[HWR logo]

Preliminary Analysis of GLOBE Soil Moisture Data

Dr. James Washburne
Department of Hydrology and Water Resources
University of Arizona

Outline

  • Motivation, Illustrations

    How does Soil Moisture Vary in Time and Space?

    Preparing Student Data for Analysis Current Status
  • Guilelines for Data Collection

    Motivation

    The soil moisture data collected by GLOBE students represents a unique and valuable resource to earth scientists. It has the potential to become the first readily accessible, long-term and consistent record of near-surface soil moisture available on a global basis. Soil moisture data can be used to test and validate satellite and model data sets by providing a true record of actual field conditions as a function of time. Satellite validation requires coincident ground and satellite sampling. Model validation requires regular measurements, particularly during periods when the soil moisture status is changing. Seasonal to weekly sample intervals are acceptable depending on the specific application. There is also a need to integrate soil moisture data into plant ecology models. This data is most useful at daily to monthly intervals, depending upon the model objectives. Thus, GLOBE protocols require data be collected from naturally moist areas either without a canopy (to be viewable by a satellite) or near a vegetated plot of interest. Further selection of a sampling strategy requires one to assess the regional variability in soil moisture levels. This is most easily done after viewing seasonal movie loops of modeled soil moisture estimates.

    How Does Soil Moisture Vary in Time and Space?

    What Climate Models Tell Us - Spatial Patterns

    The European Center for Medium-range Weather Forecasting (ECMWF) has recently published a series of CD's with a sequence of 10 ten-day average forecasts that make up a set (or ensemble) of likely (statistically-related) forecast alternatives for the 14 year period 1979-1993. I have used the GraDs graphical package and other tools to produce a 36 image sequence of the 10 day mean ensemble forecast near- surface soil moisture for the 1979-1980 period. Each model pixel represents an area approximately 2.5 by 2.5 degrees (~250 km) square. Despite this poor spatial resolution, clear seasonal patterns are evident but the timing of these seasonal changes varies by region (     Movie Loops ). Next, I selected two very different areas to illustrate these patterns in the Northeast US (PA), and Desert Southwest (AZ).

    What GLOBE Students Tell Us - Temporal Patterns

    GLOBE students were not collecting data during the same period covered by the ECMWF model reanalysis but we can try to compare what data we have to see what differences or similarities stand out. For simplicity I have abstracted the observed patterns and plotted these with patterns derived from the ECMWF analysis (Figure 4). Some patterns and relative magnitudes appear to be approximately correct. However, there are many reasons why these two plots may not be comparable.

    Preparing Student Data for Analysis

    Antecedent Precipitation Index

    Few GLOBE observations are particularly interesting taken alone. It is when taken together with other regional and diverse data to look for similarities that we learn the most. Precipitation is the variable of greatest interest to compare with soil moisture as there is a clear relationship between the two variables. Hydrologists have defined a parameter, derived from precipitation measurements, that even more closely exhibits patterns similar to those observed in soil moisture. This relationship is called the Antecedent Precipitation Index or API. One way to explain API is as a leaky bucket which loses a fixed percent of the water in it every day. New precipitation adds to the depth of the water in the bucket. The API is just a plot of the bucket water level over time and looks like a decaying or asymptotic curve between wet periods, Figure 5. GLOBE soil moisture data will help us select the best API parameters to use for a given area.

    Gypsum Block Calibration

    One limitation of the gypsum block is that it is only sensitive to relatively dry soils. In wet areas this results in "saturation" conditions where the observation is insensitive to wet soils. This saturation point appears to lie between a soil water content of 0.1 and 0.2 g/g (Figure 6), depending upon the soil texture.

    Student Data Plots

    Much of the value of GLOBE student observations of soil moisture lies in seasonal patterns that become evident in consistent and regularly-spaced data collected over periods of several years. While both gravimetric and gypsum block data are suitable for determining seasonal trends, currently the long-period (2 year) data sets in the GLOBE archive consist of gypsum block data. Some representative data are plotted (Figures 7-8) along with the Antecedent Precipitation Index, Precipitation and two cross plots useful for determining correlations between various parameters. One problem with this data in the current student data archive is that direct gravimetric observations are indistinguishable from calibration values and both calibration and raw meter reading data reside in the same archive field. We are working to fix this problem. Your help is requested to contact GLOBE schools in your area, who have made soil moisture measurements, and to have them send tabulated calibration gravimetric data to me as soon as possible.

    Current Status

    Historical Protocol Development

    Participation Overview

    After atmospheric observations, soil moisture data rank second in number (over 21400) with more than 50 schools in 11 countries and in 40 US states contributing data although 80% of the data come from roughly half of these groups. Since most of these data are gypsum block measurements, 21400 must be divided by four (5350) to get a clearer estimate of the number of school days of data that have been collected. Only 50% of the gypsum block data are calibrated, and some of these calibrations must be repeated because too few useful calibration points were used. The number of schools reporting data on any given day appears to be: 4 for Summer of '97, 10 for Spring '97 and 15 for Fall '96. This should not be equated with the maximum level of school participation, which is larger due to irregular sampling.

    Distribution of Measurements

    The new GLOBE data visualization pages (http://preview.gsfc.nasa.gov/cgi-bin/cgi.html) allow one to access the spatial variation and cumulative number of soil moisture observations over North America (Figure 1) and the rest of the globe.

    Guidelines for Data Collection

    First, please collect site metadata which will help others characterize the conditions and circumstances unique to your site. Schools making gypsum block measurements should continue to collect a calibration (30 cm) of full depth profile gravimetric sample monthly as a check on data quality and calibration validity. Gypsum block data are only being encouraged in dry, non-acidic soils; the gravimetric data is much more acceptable to the larger scientific community.
    Imagine how a soil moisture movie would look over your region. Concentrate 12 regularly spaced samples during periods of transition between wet and dry conditions. If the soil moisture in your area in high for long periods, particularly during the school year, monthly samples are adequate to characterize and generally monitor these values.
    I generally recommend that younger students to take the near-surface gravimetric measurements. Older students are able to work the auger more easily and then it becomes more worthwhile and reasonable to make depth profile measurements.