Overview

The soil environment is one of extreme heterogeneity, at even a sub-millimeter scale, while above the soil, surface-canopy-atmosphere fluxes integrate over tens of meters to kilometers.  Remote sensing energy balance models are commonly applied to estimate energy and mass fluxes of soil surfaces, however even the highest resolution of satellite data (~10 m) introduces significant errors in pixel-averaged heat flux estimations, especially for heterogeneous landscapes.

The primary goal of this research is to examine the soil spatial and temporal heterogeneity “within a pixel” by using a series of models relating aboveground microclimate and soil energy balance measurements to belowground measurements made by AMARSS.  One broad hypotheses of the AMARSS group that can be related to the aboveground NIMS studies is that soil CO2 fluxes occur at predictable soil moisture and temperature values, but become spatially and temporally complex depending on the local characteristics of the forest overstory.   Thus, soil temperature and water content, the two parameters that most influence the biological processes in the soil directly relating to CO2 flux, will be modeled from aboveground measurements.

Approach

The AMARSS project at the James Reserve is currently collecting high-resolution spatial and temporal soil data with 10 stations in an 80 m transect in the forest understory.  Each station consists of an array of belowground sensors including soil CO2, soil temperature, soil water content, and aboveground air temperature, relative humidity, and photosynthetic active radiation (see the AMARSS report for more details).  Manipulations of the soil leaf litter and moisture content will allow us to calibrate aboveground energy flux measurements to what is occurring belowground.  Using previously published and relatively simple energy models, listed below, aboveground measurements by NIMS of microclimate and soil energy balance over manipulated and undisturbed areas will be related to sub-surface temperature and moisture content measured by AMARSS.

Systems/Experiments

Initial deployments of energy balance sensors have been placed in static locations to determine the usefulness of sensors as well as the variability of data collected.  A secondary deployment included a NIMS RD mobile platform to examine finer-resolution spatial variability.  The data and discussion below are from a NIMS deployment that included a fixed-location “base station” as well as a mobile node; the mobile node was placed in one location over 48 h to provide data on temporal variations.

Aboveground Sensors and Measurements. 

The first grouping is a collection of sensors set in a fixed location (NIMS base station), the second is the collection carried on the NIMS shuttle:

• Four net radiometers for total net solar and terrestrial radiation.
• Fine gage thermocouples for soil surface temperature.
• Six soil heat flux plates at 8 cm depth
• Twelve soil temperature probes at 2 and 8 cm depths
• Cup anemometer for wind speed.
• Several pairs of Bowen-ratio thermocouples for air temperature (pending).
• One all-wave net radiometer for total and component net radiation.
• Silicon pyranometer and PAR sensor.
• One infrared thermometer for soil surface temperature.
• Aspirated air temperature and relative humidity sensors

One Licor infrared gas analyzer for Bowen-ratio water vapor flux and atmospheric CO2 gradients (pending).

 

 

The NIMS RD system at JR sends its data to the CENS database, SensorBase.org automatically.  Visualization tools are the next step towards making this data useable.  For example, this webpage (above) serves as a quick way to verify data from the AMARSS transect.  The list of sensors is displayed with their last recorded value.  Selecting one of the sensors from the list displays a graph of its values over a specified time period.  A live camera feed of the mini-rhizotron can also be displayed when the camera is active.

 

 

The webpage above shows the battery voltages on motes in the field and can be used to monitor additional deployment variables.  The time period can be changed to view patterns in the battery voltage levels.  Individual motes can be selected to view them alone to get a better view (below).

Accomplishments

Because soil temperature data collected from the original ten AMARSS sites shows significant spatial variation, the experimental design was augmented with another 11 temperature sites by Tom Schoellhammer as part of his doctoral thesis.  These sites will “fill in” the original AMARSS design, by placing one new site between each of the original five pairs of sites (each equipped with soil temperature sensors at 2 cm, 8cm, and 16cm, as well as a soil moisture sensor at 8cm), as well as instrument a new 2 m by 3 m patch along the transect (equipped with soil temperature sensors at 2 cm and 8cm, as well as a soil moisture sensor at 8cm; Figure 5 at the end of this document).  These 11 sites were networked wirelessly, allowing real-time sensor data delivery and system diagnostics.  Figure 3 (left) is an empirical semivariogram constructed from data from all 21 sites, using the temperature data from 2 cm for about one week in October 2006.  From this plot we can see that even at short distances of 1 to 2 m there is significant variation in temperature.  This is due to the spatial variability of incident radiation.  For this reason, we are also exploring the use of low-power wireless cameras, equipped with hemispherical lenses, for use in characterizing the lighting regime along the transect.

The Simple Models for Energy Balance

Below (Figure 4) is an example of the range and types of data that we have collected using this system at the James Reserve, apart from the AMARSS data.  Data were collected for 48 hr on the 25th through 27th of October, 2006 at a site near the middle of the AMARSS transect.  Components of solar and terrestrial radiation (Net radiation, net solar, net IR, net total radiation, solar radiation [silicon]) and additional AMARSS data from other sites will allow us to compare different methods and calculations over different spatial scales.

Temperature at Depth Model

A first-pass at determining temperatures at different soil depths is to use a simple sinusoidal damping model with data from temperatures at depth to test and revise the conditions for the assumptions of the model.  The damping depth, which indicates the depth of soil in which the variation in temperature is damped to e-1 of the surface value, depends on water content, although desert soils have been reported to have similar damping depths regardless of water content (Nobel 1989).  The equation below can be used to predict temperature at any depth z:

Where Tz is the soil temperature at depth z, T₀  is the daily mean soil surface temperature, DT0 is the amplitude of the daily variation in surface temperature, d is the damping depth, t is time, and p is the period.

