Data Integration for observing LAI Patterns in
Bushra Zaman
Fall 2007
Table of Contents
|
Sr. No. |
Description |
Page No. |
|
1. |
Introduction |
3 |
|
2. |
A brief introduction of the Terms Used |
3 |
|
2.1 |
Leaf Area Index |
3 |
|
2.2 |
Normalized Difference Vegetation Index (NDVI) |
4 |
|
2.3 |
Normalized Difference Water Index (NDWI) |
4 |
|
3. |
Study
Area |
5 |
|
4. |
Objectives |
5 |
|
5. |
Analysis |
5 |
|
6. |
Data
Gathering and processing |
6 |
|
6.1 |
Organizing
the data as per the availability of the Landsat
Images |
6 |
|
7. |
|
6 |
|
7.1 |
Importing
data into ArcGIS |
7 |
|
7.2 |
Generation
of spatial layers of Leaf Area Index |
7 |
|
7.2.1 |
NDVI |
7 |
|
7.2.2 |
NDWI |
7 |
|
7.2.3 |
Model –
LAI and NDVI |
8 |
|
7.3 |
Interpolation
of Precipitation data and creating a raster layer |
9 |
|
7.4 |
Extraction
of Leaf Area Index values and precipitation values at the sites |
11 |
|
8. |
Results |
12 |
|
8.1 |
Spatial
LAI layers |
12 |
|
8.2 |
Saturation
of NDVI |
13 |
|
8.3 |
Effect
of precipitation on soil Moisture |
15 |
|
8.3.1 |
Effect
of Precipitation on Soil Moisture (0-6 cm Depth) |
15 |
|
8.3.2 |
Effect
of Precipitation on Soil Moisture (30 cm Depth) |
16 |
|
8.4 |
Extraction
of slope in percentage from 1/3 arc-second DEM |
18 |
|
9. |
Summary |
21 |
|
10. |
Further
Research |
21 |
|
|
References
and Data Sources |
21 |
List of Figures
|
Figure No. |
Description |
Page No. |
|
Figure 1. |
Typical presentation
of the variation in the active (green) Leaf Area Index |
4 |
|
Figure 2. |
Adding X-Y data to ArcMap |
7 |
|
Figure 3. |
NDWI calculation using Raster
Calculator |
8 |
|
Figure 4. |
Tools for LAI & NDVI |
8 |
|
Figure 5. |
Model for LAI |
9 |
|
Figure 6. |
Model for NDVI |
9 |
|
Figure 7. |
Precipitation Stations |
10 |
|
Figure 8. |
Interpolated precipitation |
10 |
|
Figure
9. |
Tool for Extraction of values
to points |
11 |
|
Figure 10. |
Extraction of values to points |
11 |
|
Figure 11. |
Spatial LAI for |
12 |
|
Figure 12. |
Spatial LAI for |
12 |
|
Figure 13. |
Temporal variation of LAI on Corn Sites |
13 |
|
Figure
14. |
Temporal variation of LAI on Soybean Sites |
13 |
|
Figure
15 |
Saturation of NDVI on Corn
Sites |
14 |
|
Figure
16 |
Saturation of NDVI on Soybean
Sites |
14 |
|
Figure
17 |
Effect of Soil Moisture on Corn
Sites |
15 |
|
Figure
18 |
Effect of Soil Moisture on
Soybean Sites |
15 |
|
Figure
19 |
Schematic diagram of sensor installation |
16 |
|
Figure
20 |
Effect of Rain on Soil Moisture
on Field-15 (Corn Site) |
17 |
|
Figure
21 |
Effect of Rain on Soil Moisture
on Field-24(Corn Site) |
17 |
|
Figure
22 |
Effect of Rain on Soil Moisture
on Field-16 (Soybean Site) |
17 |
|
Figure
23 |
Effect of Rain on Soil Moisture
on Field-23 (Soybean Site) |
17 |
|
Figure
24 |
Variation of LAI with Soil
Moisture on site (WC03) |
18 |
|
Figure
25 |
Digital Elevation Model for the
Study Area |
18 |
|
Figure
26 |
Tool for Projection of DEM |
19 |
|
Figure
27 |
Properties of DEM |
19 |
|
Figure
28 |
Slope calculation for DEM |
20 |
|
Figure
29 |
Slope of DEM |
20 |
|
Figure
30 |
Effect of Slope on Soil
Moisture |
21 |
1. Introduction
An important aspect that has to be considered in the field of
sustainable development is the capacity of vegetation to preserve the ecosystem
functions with an efficient primary production. Among the many parameters that
describe structurally and functionally the vegetation for quantifying the
energy and mass exchange characteristics of a terrestrial ecosystem, the leaf
area index (LAI) is one of the most used (Napolitano
et al, 2001). LAI can be measured in the field in specific land cover types
and then the data can be extended to a territory when cartographic maps of land
cover are available. Today thanks to geographic information system (
Remote sensing data with the increasing imagery resolution is a
useful tool to provide information over various temporal and spatial scales.
