119
Spatiotemporal Interactions among Ecohydrological Factors
measurements of soil moisture in this study were collected at a depth of 5 cm underground using the FieldScout TDR 300 soil moisture meter (Wilson et al. 2003;
Spectrum Technologies, Inc. 2004). The time-domain reflectometry (TDR) method
has been popular, because it provides measurements of in situ soil moisture content
with high accuracy (Roth et al. 1992; Walker et al. 2001). The TDR 300 sensor rods
used in our measurements were 7.5 cm in length. We measured the soil moisture
content of soil 5 cm below the surface by inserting the probe at an angle of 40°
from the flat ground. Every two sampling locations were more than 1.5 km apart. A
Global Positioning System (GPS, GARMIN GPSMAP 76CSx) was connected to the
soil moisture meter to record the precise corresponding longitude and latitude data.
Before going to the field, the TDR probe was calibrated by a gravimetric measurement method within a range of 0%–30% moisture (i.e., converted the gravimetric to
the volumetric moisture content). An average value of three gravimetric measurements was used to calibrate each TDR measurement.
6.2.2.3 Derivation of Soil Moisture from MODIS Data
The principle of evolutionary computation (EC) is rooted from genetic algorithms
(GAs) first developed by Holland (1975), evolution strategies developed by Rechenberg
and Schwefel (Back et al. 1997), and evolutionary programming developed by Fogel
et al. (1966). All three were eventually combined into one entity called “evolutionary computation” (Gagne and Parizeau 2004). Under the EC framework, the genetic
programming (GP) is generally considered as an extension of GA. The well-known
GP approach was invented by Koza (1992), which became the best advancement to
create best selective nonlinear regression models in terms of multiple independent
variables later on. In this study, we use the GP software, Discipulus, developed by
Francone (1998) to solve the GP model.
Based on the regression relationships developed by the GP technique, soil moisture maps at a 1-km resolution over the study area can be produced (Makkeasorn et
al. 2006). Initially, 16-day EVI and daily LST/emissivity L3 Global 1-km MODIS
satellite images for the date 29 December 2009 were used as independent variable
inputs to the GP model. The GP-based nonlinear function derived in the evolutionary process uniquely links crucial input variables, including EVI and LST, with the
well-calibrated soil moisture data. The soil moisture data set was divided into a GP
model calibration with 40 data points and a GP model verification with 5 data points.
The square of the Pearson product moment correlation coefficient (R-squared) was
used to verify the effectiveness of model development.
6.2.3 eStiMation of evaPotRanSPiRation
ET is the depletion of water of the soil in vapor form through evaporation and transpiration. The loss of water through the plant’s respiration is called transpiration,
whereas the evaporation is the water loss directly through the soil. Osmosis, the diffusion driven by a salinity gradient, forces water to move from plant roots upward
to the leaves and vaporize to the air at the stomata. A high salinity level in soil
would reduce the gradient of salinity, which reduces the driving force of water movement in plants. ET rates would be lower in high-salinity soil under the same weather
Spatiotemporal Interactions among Ecohydrological Factors
measurements of soil moisture in this study were collected at a depth of 5 cm underground using the FieldScout TDR 300 soil moisture meter (Wilson et al. 2003;
Spectrum Technologies, Inc. 2004). The time-domain reflectometry (TDR) method
has been popular, because it provides measurements of in situ soil moisture content
with high accuracy (Roth et al. 1992; Walker et al. 2001). The TDR 300 sensor rods
used in our measurements were 7.5 cm in length. We measured the soil moisture
content of soil 5 cm below the surface by inserting the probe at an angle of 40°
from the flat ground. Every two sampling locations were more than 1.5 km apart. A
Global Positioning System (GPS, GARMIN GPSMAP 76CSx) was connected to the
soil moisture meter to record the precise corresponding longitude and latitude data.
Before going to the field, the TDR probe was calibrated by a gravimetric measurement method within a range of 0%–30% moisture (i.e., converted the gravimetric to
the volumetric moisture content). An average value of three gravimetric measurements was used to calibrate each TDR measurement.
6.2.2.3 Derivation of Soil Moisture from MODIS Data
The principle of evolutionary computation (EC) is rooted from genetic algorithms
(GAs) first developed by Holland (1975), evolution strategies developed by Rechenberg
and Schwefel (Back et al. 1997), and evolutionary programming developed by Fogel
et al. (1966). All three were eventually combined into one entity called “evolutionary computation” (Gagne and Parizeau 2004). Under the EC framework, the genetic
programming (GP) is generally considered as an extension of GA. The well-known
GP approach was invented by Koza (1992), which became the best advancement to
create best selective nonlinear regression models in terms of multiple independent
variables later on. In this study, we use the GP software, Discipulus, developed by
Francone (1998) to solve the GP model.
Based on the regression relationships developed by the GP technique, soil moisture maps at a 1-km resolution over the study area can be produced (Makkeasorn et
al. 2006). Initially, 16-day EVI and daily LST/emissivity L3 Global 1-km MODIS
satellite images for the date 29 December 2009 were used as independent variable
inputs to the GP model. The GP-based nonlinear function derived in the evolutionary process uniquely links crucial input variables, including EVI and LST, with the
well-calibrated soil moisture data. The soil moisture data set was divided into a GP
model calibration with 40 data points and a GP model verification with 5 data points.
The square of the Pearson product moment correlation coefficient (R-squared) was
used to verify the effectiveness of model development.
6.2.3 eStiMation of evaPotRanSPiRation
ET is the depletion of water of the soil in vapor form through evaporation and transpiration. The loss of water through the plant’s respiration is called transpiration,
whereas the evaporation is the water loss directly through the soil. Osmosis, the diffusion driven by a salinity gradient, forces water to move from plant roots upward
to the leaves and vaporize to the air at the stomata. A high salinity level in soil
would reduce the gradient of salinity, which reduces the driving force of water movement in plants. ET rates would be lower in high-salinity soil under the same weather