Using the 48-hours of data collected above, an estimate of the damping depth is 2.15 ± 0.27 cm, using surface and 2 cm depth values for three different locations.  Using this value, we can calculate a predicted temperature at 8 cm (Figure 5):

Figure 5.  The predicted temperature value is based on a symmetric sine wave and thus introduces errors in shaded environments like AMARSS.  A refinement of the model would be based on the observed surface patterns rather than a sine wave.

Moisture Balance Model

A first-pass at modeling the moisture balance as measured with the sensors on NIMS and AMARSS is to use a simplified model of energy balance.  The simplified soil-surface energy balance is the sum of the components of energy flux (in W m-2), which should be zero:

Net radiation (Rn) − soil heat flux (G) − sensible heat flux (H) − latent heat flux (LE) = 0

Rn is the incoming shortwave + longwave radiation minus the shortwave + longwave radiation that is reflected or emitted from the surface and can be measured with a net radiometer. Outgoing longwave radiation is emitted from the soil surface and is a function of soil surface temperature, which can be measured with surface thermocouples or an infrared thermometer. Soil heat flux is the energy flux passing conductively through the soil profile and is measured with buried soil heat flux plates. Sensible heat flux is the energy flux transferred convectively between the soil surface and the layer of air over the soil and is best calculated using wind speed and air temperature differences.   Latent heat flux is the flux of energy associated with evaporation from the soil surface, and as such the energy balance can be used to estimate evaporation. It is the relationship between latent heat and evaporation that links energy balance measurements and the water balance of the soil.

Using the middle 24 h of data, a first estimate of the energy loss due to evaporation using the simple energy balance model can be calculated.  Sensible heat is calculated using the single layer (bulk resistance) approach (Moran et al. 1994) as:

where Tsoil is the temperature of the soil surface (K), Tair is the air temperature, r is the density of air (1.21 kg m-3), cp is the specific heat of air (1.010 kJ kg-1 K-1), and r is the aerodynamic resistance of the soil surface to heat transport (s m-1).  Aerodynamic resistance can be calculated empirically (Ben-Asher et al. 1983) as above, where U is wind speed (m s-1).

 

Data was collected every 5 min and was integrated over 24 h to determine total latent heat loss from the soil of 2.8 MJ m-2 d-1.   Figure 6.

 

The instantaneous values of latent heat are reasonable for a wet soil in October.  Negative values during the evening could be associated with dew or frost formation, although this may also be anomalous and with use of refined models may change.  The value of 2.79 MJ m-2 d-1 of latent heat loss can be converted to an evaporative flux using the latent heat of evaporation of water at the soil surface temperature.  In the temperature range 8-12 °C the evaporation heat of water equals 2.43 kJ g-1, thus the amount of water lost from 1 m2 in the AMARSS transect equals 1.1 kg of water, equivalent to about 1.1 mm precipitation.

These values are preliminary, but reasonable for a wet soil, and investigation into more complex models is the next step.

Future Directions

Manipulations to the soil surface will be made, including changes to leaf litter thicknesses and composition (e.g. pine needles or oak leaves) in an area adjacent to the AMARSS transect.  Additionally, an area of soil that is kept dry and an area that is maintained wet will allow us to calibrate above-ground measurements with known below-ground conditions.

Further exploration into more complicated models and larger data sets is planned.  Several target models have already been identified:

Temperature Models.  The goal of this set of models and measurements is to predict soil temperature at depth in locations that have different canopy cover and different leaf litter thicknesses and composition.

Santanello and Friedl 2003: a simplification of the Simultaneous Heat and Water (SHAW) model (that would require eddy covariance measurements above the canopy); the Santanello and Friedl model correlates soil heat flux with net radiation only.

• Bond-Lamberty et al. 2005: a multiple moving average (MMA) model requiring air temperature and overstory leaf area index using hemispherical canopy images.
• Ogée et al. 2001: a model requiring soil surface temperature, humidity, and soil bulk density measurements.  Soil thermal conductivity is calculated using soil temperature at three depths over a seasonal change in soil moisture content.
• Balisky and Burton 1995: soil temperature is predicted using mean surface temperature, leaf litter thickness, and moisture content of the mineral soil.
• Quin et al. 2002: standard energy balance model that requires soil water content and soil physical properties in addition to air and soil temperatures, wind speed, and radiation measurements.  A numerical solution is provided.

Moisture Models.  The goal of this set of models and measurements is specifically to predict soil moisture at depth, although some of the models above can also be used in such calculations.

• Evett et al. 1994: a model of evaporation at the soil surface that requires wind speeds, soil surface temperatures, and a reference cylinder of dry soil buried near the test locations under similar conditions.
• Suleimanand Crago 2004: a model used in remote sensing that uses air temperature and changes in soil surface temperatures to calculate evapotranspiration.
• Abrahamsen and Hansen 2000: Daisy is a crop/soil simulation model with open-source code available at http://www.dina.dk/~daisy/ 

 

Additional planned NIMS RD transects, perpendicular to the long axis of the AMARSS transect, will allow us to correlate additional NIMS and AMARSS data (Figure 7).

 

People

Faculty:

• Bill Kaiser, UCLA, Electrical Engineering
• Phil Rundel, UCLA, Ecology and Evolutionary Biology

Staff:

• Eric Graham, Eric Yuen, and Yeung Lam; UCLA

Graduate Students:

• Thomas Schoellhammer, UCLA