The motivation for this paper comes from the fact that part of my research work
is closely related to this project. Studying the physiological and spectral
characteristics of the data was of importance, especially, because I plan to
use the SMEX 2002 data for my research and I needed to study the behavior of
the data.. Also data fusion forms an important part of my research and assimilation
procedures often require remote sensing data over different spectral domains to
retrieve input parameters which characterize surface properties such as albedo, emissivity or Leaf Area Index.
2. A
brief introduction of the Terms Used
2.1 Leaf Area Index
The Leaf Area Index (LAI), a dimensionless quantity, is
the leaf area (upper side only) per unit area of soil below it. It is expressed
as m2 leaf area per m2 ground area. The active LAI is the index of the leaf
area that actively contributes to the surface heat and vapor transfer. It is
generally the upper, sunlit portion of a dense canopy. The LAI values for
various crops differ widely but values of 3-5 are common for many mature crops.
For a given crop, green LAI changes throughout the season and normally reaches
its maximum before or at flowering (Figure 8). LAI further depends on the plant
density and the crop variety ( Allen et
al., 1998, FAO 56 method).
Figure 1 : Typical presentation of the
variation in the active (green) Leaf Area
Index over the
growing season for a maize crop (Ref:
Allen et al., 1998, FAO 56 method)
2.2
Normalized Difference Vegetation Index
(NDVI)
NDVI is the acronym for normalized difference vegetation
index. It is a simple formula using two satellite channels. If one band is in
the visible region (
The reason NDVI is related to vegetation is that healthy vegetation
reflects very well in the near infrared part of the spectrum. Green leaves have
a reflectance of 20 percent or less in the 0.5 to 0.7 micron range (green to
red) and about 60 percent in the 0.7 to 1.3 micron range (near infra-red). The
visible channel gives you some degree of atmospheric correction. The value is
then normalized to the range -1<=NDVI<=1 to partially account for
differences in illumination and surface slope.
NDVI provides a crude estimate of vegetation health and a
means of monitoring changes in vegetation over time. The possible range of
values is between -1 and 1, but the typical range is between about -0.1 (NIR
less than
2.3
Normalized Difference Water Index
(NDWI)
NDWI is equal to [R(0.86 micrometers ) – R(1.24
micrometers )]/[R(0.86 micrometers ) + R(1.24 micrometers )], where R
represents the apparent reflectance. At 0.86 micrometers and 1.24 micrometers ,
vegetation canopies have similar scattering properties, but slightly different
liquid water absorption. The scattering by vegetation canopies enhances the
weak liquid water absorption at 1.24 micrometers . As a result, NDWI is
sensitive to changes in liquid water content of vegetation canopies (Gao, 1996).
The project uses
the LAI, Soil Moisture and Precipitation data gathererd
during the Soil
Moisture Experiments in 2002 (SMEX 2002) conducted in Walnut Creek Watershed,
4. Objectives
The objectives of this project are to:
-
Use the
existing models with Airborne and Satellite Imagery to create spatial LAI
layers. To match up the LAI spatial layers to the Soil moisture ground
measurements and observe the relation between the two.
-
To get the
precipitation data for the study area and to interpolate the precipitation
values at various stations in the area within the
-
Map the study
area with a digital elevation model (DEM) and generate a raster layer for Slope
and observe if soil moisture shows an organized pattern.
