LIFE ENVIRONMENT STRYMON
Ecosystem Based Water Resources Management to Minimize Environmental Impacts from Agriculture Using State of the Art
Modeling Tools in Strymonas Basin
LIFE03 ENV/GR/000217
Task 2. Monitoring Crop Pattern, Water quality and Hydrological Regime
SHYLOC implementation in Strymonas basin
Volume 3
Period covered by the report: from 1/9/2005 to 31/8/2006
Date of submission of the report: 31/8/2006
The present work is part of the 4-years project: “Ecosystem Based Water Resources Management to Minimize Environmental Impacts from Agriculture Using State of the Art Modeling Tools in Strymonas Basin” (contract number LIFE03 ENV/GR/000217). The project is co-funded by the European Union, the Goulandris Natural History Museum - Greek Biotope/Wetland Centre (EKBY), the Prefecture of Serres – Directorate of Land Reclamation of Serres (DEB-S), the Development Agency of Serres S.A. (ANESER S.A.) and the Local Association for the Protection of Lake Kerkini (SPALK).
This document may be cited as follows: Hatziiordanou, Eleni. 2006. SHYLOC implementation in Strymonas basin - Volume 3. Greek Biotope/Wetland Centre (EKBY). Thermi, Greece. 42 p.
PROJECT TEAM
Greek Biotope/Wetland Centre (EKBY)
Dimitrios Papadimos (Project Manager) Iraklis Chalkidis (Agricultural Engineer)
Antonios Apostolakis (Geographic Information System Expert) Eleni Hatziiordanou (Geographic Information System Expert)
Prefecture of Serres – Directorate of Land Reclamation of Serres (DEB-S)
Christos Metrzianis (Scientific Coordinator) Athanasios Taousianis (Scientific Coordinator)
i
CONTENTS
CHAPTER 1: INTRODUCTION ........................................................................... 2
CHAPTER 2: METHOD ........................................................................................ 3
2.1 General ............................................................................................................ 3
2.2 Image acquisition............................................................................................. 4
2.3 Image preprocessing ........................................................................................ 5
2.4 Water quantity monitoring network.................................................................. 6
2.5 Third application of SHYLOC to the Strymonas River Basin........................... 9
2.5.1 Application of SHYLOC on SPOT Scene 9..............................................13
2.5.2 Application of SHYLOC on SPOT Scene 10............................................18
CHAPTER 3: RESULTS .......................................................................................23
3.1 Ditch indexes ..................................................................................................23
3.2 Results of water depth estimation....................................................................28
3.2.1 Measured water depths.............................................................................29
3.2.2 Satellite-derived water depths...................................................................29
3.3 Deviations between measured and satellite-derived water depths ....................32
3.4 Deviations between measured and satellite-derived water widths ....................37
CHAPTER 4: DISCUSSION AND CONCLUSIONS...........................................41
REFERENCES.......................................................................................................42
2
SHYLOC IMPLEMENTATION IN STRYMONAS BASIN VOLUME 3
CHAPTER 1: INTRODUCTION
The Strymonas River basin has significant values, in terms of biodiversity,
agriculture, tourism and economic assets. The irrigation and drainage of the
Strymonas River basin is elaborated through a dense network of canals and ditches.
Agricultural activities constitute the main threat to surface waters and groundwater in
the basin. One of the main problems occurring in the area is the unsustainable
management of surface waters with significant consequences: loss of water,
salinisation of agricultural soils, alterations in the hydroperiod of Lake Kerkini, high
nutrients concentration, sea intrusion.
The Life Strymon project aims at the sustainable management of surface waters and
groundwater in Strymonas River Basin. To accomplish that, a monitoring system for
the calibration of the modeling system of Strymonas River has been designed
(Chalkidis et al. 2004). A sufficient network of stage boards, sensors and sampling
stations for monitoring the quantitative and qualitative water flow properties is being
established. Spatial and temporal variations of the impacts of agricultural activities on
surface waters and groundwater are assessed using State of the Art Modeling Tools.
State of the art methods are also used to elaborate optimum water resources
management plans.
