Soil Measurements for Carbon Science

Background

The ARM program has been collecting soils data at a number of its facilities in the Southern Great Plains site.  These data are valuable because water is an important factor limiting primary production.  The combination of carbon flux data, atmospheric climate data, and soil information at these sites is unique.

The Soil Water and Temperature System (SWATS) collects soil measurements from a profile of soil depths.  At each site and depth, hourly measurements are produced at two, side-by-side locations (designated east and west).   Table 1 shows the variables measured and their variable names in the database. The measurements are further described on the ARM website: http://www.arm.gov/instruments/static/swats.stm.

Table 1. Measurements in the SWATS data

Available Data

In addition to the raw SWATS data, which are available from the main ARM archive pages (  URL ), we produced user-friendly datasets for the purpose of validating Carbon models that predict soil moisture.  Each file below contains our best estimates of the variables above for hours in which all quantities were available (i.e., excluding missing values).  Our methods of assigning flags are described below: #Criteria for removing measurements.

To produce filtered datasets for validation, we apply the following rules.  We first screen the data to see if TRISE is missing or if both east and west measurements of WATCONT are flagged.  If so, all data for that hour are removed.  In the remaining data, we compare east and west measurements for each variable.  If neither measurement is flagged, we average the two measurements.  If one non-flagged measurement is available, we use it.  The resulting data have one non-missing value for each measurement.  Flagged raw data, filtered data and plots can be obtained from the URLs below:

Filtered data: http://www.archive.arm.gov/Carbon/dataproducts/soils/BEST/

Raw, flagged data: http://www.archive.arm.gov/Carbon/dataproducts/soils/RawSWATS/

Quicklook plots:  http://www.archive.arm.gov/Carbon/dataproducts/soils/QuickLooks/

Criteria for removing measurements

Problems with ice:  As described earlier, we excluded hours with TRISE flagged because these indicate ice on the sensor and imply that all measurements are unreliable.  Because TRISE flags did not remove all apparent anomalies during winter, we also flagged measurements of WATCONT with values exceeding the 99^th percentile and soil temperatures less than 1 degree C.

Spikes:  We flagged WATCONT measurements with spikes, defined as a change of 20% or more in one hour, followed by another change of 20% or more during the next hour.  The rationale is that "real" changes of this magnitude should persist longer than one hour.

Rain flags:  We compared soil data against meteorological data to ensure that water content measurements were consistent with recent precipitation measurements.  These flags were only applied in soil layers shallower than 1 m.  We flagged WATCONT measurements exceeding the 99^th percentile if it had not rained within 96 hours.  We also flagged WATCONT measurements below the 1^st percentile between 4 and 96 hours after rain.  The lower limit of 4 hours was imposed to allow for a delay in wetting of soil layers at intermediate depths.

E-W comparison flag:  Finally, we flagged both east and west measurements of WATCONT if the two values differed by more than 20% from the maximum of the two measurements.

Table 2.  Number of measurements by year for each facility in the ARM/CART SGP site.

Soil Texture Analysis

Carbon models, such as SIBD, often require descriptors of soil properties.  For the ARM/CART facilities, an analysis of texture at various depths was conducted by Oregon State University.  The results are summarized below.  More information is available at http://www.arm.gov/instruments/static/swats.stm.

Table 3.  Soil texture at several depths for each ARM/CART facility.