5. Analysis
The
following analyses have been done:
-
For creating
the spatial LAI layers, I have built a model that calculates the spatial LAI
for a particular data and the input parameters are the Landsat
image bands (i.e. NIR and
-
The
precipitation values have been interpolated using Kriging
and then further analysis has been done as will be shown next under
-
The
Experimental Sites have been located on the DEM (1/3 Arc Second, Seamless
Server) and the Landsat Imagery. The projection has
been adjusted to represent the sites correctly.
-
The data
layers that have been included in ArcMap are:
- Precipitation
stations
- NED_53082904
(1/3 Arc second DEM)
- Experimental
Sites
- Soil Moisture
ground measurements
- Landsat Images for 6th June, 23rd
June, 1st July, 8th July and 17th July
2002
6. Data Gathering and processing
6.1 Organizing the data as per the availability of
the Landsat Images
Data organization was difficult because I had to
match the data temporally for the different features that I was using e.g soil moisture, LAI etc with the Landsat
image dates.
-
Most of the
data for this study has been downloaded from this website http://nsidc.org/data/amsr
_validation/ soil_moisture/smex02/
-
The Images
from the Landsat TM for five dates were obtained in
proper format i.e. img format. The Landsat TM image for 6th June was in the geotiff format, so it was imported into ERDAS Imagine and
changed to .img format before they were used in ARcGIS.
-
The Digital
Elevation Model for the Study area is downloaded from the USGS seamless server http://seamless.usgs.gov/
. The spatial coverage of the study area (i.e. 41.7°N to 42.7°N, 93.2°W
to 93.8°W) was fed into the server to retrieve the DEM. The DEM is 1/3
Arc-second and the coordinate system is NAD83.
-
There were
some
-
The
precipitation data for 22 precipitation stations were downloaded from the SMEX
2002 website. One SCAN station was also included. Since the precipitation
stations did not cover the entire study area, data at 4 other locations were
downloaded from the DAYMET website.
7.
7.1
Importing
data into ArcGIS
7.2
Generation of
spatial layers of Leaf Area Index (LAI) and Normalized Difference Vegetation
Index (NDVI).
7.3
Interpolation
of Precipitation data and creating raster layers.
7.4
Extraction of
Leaf Area Index values and precipitation values at the sites
7.5
Extract the
slope in percentage from 1/3 arc-second DEM.
7.1
Importing data into ArcGIS
First, the shapefiles
available on the website were downloaded and directed added to ArcGIS. Thereafter, the soil moisture data was downloaded
from the website and inserted into a spreadsheet. Next, it was necessary to convert the
degrees-minutes-seconds data to decimal degrees with four digits to the right
of the decimal point. After the data
preparation, the soil moisture points were imported into ArcMap
using the ‘Add X-Y Data’ function. At
this point it was necessary to specify the spatial reference of the data, which
was WGS 1984. A visual inspection
revealed that the points were in the correct location. Similarly the Rain Gauge
stations were imported to Excel and were added to ArcMap.
Figure 2: Adding X-Y data to ArcMap
7.2
Generation of spatial layers of Leaf Area Index
There were a total of five images obtained from
the website for the dates 6th June, 23rd June, 1st
July, 8th July, 16th July and
The bands were imported into ArcGIS
separately, so that each band becomes a separate raster layer to be used in the
raster calculator. The NDVI and NDWI are calculated using Raster Calculator as
shown in the figure below. The following formulas were used:
7.2.1
NDVI = NIR –
Where, NDVI – Normalized Difference Vegetation
Index
7.2.2
NDWI = NIR –
Where, NDWI – Normalized Difference Water Index
The figure below shows the snapshot of the formula
entered into the raster calculator. The final product is the raster layer for
NDVI and NDWI.
[Here, Band 1 is NIR, Band 2 is
Figure 3: NDWI calculation using
Raster Calculator
Since there are six different dates and a raster
layer for NDVI, NDWI and LAI is needed for each date, a new toolbox was made
that calculates all three of these and takes the bands as inputs.
Figure 4: Tools for LAI &
NDVI
7.2.3
Model – LAI
and NDVI
The following figures show the models that are built
for LAI and NDVI
Figure 5: Model for LAI
Figure 6: Model for NDVI
The input to the LAI model is band-1 (NIR) and
band-3 (
7.2.4 Regression
Equation for Leaf Area Index
The following regression equation is used for
calculating the value of LAI.