The objective of this work is to examine the possibility of using a satellite image
analysis method to provide water levels at certain positions in surface waters of the
Strymonas river basin. In particular, satellite images were used to estimate the water
widths of certain ditches and these values were correlated with the recordings of water
depth instrumentation at specific cross sections and on specific dates. The expected
output is to apply linear correlations in order to estimate water level at certain
positions of the drainage and irrigation network of the study area and use these
estimations to calibrate the modeling tools.
3
CHAPTER 2: METHOD
2.1 General
To assess water depth in surface waters of the river basin, 12 instruments measuring
water depths were installed at 12 cross sections of the irrigation and drainage network
of the Strymonas River basin. Satellite images that cover the basin area, were
processed and water widths of the parts of the irrigation and drainage network where
instrumentation is installed, were derived. These water widths were correlated with
ground measurements of water depths from different dates, in order to test the
accuracy of the SHYLOC results and examine the possibility of using them for the
calibration of the MIKE SHE/MIKE 11 models. Satellite-derived water depths were
also estimated using the SHYLOC water width calculations, taken into account that
most ditches have trapezoid cross sections and their inclinations and bases are known.
Ten satellite images have been processed so far, covering the Strymonas river basin.
The images were acquired at different dates covering spring and summer periods of
2004, 2005 and 2006.
Both satellite image processing and the calculation of the ditch water depths were
performed using SHYLOC (System for HYdrology using Land Observation for model
Calibration). The functionality of this software has been described in “SHYLOC
Implementation in Strymonas Basin-Volume 1”.
The satellite images were firstly geo-processed and then used to estimate the water
width of rivers, through SHYLOC. The first application of SHYLOC was performed
to satellite imagery of 2004 and is described in the first SHYLOC technical report
(Hatziiordanou and Papadimos 2004). SHYLOC application to the remaining satellite
images of 2004 and those of 2005, was implemented in 2005 and is described
thoroughly in the second SHYLOC technical report (Hatziiordanou 2005). The
present report describes the SHYLOC application to two satellite images that were
acquired in summer of 2006.
The processing stages and the results of the third SHYLOC application are described
in the following chapters.
4
2.2 Image acquisition
The selection of the appropriate satellite imagery was based on its spatial resolution,
its coverage and its ability to distinguish water from land. SPOT (Satellite Pour l’
Observation de la Terre) imagery outweighed other space and airborne remote sensing
data on all the above matters. A further advantage of SPOT is that its images can be
acquired under exact acquisition programming request.
Thus, 10 panchromatic SPOT images that cover the whole study area were requested
to be acquired at certain time periods (Table 2.2.1). The programming request
included detailed descriptions and technical requirements of the imagery needs, such
as survey period, survey area and repeated acquisitions at specified time intervals for
crop monitoring. The image acquisition was programmed for the spring and summer
of 2004, 2005 and 2006, in order to avoid cloud and ice coverage. More precisely,
five sets of images were purchased, each including two scenes, one from the
northeastern part and one from the southwestern part of the basin. Most of them were
acquired by SPOT-4 and some by SPOT-5, depending on the time availability of the
satellite’s pass at the requested time period. A minimum radiometric correction was
already performed to them by ‘SPOT Image France’ (level of processing: 1A).
Table 2.2.1 Technical information and exact date and time of acquisition of the
satellite images.
