI sonic; OPHIR hygrometer
  8. ATI sonic; OPHIR hygrometer
We are especially grateful to Chris Fairall of NOAA/ETL for the use of 4 of these sonic anemometers and 2 hygrometers.

Soil measurements were made to characterize the first 10 cm depth. The temperature probe was inserted on a slant to average from 1-9cm. The (automatic) soil moisture probe was inserted horizontally at 5cm and its radiation pattern roughly integrates from 3-7cm. Soil moisture comparison measurements were made manually using (TRIME) probes inserted on a slant to integrate between 0-10cm and using core samples taken from 1-7cm (note that this is weighted higher than the other measurements).

Several non-standard sensors were used for this program:

Known Instrument Problems

IN PROGRESS.

Rain Gauges

The rain gauge field calibration checks are summarized in a table. Most gauges read within 5% of the nominal reading, which is considered acceptable. The gauge at site 5 was moved when this site was finally deployed at which time the calibration shifted by 10%. We will do a post- calibration to determine if this was due to mis-leveling or an actual change in the calibration. In any case, the data probably should be corrected by 8% (the average of the two calibrations). This has not been applied to the NetCDF files, but may in the future. Similarly, it may be desirable to adjust the calibration at site 6.

I will remove the jumps in the data caused by these calibrations from the final data set.

A plot of the accumulation time series for the period 18 April - 26 May is available (without any adjustments). The total accumulations (both with and without adjustments for the above calibrations) at the end of this period are shown in another table. Note that both ETI gauges read less than the MRI gauges by about 30%. (Although Ed Brandes has noted that the ETI is sometimes higher than the MRI for an individual rain event.)

Soil Moisture

The soil moisture field calibration checks are summarized in a table. These have been used to create new calibrations, which is described in a report.

Pyrgeometers

Pyrgeometers were deployed at site #7 to measure long-wave (infrared) radiation. The sensor has a thermopile to measure the difference in temperature between an exposed black plate and the metal housing. The temperature of the housing, along with the dome and the case of the instrument, are measured seperately with thermistors. The equation normally used to compute the radiation from these measurements is:

Rlw = Rpile + sigma Tcomp^4 - [1.11 sigma (Tdome^4 - Tcase^4) + 0.036 Rsw]

However, the Tdome signal from the downward-looking radiometer (measuring pyg.out) was bad, with an offset of about -8 C and noise. Since the sensor operated normally in the lab, and the A/D channel passed its check-out, the problem presumably was an electrical connection. This isn't a serious problem, since we expect Tdome to be nearly equal to Tcase for the downward-looking sensor. Thus, the Tdome^4 - Tcase^4 term above should be neglected for pyg.out.

Chronology

Day: Action

Data Processing Notes

STILL UNDER PROGRESS.

Sonic Anemometer Tilt Correction

Tilt corrections will be applied to the final data set (they have not been done in the preliminary data sets). This correction compensates for any misalignment of the sonic anemometers with respect to a plane parallel to the surface of the ground. Generally, the anemometers are set up so that the vertical velocity is reasonably well aligned (within 1 degree) to gravity. However, our alignment of these sensors may not be perfect, the mount may sag with time, and the surface usually is not perfectly horizontal.

We generate this correction by fitting to the equation: W = Wo + a*U + b*V, where U, V, W are 5-minute averages of the measured wind components and the fit is done over the entire period that the instrument was not moved. The fit is reported as an offset in W, a "lean angle" (where 0 is vertical), and a "lean direction" (where 0 is along the azimuth of the anemometer u-component). An example of this fit is given here for the CSAT anemometer.

We use one fit for the entire data set, unless we know of a discontinuity. There were only 2 such events during CASES; the u-component was readjusted (and bias might have changed) at site 3 on 21 April and the entire anemometer was raised at site 6 on 7 May. However, neither event appeared to affect the fit significantly (based on examining the time series of the fit coefficients when a separate fit was generated for each day), so only one fit was used. For a few other sites, the fit differed as a function of time, however this probably was due to limited wind directions, and data composited over the entire period had a reasonably low amount of scatter. The resulting coefficients from this process are given in a table. Note that many of the biases were rather large (about 5 cm/s), which may have been the result of preferentially locating the towers at the top of ridges.

Daily Weather Plots

THE PLOTS IN THIS SECTION ARE NOT FINAL.

The following plots summarize conditions at each station during each day of the project. Each plot covers one and a half days (0100-1300+1 CDT) and is labeled with time in GMT at the bottom and local time (CDT) at the top. The top panel displays temperature and specific humidity measured at 2m, pressure, and precipitation rates (if present). Below that is a plot of wind speed and direction measured at 10m, with dotted lines showing the directions in which flow distortion by the towers should not be a problem (there may be more). The next panel shows net radiation, and sensible and latent heat flux. The bottom panel shows the Monin-Obukhov stability parameter, z/L, the friction velocity, u*, and the Bowen ratio calculated from the flux data. Since these fluxes and derived parameters are based on smoothed, 5-minute average statistics, they should not be used quantitatively and are only shown for guidance in selecting periods to analyze further.

Other plots

Rain

Soil Moisture

Downwelling radiation

These are plots of downwelling shortwave radiation measured by the Li-Cor sensors at each site versus that from the Epply at site 7 for mostly clear cases. If there were no clouds, we expect these to be nearly the same. A second-order fit was applied to these data, which shows some systematic differences, though all within the +/-5% specification for these radiometers by Li-Cor. It is possible to correct these data using these fits, but be cautioned that we don't know the reason for these differences. All of this analyses were contributed by Peggy Lemone.

Site 1, Site 2, Site 3, Site 4, Site 5 (first part), Site 5 (entire period), Site 6, Site 7, Site 8

Roughness length

This is a plot of histograms of the log10 of zo for each site, computed from the near-neutral 30-min average statistics by:
 zo <- 10*exp(-0.4*dat("Spd")/dat("u*"))
 zo[zo<1e-6] <- NA
 ineut <- abs(dat("w't'")<.01
 par(mfrow=c(3,3))
 for (i in 1:8) {
    hist(log10(zo[ineut[,i],i]),n=30)
    title(paste("#",i,"; ",round(10^median(log10(zo[ineut[,i],i]),na.rm=T),
	dig=4),sep="")) }
The title is the site number and the median roughness length (in meters). The values are (in cm):
Site:		1	2	3	4	5	6	7	8
Cover:		range	range	crn/bns dirt	wheat	wheat	wheat	range
zo (cm)	median	0.6	0.2	0.7	1.3	0.8	5.3	1.4	0.8
zo (cm)	maxhist	1.0	0.4	0.7	2.2	4.5	7.1	7.1	2.5
I don't quite believe these results, which says that wheat had a roughness of about 5cm, dirt about 2cm, rangeland and planted crops about 1cm. Three weaknesses with this analysis are that all wind directions have been lumped together, only heat flux is used to characterize near-neutral, and z was set to 10m, ignoring displacement height (this last should be a relatively minor problem).

[an error occurred while processing this directive]
This page was prepared by Steven Oncley, NCAR Research Technology Facility