Spatial and Temporal Analysis of Soil Moisture using MODIS NDVI and LST Products

1
Spatial and Temporal Analysis of Soil Moisture
usingMODIS NDVI and LST Products J.M. Shawn
Hutchinson1, Thomas J. Vought1, and Stacy L.
Hutchinson21Department of Geography and
2Department of Biological and Agricultural
Engineering, Kansas State University, Manhattan,
Kansas 66506
Impact of Maneuver Training on NPS Pollution
Military readiness depends upon high quality
training. Effective maneuver training requires
large areas of land and creates intense stress on
this land. Environmental protection requirements
place additional restrictions on land use and
availability. Because military training
schedules are set well in advance to make the
best use of installation training facilities and
National Training Centers, there is little
flexibility to modify training events and
maintain readiness. In order to avoid maneuver
restrictions, proactive management plans must be
developed giving commanders the information they
need to assess the environmental cost of training
and management practices that reduce the
environmental impact. Non-point source (NPS)
pollution has been called the nations largest
water quality problem, and its reduction is a
major challenge facing our society today. As of
1998 over 290,000 miles of river, almost
7,900,000 acres of lake and 12,500 square miles
of estuaries failed to meet water quality
standards. Military training maneuvers have the
potential to significantly alter land surfaces in
a manner that promotes NPS pollution, resulting
in the inability of military installations to
meet water quality standards and the decline of
training lands. The overall objective of the
parent project of this research, funded through
CP1339 (Characterizing and Monitoring Non-point
Source Runoff from Military Ranges and
Identifying their Impacts to Receiving Water
Bodies) is to identify sources of NPS pollution
resulting from military activities, assess the
impact of this pollution on surface water
quality, and provide information for commanders
to lessen the impact of training on water quality
(Figure 1). Investigators are assessing the
impact of two major sources of NPS pollution on
surface water quality at Fort Riley, Kansas (1)
erosion from upland training areas and (2)
channel erosion at stream crossing sites.
Researchers are using watershed water quality
models in conjunction with remotely sensed
information and geographic information systems
(GIS) to assess the impact of training on water
quality, in particular on the amount of soil
erosion. A pair of decision matrices, the first
addressing the generation of NPS pollution and
the second the potential to exceed TMDL
regulations for NPS pollutants (i.e., an
Environmental Decision Support Tool), will be
created for assessing the environmental cost of
training maneuvers (Figure 2). In addition,
researchers are collecting surface runoff at
three buffer sites to determine the effect of
vegetated buffers for controlling NPS pollution
and using new real-time data collection systems
to assess the impact of vehicle crossings on
stream water quality and erosion dynamics at Low
Water Stream Crossings (LWSCs). Soil Moisture
A Critical VariableSoil moisture is a critical
variable that contributes to the physical
processes, biogeochemistry, and human systems
that influence global change (Henderson-Sellers
1996). Increasingly, remotely sensed data are
being used in land surface climatology research
and modeling efforts. In addition, antecedent
soil moisture conditions affect the hydrologic
behavior of an area through the partitioning of
precipitation into runoff and storage terms.
However, the value of soil moisture as an
environmental descriptor or as model input is
lessened by our inability to measure it in a
consistent and spatially comprehensive manner. At
the root of this problem is the natural spatial
and temporal variability of soil moisture
conditions, caused by the heterogeneity of soil
properties, topography, land cover, and
precipitation. In remote sensing, plant
spectral reflectance characteristics permit
ability to sense variations in green biomass
while the small thermal mass of plant leaves
distinguishes green vegetation from soil
backgrounds (Tucker 1979, Goward et al. 1985,
Carlson et al. 1995). For many years, surface
radiant temperature measurements used to define
model parameters of soil moisture availability
and thermal inertia (Gillies and Carlson 1995).
Other research has shown that a strong negative
relationship between surface temperature (Ts) and
normalized difference vegetation index (NDVI)
over different biomes, the slope of which can be
used as landscape-level proxy for canopy
resistance (Rc) and wetness (Nemani and Running
1989 Nemani et al. 1993) (Figure 3). Others have
illustrated inversion methods for computing
surface soil water content from measurements of
surface temperature and a vegetation index
(Gillies et al. 1997). Estimating Soil Moisture
via Remote SensingThe objective of this subtask
of the SERDP-funded project, Impact of Maneuver
Training on Water Quality and NPS Pollution, is
to develop spatially- and temporally-distributed
estimates of near surface soil moisture. These
estimates will be used to evaluate antecedent
soil moisture conditions and used as input into a
landscape-scale surface water quality model to
evaluate the effectiveness of riparian buffers in
filtering sediments transported from upland sites
within mechanized military training
areas. Specifically, the relationship between
land surface temperature (LST) and NDVI will be
investigated (see Figure 3) and, if valid, each
satellite image product will be used as
independent variables in a linear regression
model to predict volumetric water content (VWC)
of near surface soils. The study site is Fort
Riley, Kansas (Figure 4). Located in the
northeastern portion of the state, Fort Riley is
an Army training installation, approximately
39,800 hectares in area, for multiple brigades of
armored and mechanized infantry units. Near real
time MODIS satellite data products were obtained
from the EOS Data Gateway. Land surface
temperature (Figure 5) and NDVI data (Figure 6)
is in the form of 8-day and 16-day maximum value
composites, respectively, at a spatial resolution
of 1 kilometer. Soil moisture was measured
weekly at 80 control points using a portable
time-domain reflectometer (TDR). If multiple
points were located within the same 1 km x 1 km
image grid cell, TDR measurements were averaged
to define soil moisture at that location. In a
geographic information system (GIS), LST, NDVI,
and TDR points were overlayed to create a table
of values that could be analyzed statistically.