Set Scene Satellite Instrument Resolution Acquisition date
Acquisition time
1 1 SPOT 4 HRVIR 2 10 m 23 April 2004 09:44:54
1 2 SPOT 4 HRVIR 1 10 m 29 April 2004 09:29:25
2 3 SPOT 4 HRVIR 1 10 m 25 May 2004 09:29:34
2 4 SPOT 4 HRVIR 2 10 m 14 June 2004 09:45:09
3 5 SPOT 5 HRG 2 10 m 14 July 2004 09:41:40
3 6 SPOT 5 HRG 2 10 m 25 August 2004 09:34:04
4 7 SPOT 5 HRG 2 10 m 22 June 2005 09:43:44
4 8 SPOT 4 HRVIR 2 10 m 9 July 2005 09:46:14
5 9 SPOT 5 HRG 2 10 m 17 June 2006 09:18:50
5 10 SPOT 5 HRG 2 10 m 7 July 2006 09:34:14
5
2.3 Image preprocessing
SPOT images of the fifth set were georeferenced to the Greek Geodetic Reference
System EGSA�871. “Image to map” and “image to image” coordinate transformations
were applied using well defined ground control points from topographic maps (scale
1: 50.000) and the first order polynomial method. The bilinear interpolation was
selected for the image resampling, because of its better spatial accuracy and its
suitability for SHYLOC application against other available methods (Hatziiordanou
and Papadimos, 2004). Figures 2.3.1 and 2.3.2 display the resulted images. The
images had to be reprojected to the Word Geodetic System (WGS84)2, in order to be
used as input in SHYLOC.
Figure 2.3.1 Scene 9 (June 17, 2006) that covers the NW part of the study area,
georeferenced to EGSA�87.
1 The Greek Geodetic Reference System (EGSA�87) is a Tranverse Mercator projection that uses the
spheroid of GRS80 and a scaling factor of 0.9996. It is the main reference system that is used in Greece
and it measures in meters.
2 The World Geodetic System 1984 (WGS84) is a geocentric geodetic datum used for the determination
of geographical coordinates developed by the United States Department of Defence.
6
Figure 2.3.2 Scene 10 (July 7, 2006) showing the SE part of the study area,
georeferenced to EGSA�87.
2.4 Water quantity monitoring network
The main objective of the water level monitoring network is to provide an adequate
number of water depth time series for the calibration and validation of the hydraulic
model of the catchment (Chalkidis et al. 2004).
The network includes 12 water level auto-recorders that are established at the inlets
and outlets of either the natural water bodies (e.g. Strymonas River, Lake Kerkini, etc)
or the irrigation and drainage networks in the Strymonas basin (Figure 2.4.1). Four of
them (No 9, 10, 11 and 12) are installed at earthen canals, whereas the remaining 8
(No 1 to 8) are installed at concrete canals.
The 1st water level recorder (No 1) has been established in Strymonas River just
upstream the flow control structure “Ypsilon 1 (Y1)” aiming at the monitoring of
Strymonas inflows into the catchment.
7
Figure 2.4.1 The water quantity monitoring network (yellow spots) and the study area (orange line).
8
Three water level recorders (No 2, 3 and 4) have been established in the upper end of
the canals of “Kentriki dioriga”, “Anatoliki dioriga” and “Ditiki dioriga”, aiming at
monitoring the discharges that are supplied to the 1st irrigation network of Serres plain
and to the region of “Hrisohorafa”.
Water level auto-recorder No 5 has been established about 3 km downstream the
upper end of the canal “2K” that supplies the irrigation networks of Sidirocastro
region.
The 6th water level recorder (No 6) has been established just downstream the flow
control structure “Ypsilon 2 (Y2)” under the bridge of “Enotiki Dioriga” that supplies
with water the 2nd irrigation network of Serres.
The 7th water level auto-recorder (No 7) has been established downstream the
“Ypsilon 3 (Y3)” flow control structure through which water diverts to the “5K”
canal. The canal “5K” supplies with water the 4th irrigation network of Serres and the
“Dimitritsi” irrigation network.
The 8th, 9th and 10th water level auto-recorders (No 8, 9 and 10) are located in
“Belitsa” drainage ditch. Instrument No 8 has been established in “Belitsa” ditch just
before the outlet of “Annageniseos” ditch, aiming at the monitoring of the water flow
at its upper end. Instrument No 9 has been established under the bridge near the
villages “Ano Mitrousi” and “Kato Mitrousi”. The water flow at this point comes
from drainage water from the upstream cultivated areas during the summer while
during the rest period, comes mainly from the upstream torrents. Instrument No 10 is
located under the bridge of village “Skoutari”.
The 11th water level auto-recorder (No 11) has been established at “Agitis” River near
the “Agistas” Railway Station. At this point the net inflow of “Agitis” into the
catchment can be estimated.