Y = (a*VI +b) * (1 + c*exp[d*VI])………………………………………(3)
Where,
Y = LAI, and
a = 2.88; b=1.14; c=0.104; d = 4.1
(Ref:Anderson, M. C., Neal, C. M. U., Li,
F., Norman, J. M., Kustas, W. P., Jayanthi,
H., & Chavez, J. (2004). Upscaling ground
observations of vegetation water content, canopy height, and leaf area index
during SMEX02 using aircraft and Landsat imagery.
Remote Sensing of Environment. Pii:S0034-4257(04)00176-2)
7.3
Interpolation of Precipitation data and creating a
raster layer
This precipitation data set
obtained from the website http://nsidc.org/data/nsidc-0236.html
contains hourly precipitation data taken at 22 rain gauge locations which was
recorded as a part of the Soil Moisture Experiment 2002 (SMEX02). This study
was conducted during June, July, and August 2002 in the
It was observed that the rain gauge stations only
partially covered the study area, hence a few more stations were downloaded
which are marked in red circles
Figure 7: Precipitation Stations
An Excel file was prepared containing data only
for the dates on which the area received rainfall. The file contained a total
of 15 dates and the precipitation values for all the 27 stations on these
dates. The latitude and longitude values of the stations were also listed. The
excel sheet was then imported into ArcGIS and was
converted to a shapefile.
The precipitation was interpolated using spatial
analyst, for those dates that could be linked to the satellite images. Kriging was used to interpolate the precipitation data. The
figure below shows the interpolated surface for precipitation.
Figure 8: Interpolated precipitation
The
points corresponding to the on-ground soil moisture measurements were extracted
from the precipitation raster layer using the Extraction tool in the Spatial
Analyst toolbox as shown below.
7.4
Extraction of Leaf Area Index values and precipitation
values at the sites
After the generation of spatial leaf area index
layers and the precipitation layers, the points on the raster corresponding to
the ground soil moisture measurements were extracted. The spatial analyst Extraction
tool is used for this analysis. This is done by using the tool in the following manner,
Spatial Analyst tools à Extraction à Extract Values to Points. Henceforth, the raster values of LAI, raster values for precipitation
and Soil Moisture point values are available for the latitude and longitude
points on the study area.
The toolbox used for this analysis is shown below.
Figure 9: Tool for Extraction of
values to points
The input point features are the Soil Moisture
layer feature class. The input raster is the ratser
from which we need to extract values at the points of interest. The output
point feature class is stored in a separate Shape file which can be added to ArcMap.
Figure 10: Extraction of values
to points
The process is repeated for all the six dates and
the LAI raster values and precipitation values are obtained.
8. Results
8.1 Spatial LAI layers.
The results from the exercise are
spatial LAI layers for the dates for which the image was acquired. The spatial
LAI layers generated through the model (Figure 11 & 12) are shown below.
The figures show Fields 17, 21, 32 and 33. Fields 17 and 33 are Corn Fields and
21 and 32 are Soybean Fields.
The figure shows spatial LAI for The value of Leaf Area Index at the
location of the field is shown.
Figure 11:
Spatial LAI for
The figure shows spatial LAI for
Figure 12:
Spatial LAI for
From the above figures it was observed that there
is an increase in Leaf Area Index as we proceed through the growing season. The
process of generating the spatial LAI layer was repeated for all the six dates
for which the Landsat imagery was available.
The figure below shows temporal variation of LAI
through the growing season for both the Corn and Soybean Fields. The plot for
Corn Sites shows a proper pattern for all the dates except 8th July
for which the pattern does not look reliable.
Figure 13: Temporal variation
of LAI on Corn Sites
The plot below shows the pattern of LAI for
Soybean Sites. It is observed that the pattern of LAI is fine for all the dates
except for 8th July when it shows greater value than the LAI on 16th
and 17th July. So the values for 8th July don’t look
reliable for Soybean sites too.
Figure 14: Temporal variation
of LAI on Soybean Sites
8.2
Saturation of NDVI
Saturation of NDVI at high LAI values is well
documented, with such saturation occurring at values substantially below typical
LAI in high productivity sites. Limitations of current methods for deriving
canopy biophysical (LAI) and leaf biochemical constituents, such as
chlorophyll, nitrogen, and water content, limits accuracy when estimating the
significance of photosynthetic processes in regulating some components of the
carbon cycle (Tejad and Ustin, 2001).