Overall Technical Approach
DATA COLLECTION
Assess/Identify NPS Pollution
Characterize Stream Sediment
Buffer Field Study
MODELING DESIGN
Real-Time Sediment Load Sensor
Buffer Model Development
Quantify Vegetation Impacts
Stream Crossing Evaluations
NPS PollutionModeling
ASSESSMENT
EnvironmentalDecision SupportTool
DELIVERABLE
VWC 19.204 0.091 (NDVI) 0.039 (LST) R2
0.11 SE 9.3
Figure 1. Technical approach of the project,
Assessing the Impact of Maneuver Training on NPS
Pollution and Water Quality.
Figure 7. Predicted VWC values for Fort Riley
during the 16 day composite period including June
9, 2004. Field sampling sites shown as point
features.
Preliminary FindingsLST and NDVI values vary
significantly among image dates within study
area, indicating a scaling technique may be
necessary for a single regression-based model to
be applicable. The field sampling date/composite
period of June 9 with the largest variation in
NDVI, LST, and soil wetness values produced the
best linear regression model (Figure 7), despite
weakest negative correlation between LST and
NDVI. Other dates showed a significant negative
relationship between LST and NDVI. However, the
more more homogeneous dry or wet conditions
yielded poor model results. Field data collection
will continue and more accurate and precise
measurement techniques will be incorporated
(e.g., gravimetric sampling). Concurrent
research is comparing MODIS enhanced vegetation
index (EVI) with NDVI by composite period, and
over time, to assess the suitability of EVI as a
replacement vegetation index. In addition to
testing various data scaling techniques to
standardize LST and VI values, nonlinear
regression models will be explored to improve
variable significance and the accuracy of
predicted VWC values. References Carlson, T.N.,
R.R. Gillies, and T.J. Schmugge. 1995. An
interpretation of methodologies for indirect
measurement of soil water content. Agricultural
and Forest Meteorology 77(3-4)191-205. Gillies,
R.R. and T.N. Carlson. 1995. Thermal remote
sensing of surface soil-water content with
partial vegetation cover for incorporation into
climate models. Journal of Applied Meteorology
34(4)745-756 Gillies, R.R. T.N. Carlson, J. Cui,
W.P. Kustas, and K.S. Humes. 1997. A
verification of the triangle method for
obtainin surface soil water content and energy
fluxes from remote measurements of the normalized
difference vegetation Index (NDVI) and surface
radiant temperature. International Journal of
Remote Sensing 18(15)3145-3166. Goward, S. N.,
C. J. Tucker, and D. G. Dye. 1985. North
American vegetation patterns observed with the
NOAA-7 advanced very high resolution radiometer,
Vegetatio 643-14. Henderson-Sellers, A. 1996.
Soil moisture A critical focus for global
change studies. Global and Planetary Change
133-9. Nemani, R.R. and S.W. Running. 1989.
Estimation of regional surface-resistance to
evapotranspiration from NDVI and thermal-IR AVHRR
data. Journal of Applied Meteorology
28(4)276-284. Nemani, R.R., L. Pierce, S.W.
Running, and S. Goward. 1993. Developing
satellite-derived estimates of surface moisture
status. Journal of Applied Meteorology
32(3)548-557. Tucker, C.J. 1979. Red and
photographic infrared linear combinations for
monitoring vegetation. Remote Sensing of the
Environment 8127-150. Acknowledgements This
project is funded by the Strategic Environmental
Research and Development Program through CP1339
(Characterizing and Monitoring Non-point Source
Runoff from Military Ranges and Identifying their
Impacts to Receiving Water Bodies).
Co-investigators of this project are (from Kansas
State University) James M. Steichen, Phillip L.
Barnes, Naiqian Zhang, Charles G. Oviatt, Naiqian
Zhang, and (from the Fort Riley Integrated
Training Area Management (ITAM) Program) Philip
B. Woodford. Field work during year one of this
research effort was performed by graduate
students Scott Leis, Ben White, and Brooke
Stansberry (Department of Geography, Kansas State
University).
Potential NPS Pollution Generation
Environmental Decision Support Tool
Graphs
Figure 2. Decision support tools designed to
assist installation officials better evaluate the
potential environmental impact of scheduled
training activities.
Figure 5. NDVI values for Fort Riley from the
composite period including June 9, 2004 Field
sampling sites shown as point features.
Normalized Difference Vegetation Index, NDVI
(Min -1.0 to Max 1.0)
NDVI Image
LST Image
Land Surface Temperature, LST (oC)
Figure 3. Scatterplot of LST and NDVI from three
image dates for Fort Riley and surrounding
counties showing typical relationship between the
two measurements.
Figure 6. LST values for Fort Riley from the
composite period including June 9, 2004. Field
sampling sites shown as point features.