Water level auto-recorder No 12 has been established 2 km upstream of Strymonas
outlet into Strymonikos Gulf aiming at monitoring the total runoff of Strymonas
catchment into the sea.
9
2.5 Third application of SHYLOC to the Strymonas River Basin
The third application of SHYLOC involved application to the fifth set of SPOT
images, acquired on summer of 2006 (Table 2.5.1 and Figure 2.5.1).
Table 2.5.1 SPOT images that were used for the third SHYLOC application.
Set Scene Satellite Resolution Band used for the
SHYLOC application Acquisition
date
5 9 SPOT 5 10 m 4 17 June 2006
5 10 SPOT 5 10 m 4 7 July 2006
The 4th spectral band (Short Wave Infrared) of SPOT was used for the SHYLOC
calculations, due to the ability that offers to distinguish water from land, as it takes
advantage of their different reflectance at this spectral range. Figure 2.5.2 displays the
study area (highlighted in orange polyline), the hydrographic network (highlighted in
blue polyline) and the 12 instruments (in red) on Band 4 of the fifth set of images.
The SPOT images were opened at SHYLOC in SceneType and their corner
coordinates were entered in WGS84 Longitude/Latitude. They were then saved in
bitmap format and opened in Shyloc Type mode from the SHYLOC Calculation bar.
The ditch4 data (polyline vector of the hydrographic network in shapefile format)
were transformed from EGSA’87 to WGS84, to overlay the SPOT images.
4 Although instruments were installed in canals, the term “ditch” will be used hereafter, due to its
compatibility to SHYLOC.
10
Figure 2.5.1 Fifth set of SPOT images. Scene 9 at the left (June 17, 2006) and scene 10 at the right (July 7, 2006) displayed in RGB
combination, using the 4th and 2nd spectral bands.
11
Figure 2.5.2 The study area (orange polyline), the hydrographic network (blue polyline) and the 12 instruments (red symbols) on Band 4 of Set
5 (9th SPOT scene at the left and 10th SPOT scene at the right).
12
SHYLOC was applied to 10 of the 12 water level recorders of the water level
monitoring network. It was not applied to instruments 8 and 1, because the 8th water
level recorder could not be detected by the satellites and the 1st is established at an
orthogonal ditch. It should be noted that SHYLOC calculations for water level
recorders No 9, 10, 11 and 12 that are installed at earthen ditches are not accurate,
because these ditches do not have stable cross sections. Nevertheless SHYLOC was
applied to water level recorders No 2, 3, 4, 5, 6 and 7, that are installed at trapezoid
concrete ditches with known inclinations and bases.
For the SHYLOC calculations the land reference value was automatically computed
by the software for each cross section on the images. The pure water reference value
was stable for each image. The water widths of the cross sections that were computed
for each land method were used to estimate the corresponding water depths. The
water depths that were estimated by SHYLOC were compared with the measurements
of water depths from the water level recorders at the exact day and time that the
satellite passed over the instruments.
To identify the ditch carrying pixels and calculate the ditch index and the ditch width,
10 boundaries where drawn around the locations of the 10 instruments, using the
Polyline Editor tool of SHYLOC. Thus the ditch index and wet ditch width
calculations were limited to the ditch carrying pixels located within the defined
boundaries.
The SHYLOC Calculation bar (Figure 2.5.4) was used for raster and vector data
input. Raster data (Image file) had to be in bitmap format, vector data that represents
hydrographic network (Ditch Data) had to be in shapefile format and vector data that
represents the defined boundaries around the instruments (Boundary) had to be in
DXF format.
A slight shift of the images was necessary, in order to optimise the fit between the
satellite image and the vector ditch data. The “Shift/rotate” tool of SHYLOC was used
in order to recalibrate the images.
13
Figure 2.5.4 Input of raster and vector data in Shyloc Type mode, to the SHYLOC
calculation bar.
2.5.1 Application of SHYLOC on SPOT Scene 9
The 9th SPOT Scene that was acquired on 17 June of 2006, covers the NW part of the
study area. Nine water level recorders are installed at ditches around the area covered
by the image (Figure 2.5.5). The 8th and the 1st water level recorders were excluded
from the SHYLOC calculations. Thus, SHYLOC was applied to 7 cross sections
(around water level recorders No 2, 3, 4, 5, 6, 7 and 9).