A number of studies have shown that the NDVI saturates at LAI of 3-4 [Sellers P.J, 1986]. If a saturation threshold is defined where a
given VI, averaged within LAI bins of width 0.5, reaches 95% of the full range
in binned VI, then NDVI saturates at LAI≈3.5, OSAVI at LAI≈4.0, and
NDWI at LAI≈4.5 (Anderson et al.,
2004).
To observe the saturation of LAI, the NDVI was
plotted against LAI. It was observed that NDVI does show saturation after LAI
attains a value of approximately 3.5, on the Corn sites.
Figure 15: Saturation of NDVI on
Corn Sites
On the Soybean sites, it is observed that the
maximum LAI achieved is nearly 3.5 and hence a pattern is hard to access.
Figure 16: Saturation of NDVI on
Soybean Sites
8.3 Effect
of precipitation on soil Moisture
8.3.1 Effect of Precipitation on Soil Moisture
(0-6 cm Depth)
The
soil moisture data used here resulted from daily measurements of surface (0-6
cm) volumetric soil moisture. The volumetric soil moisture content from soil
specific calibration obtained for 0-6 cm depth was plotted for the Corn and
Soybean sites for 5th and 8th of July 2002. The rainfall
for
Figure 17: Effect of Soil
Moisture on Corn Sites
Figure 18: Effect of Soil
Moisture on Soybean Sites
8.3.2
Effect of Precipitation on Soil Moisture (30 cm
Depth)
Soil profile stations were deployed at four sites
in the Walnut Creek Watershed: Sites 15 and 24 were corn fields and 16 and 23
were Soybean sites. The stations were
instrumented identically to measure soil temperature and moisture profiles as
well as rainfall. All measurements were
made at 10-second intervals and were averaged or totalized over 15 minute
output intervals.
At each site, two pits were dug within a few
meters of the enclosure to a depth of 30 cm (Fig. 19). Pit A was dug between the corn rows with a
clean vertical face perpendicular to rows.
Pit B was adjacent to a row with a clean face 15 cm from the stem of the
row crops. For soybean fields, the two
profiles were established with no regard to row structure. In each pit, soil moisture and temperature
measurements were made at six depths: 2, 5, 10, 15, 20 and 30 cm. Soil moisture was measured at each depth
using a Water Content Reflectometer (WCR), a device
based on time domain reflectometry, in which an
electrical pulse is transmitted down a pair of steel probes, reflected from the
probe ends and subsequently received at the base.
Figure 19 : Schematic diagram of sensor installation (Ref: Data Set Description - Soil Profile
Stations, SMEX 2002)
The soil moisture at 30cm depth measured in pit B
was used from this dataset for further analysis.
The Soil moisture at 30cm depth and rainfall were
plotted for six days of the year i.e. 23rd June, 30th
June, 1st July, 5th July and 8th July . It is
observed from the graphs shown below that the Corn sites exhibit an enhanced
effect of the rainfall on soil moisture at 30 cm depth. The soil moisture
increases after the precipitation event so that the rainfall covers the
depletion in soil moisture due to crop growth. However on the soybean sites, it
is observed that the moisture shows a constant value after the precipitation
event., so it might be concluded that the rainfall just makes up for the soil
moisture consumed for the growth of crops.
Figure 20: Effect of Rain on Soil
Moisture on Field-15 (Corn Site)
Figure 21: Effect of Rain on Soil
Moisture on Field-24(Corn Site)
Figure 22: Effect of Rain on Soil
Moisture on Field-16 (Soybean Site)
Figure 23: Effect of Rain on Soil
Moisture on Field-23 (Soybean Site)
The leaf area index was plotted against soil
moisture for site WC03. The aim was to see a decrease in soil moisture with the
increase in leaf area index. The LAI versus Soil Moisture was plotted for three
dates i.e. 23rd June, 1st and 8th july. In the figure below it is seen that the soil moisture
decreases between 23rd June and 1st July but it shows an
increase from 1st July to 8th July. It was observed that
there was a precipitation event between these two dates that caused the soil
moisture to rise.