Image pre-processing included a shifting of the image, in order to optimise the fit
between the satellite image and the vector ditch data. Figure 2.5.6 (up) shows a detail
of the image where the digitized hydrographic network (vector file outlined in red) is
not overlaying well the ditches (shown in black color on the satellite image).
Recalibration was applied using the “Shift/rotate” tool in SHYLOC and the fit
between the image and the overlaying vector file was improved (Figure 2.5.6 down).
14
Figure 2.5.5 The 9 water level recorders (red spot) on SPOT Scene 9 (June 17, 2006).
15
Figure 2.5.6 Up: Detail from the 9th SPOT scene (June 17, 2006) where the digitized
hydrographic network (outlined in red) is not overlaying well the
satellite image (ditches shown in black). Down: The same scene, after its
coordinates have been modified using the shift tool.
To enable faster and easier data processing, the image was cut in smaller areas defined
by selections, using the rectangle selection tool. The areas surround the boundaries
that were drawn around the water level recorders. (Figure 2.5.7).
16
Figure 2.5.7 Detail from the boundaries (in yellow) that were drawn with the Polyline
Editor around the water level recorders 2, 3, 4, 5, 6, 7 and 9 for SPOT
scene 9.
2nd 3rd
4th 5th
6th
9th
7th
17
Following the image pre-processing, the Pure Water Reference Value and the Land
Reference Value were determined.
The Pure Water Reference Value was determined using the rectangular selection
button, by defining an area inside Kerkini Lake. The minimum digital number inside
this area corresponds to the value of the reference water pixel (W) and was found to
be 0 (Figure 2.5.8).
Figure 2.5.8 Selected pure water area and minimum digital number (highlighted at
the pixel selector table) of pixels found inside this area. This minimum
number corresponds to the pure water reference number of the 9th
SPOT scene.
Regarding the calculation of the digital number of the land reference value (Li), the
automatic method was applied to calculate it, using eight methods (described at
section 2.4 of “SHYLOC Implementation in Strymonas Basin-Volume 1”). The main
scope was to examine the effects of the different land calculation methods upon the
statistical relationships between satellite-derived and measured water depths, obtained
on the day of the satellite overpass.
A SHYLOC calculation index was applied to the 7 user-defined boundaries around
the ditches where water level recorders are installed. The calculation was applied 8
Selected pure water area
Highlighted pure water DN at the pixel editor window
18
times for each one of the boundaries. In all cases the same water reference value was
declared at the calculation bar, but the land reference value was dependent on the
selected land calculation method. In some cases, editing of the ditch pixels was
necessary in order to correct some slight errors, caused by inaccurate digitization.
2.5.2 Application of SHYLOC on SPOT Scene 10
The 2nd SPOT Scene was acquired at 29 April 2004. It covers the NW part of the
study area. Four water level recorders are installed at ditches around the area covered
by the image (Figure 2.5.9). The 9th water level recorder was excluded from the
SHYLOC calculations, because it was located at the edge of the image. Thus,
SHYLOC was applied to 3 cross sections around water level recorders No 10, 11 and
12 that were all installed at earthen ditches.
During the SHYLOC application, a slight shift of the image was needed for
improving the overlay with the vector ditch data. Figure 2.5.10 displays an example
where the digitized hydrographic network did not overlay well the ditches (shown in
black color on the satellite image).
19
Figure 2.5.9 The 4 water level recorders (red spots) on SPOT Scene 10 (July 7, 2006).
20
Figure 2.5.10 Up: Detail from the 10th SPOT scene (July 7, 2006) where the digitized
hydrographic network (outlined in red) is not overlaying well the
satellite image (ditches shown in black). Down: The same scene, after
image shifting.
The image was then cut in smaller areas around the 3 water level recorders for faster
and easier data processing (Figure 2.5.11).