Figure
24: Variation of LAI with Soil Moisture on site WC03
8.4
Extraction of slope in percentage from 1/3 arc-second DEM
The
following figure shows the DEM having the same spatial extent as the study
area, as obtained from the seamless server.
Figure 25: Digital Elevation Model for the Study Area
Projecting the DEM - To
perform slope calculations the DEM was projected into the Albers equal area
projection. This was done using the Data Management Tools à Projections and Transformations à Raster à Project Raster in the
toolbox. The
output cell size is set to 10m:
The 1/3Arcsecond DEM downloaded from
the seamless server website was projected
Figure 26: Tool for Projection of DEM
The snapshot below shows the
properties of the DEM.
Figure 27: Properties of DEM
This NED data is at 1/3 arc second spacing so a
cell size 10m was specified. The CUBIC interpolation method was used
because CUBIC refers to the cubic
convolution method that determines the new cell value by fitting a smooth curve
through the surrounding points and this works best for a continuous surface
like topography.
The slope
of the DEM is calculated using Spatial Analyst in the following way,
Spatial Analyst à Surface Analysis à Slope
Figure 28: Slope calculation for DEM
This
results in a grid of slope values in percentage as shown below.
Figure 29: Slope of DEM
The
percent slope for each location is extracted in the same manner as explained
earlier in the report. The values obtained are scaled so that when plotted with
the soil moisture, an observation could be made. It was observed that the slope
of the area varies from 0 to 80 percent.
A vertical bar graph was plotted and it was found
that no pattern was observed when the slope and the surface soil moisture
(0-6cm). The change in soil moisture could not be associated with the
irregularity of the slope. The probable reason might be the discontinuity in
the area of study.
Figure 30: Effect of Slope on
Soil Moisture
9. Summary:
The aim is to look into the relation between Leaf
Area Index that was derived from the remote sensing images and how well they
correlate with the other plant physiological characteristics such as soil
moisture and Vegetation Water Content. Also the effect of precipitation on Soil
moisture and crop growth was observed. It was clear from the analysis that
strong relationships exist between these parameters. The LAI patterns establish
the fact that analysis based on remotely sensed imagery is completely
dependable.
10. Further Research
The further research that I plan to do is
calculation of Evapo-transpiration using Energy
Budget. I also plan to build a predictor set keeping in mind all that I have
done for this project and apply some learning machines for direct estimation of
Soil moisture and ET.
References
and Data Sources
-
Anderson, M.
C., Neal, C. M. U., Li, F., Norman, J. M., Kustas, W.
P., Jayanthi, H., & Chavez, J. (2004). Upscaling ground observations of vegetation water content,
canopy height, and leaf area index during SMEX02 using aircraft and Landsat imagery.
Remote Sensing of Environment. pii:S0034-4257(04)00176-2.
-
-
Crosson,W.L.,Limaye,
A.S.,and Laymon C.A.,
2005,Member, IEEE Parameter Sensitivity of Soil Moisture Retrievals From
Airborne C- and X-Band Radiometer Measurements in SMEX02, IEEE Transactions On Geoscience And Remote
Sensing, 43(12):2842-2853,
-
Document URL: http://nsidc.org/data/docs/daac/nsidc0187_smex_veg_watershed.gd.html
-
Gao, B.-C. (1996).
NDWI—a normalized difference water index for remote sensing of vegetation liquid
water from space. Remote Sensing of Environment, 58, 257– 266.
-
http://nsidc.org/data/amsr_validation/soil_moisture/smex02
-
http://nsidc.org/data/nsidc-0199.html
-
Napolitano R.
et al.(2001). Data Integration for Leaf Area Index Prediction in Function of
Land Cover Change. “Decision Support
System for Sustainable ECOsystem MANagement
in
-
Ray, S.S., Wadhwal,V.K., 2001,Estimation of crop evapotranspiration
of irrigation command area using remote sensing and
-
Sellers, P.
J., Y. Mintz, Y. C. Sud,
and A. Dalcher, A simple biosphere model (SiB) for use within general circulation models, J. Atmos.
Sci., 43, 505-531, 1986.
-
Tejad Zarco a P. J and Ustin S.L.,
Modeling Canopy Water Content for Carbon Estimates from MODIS data at Land EOS
Validation Sites, IEEE, 342-344, 2001