21
Figure 2.5.11 Detail from the boundaries (in yellow) that were drawn with the
Polyline Editor around the water level recorders 10, 11 and 12 for
SPOT scene 10.
The Pure Water Reference Value for the 10th SPOT scene was determined by defining
an area inside Kerkini Lake, using the rectangular selection button. The minimum
digital number inside this area that corresponds to the value of the reference water
pixel (W) was found to be 0 (Figure 2.5.12).
10th 11th
12th
22
Figure 2.5.12 Selected pure water area and minimum digital number (highlighted at
the pixel selector table) of pixels found inside this area. This minimum
number corresponds to the pure water reference number of the 10th
SPOT scene.
The automatic method was applied to calculate the digital number of the land
reference value (Li), using all eight available methods. Finally, a SHYLOC
calculation index was applied to the user-defined boundaries around the ditches where
water level recorders are installed. The calculation was applied 8 times for each one of
the 3 boundaries, using the same water reference value (W=0) and the land reference
value that was calculated from 8 land calculation methods. Editing of the ditch pixels
was necessary in order to correct digitization errors.
Selected pure water area
Highlighted pure water DN at the pixel editor window
23
CHAPTER 3: RESULTS
3.1 Ditch indexes
SHYLOC calculation of the “Ditch Index” includes determination of the wet ditch
width inside user-defined boundaries. The “Ditch Index” calculations were applied to
both satellite images, for user-defined boundaries around all water level recorders
situated at certain ditches that were detected at the images. The calculation, which was
made for each one of the water level recorders and for each of the images, was
repeated 8 times, using the pure water reference digital value of each image, and each
time the land reference digital value that was calculated using 8 methods.
All automatic methods of the software were used to calculate the land reference value:
(1) moving average 5x5, (2) moving average 3x3, (3) fixed average inside boundary,
(4) maximum average inside boundary, (5) fixed average in 1 nearest pixel, (6)
maximum average in 1 nearest pixel, (7) fixed average in 2 nearest pixels and (8)
maximum average in 2 nearest pixels. The reason for using all the methods was to
find out which give tighter correlations between the water depths that are derived by
SHYLOC and those derived by the water level recorders.
The results from the SHYLOC calculation applied to the 6th water level recorder for
SPOT scene 9 and the 10th for SPOT scene 10 are displayed in Figures 3.1.1 and
3.1.2 correspondingly. The different colored pixels indicate the percentage of the
pixel water coverage. The deeper the color of the category, the greater the pixel water
content is. Orange color indicates dry pixels. The default value (100 %) of the
maximum pixel water coverage was used for the calculations.
Tables 3.1.2 and 3.1.4 that follow, show the water widths (in meters) that were
derived by SHYLOC applications on both images for cross sections 2, 3, 4, 5, 6, 7, 9,
10, 11 and 12, using as land reference values those that were calculated by 8
automatic methods. The results of the land reference values are shown at Tables 3.1.1
and 3.1.3
24
moving average 5x5 moving average 3x3 fixed average inside boundary
maximum average inside boundary fixed average in 1 nearest pixel fixed average in 2 nearest pixels
maximum average in 1 nearest pixel maximum average in 2 nearest pixels
Figure 3.1.1 The ditch index calculation for the 6th instrument, calculated using the 8
automatic land calculation methods for SPOT scene 9 (June 17, 2006).
25
moving average 5x5 moving average 3x3 fixed average inside boundary
maximum average inside boundary fixed average in 1 nearest pixel fixed average in 2 nearest pixels
maximum average in 1 nearest pixel maximum average in 2 nearest pixels
Figure 3.1.2 The ditch index calculation for the 10th instrument, calculated using the
8 automatic land calculation methods for SPOT scene 10 (July 7, 2006).
26
Table 3.1.1 Land reference values derived by 8 calculating methods for selected boundaries corresponding to instruments 2, 3, 4, 5, 6, 7
and 9 of the 9th SPOT scene (June 17, 2006).
Cross Section
Date of SPOT image
overpass
Moving average 5x5
window
Moving average 3x3
window
Fixed average inside boundary
Maximum value inside boundary
Fixed average in 1 near pixel
Maximum value in 1 near pixel
Fixed average in 2 near pixel
Maximum value in 2
near pixels 2 17 June 161.4 152.44 179 255 159 231 168 255 3 17 June 163.68 146.2 157 255 147 211 162 255 4 17 June 175.083 160.125 203 255 175 255 184 255 5 17 June 132.778 124.444 134 255 127 192 137 255 6 17 June 36.8061 22.497 106 234 36 137 57 162 7 17 June 103.023 91.0747 126 238 98 171 111 204 9 17 June 133.438 107.884 204 255 128 236 160 255
Table 3.1.2 Water widths (m) corresponding to instruments 2, 3, 4, 5, 6, 7 and 9, derived by a SHYLOC application to the 4th band of
the 9th SPOT scene (June 17, 2006). The water reference value used for the calculation is W=0 and the land reference
values are shown in Table 3.1.1.
Cross Section
Date of SPOT image
overpass
Moving average 5x5
window
Moving average 3x3
window
Fixed average inside boundary
Maximum value inside boundary
Fixed average in 1 near pixel
Maximum value in 1 near pixel
Fixed average in 2 near pixel
Maximum value in 2
near pixels 2 17 June 6.22 4.25 9.74 18.74 6.09 16.53 7.81 18.74 3 17 June 7.32 4.78 6.84 16.13 5.20 13.03 7.59 16.13 4 17 June 5.40 3.16 8.78 13.05 5.44 13.05 6.62 13.05 5 17 June 3.29 1.94 3.55 13.06 2.46 9.61 3.99 13.06 6 17 June 59.82 54.52 64.90 66.95 58.07 65.75 61.68 66.20 7 17 June 9.32 5.98 14.65 25.46 8.66 20.69 11.69 23.43 9 17 June 16.61 11.19 24.17 27.18 15.23 26.21 20.03 27.18
27
Table 3.1.3 Land reference values derived by 8 calculating methods for selected boundaries corresponding to instruments 10, 11 and 12
of the 10th SPOT scene (July 7, 2006).
Cross Section
Date of SPOT image
overpass
Moving average 5x5
window
Moving average 3x3
window
Fixed average inside boundary
Maximum value inside boundary
Fixed average in 1 near pixel
Maximum value in 1 near pixel
Fixed average in 2 near pixel
Maximum value in 2
near pixels 10 7 July 111.841 99.3175 147 255 110 187 127 198 11 7 July 99.3651 93.8008 148 255 98 211 106 211 12 7 July 71.336 61.856 152 238 73 127 87 152
Table 3.1.4 Water widths (m) corresponding to instruments 10, 11 and 12 derived by a SHYLOC application to the 4th band of the 10th
SPOT scene (July 7, 2006). The water reference value used for the calculation is W=0 and the land reference values are
shown in Table 3.1.3.
Cross Section
Date of SPOT image
overpass
Moving average 5x5
window
Moving average 3x3
window
Fixed average inside boundary
Maximum value inside boundary
Fixed average in 1 near pixel
Maximum value in 1 near pixel
Fixed average in 2 near pixel
Maximum value in 2
near pixels 10 7 July 10.01 5.85 18.00 28.79 9.51 23.45 13.99 24.56 11 7 July 0.01 0.01 0.04 0.07 0.01 0.06 0.02 0.06 12 7 July 20.00 12.23 44.87 53.04 20.87 40.42 28.03 44.87
28
3.2 Results of water depth estimation
SHYLOC water width calculations were used to estimate the water depths of cross
sections. These depths were then compared with the depths that were measured by the
water level recorders at the exact days that SPOT satellite passed over the study area.
This enabled examining the effects of different land calculation methods upon the
statistical relationships between satellite-derived and measured depths.
Satellite-derived water depths were estimated from the water width calculations, taken
into account that ditches are trapezoids with known inclinations and bases (Figure
3.2.1). Table 3.2.1 shows the inclinations of the side walls (c) and the widths of the
bottom base (b) of the concrete ditches where water level recorders No 2 to 7 are
installed. Water level recorders 9, 10, 11 and 12 are installed at earthen ditches that do
not have stable dimensions.
Figure 3.2.1 Representation of a trapezoid shaped water ditch, filled with water. The
water depth (measured quantity) is symbolized with the letter d, the
water width with w, the inclination of the side walls of the ditch (known
value) with c and the small base of the trapezoid (known value) with b.
Table 3.2.1 Inclinations of side walls and widths of the bottom base of the concrete
covered trapezoid ditches.
Water level recorder ID Inclination of side walls Bottom base width (m) 2 1 1.5 3 1.49 0.6 4 1.55 0.6 5 1.5 2 6 2 23.6 7 1.53 4.8
c
b
d
w
29
3.2.1 Measured water depths
Table 3.2.2 shows the measured water depths for the acquisition date of SPOT scene 9
(June 17, 2006) and Table 3.2.3 for the acquisition date of SPOT scene 10 (July 7,
2006).
Table 3.2.2 Ground measured water depths (in meters) from the 7 water level
recorders that were located at the area covered by SPOT scene 9 at the
exact date and time of the SPOT image acquisition.
Table 3.2.3 Ground measured water depths (in meters) from the 3 water level
recorders that that were located at the area covered by SPOT scene 10 at
the exact date and time of the SPOT image acquisition.
3.2.2 Satellite-derived water depths
The next step was to estimate the water depths at each cross section, by using the
SHYLOC-derived water widths, for those water level recorders that were detected by
SPOT scenes 9 and 10.
Regarding water level recorders 2, 3, 4, 5, 6 and 7 that are located in concrete
trapezoid ditches, the water depth was easily computed. The resulted water depths
corresponding to the 8 different land methods are included in Table 3.2.4.
Instrument Measured water depth on June 17, 2006 (d) 2 0.51 3 0 4 1.048 5 0.776 6 1.789 7 1.81 9 0.82
Instrument Measured water depth on July 7, 2006 (d) 10 1.1 11 0.992 12 0.877
30
Table 3.2.4 Satellite-derived water depths D (in meters) for concrete cross sections
where instruments No 2 to 7 are installed. The measured depths (in
meters) for the day of 17th of June 2006 that the 9th SPOT scene was
acquired are included in Column 2. Columns 3 to 10 show satellite-
derived water depths, estimated from the water widths, for 8 different
land calculation methods.
Regarding those water level recorders (No 9, 10, 11 and 12) that were installed at
earthen ditches, that do not have stable dimensions, water depth could not be easily
computed. Thus, the effectiveness of the land calculation method was tested using the
satellite-derived water widths and the water widths that were estimated from the
measured depths. It should be mentioned that these widths are not very accurate, since
they were estimated graphically, using sketches of the 4 cross sections (Figures 3.2.2
to 3.2.5).
0.00
1.00
2.00
3.00
4.00
5.00
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0
Figure 3.2.2 Cross section at Belitsa Ditch where the 9th water level recorder has
been established (Chalkidis et al. 2004).
Cross Section
Measured depth (d)
D1 Moving average 5x5
window
D2 Moving average 3x3
window
D3 Fixed
average inside
boundary
D4 Maximum
value inside
boundary
D5 Fixed
average in 1 near
pixel
D6 Maximum value in 1 near pixel
D7 Fixed
average in 2 near
pixel
D8 Maximum value in 2
near pixels
2 0.51 2.36 1.38 4.12 8.62 2.30 7.52 3.16 8.62 3 0 5.01 3.11 4.65 11.57 3.43 9.26 5.21 11.57 4 1.048 3.72 1.98 6.34 9.65 3.75 9.65 4.67 9.65 5 0.776 0.97 -0.05 1.16 8.30 0.35 5.71 1.49 8.30 6 1.789 36.22 30.92 41.30 43.35 34.47 42.15 38.08 42.60 7 1.81 3.46 0.90 7.54 15.80 2.95 12.16 5.27 14.25
Water level recorder No9
Bridge at Mitrousi
31
0.001.002.00
3.004.005.006.00
7.008.00
0 5 10 15 20 25 30 35 40 45
Figure 3.2.3 Cross sections at Belitsa Ditch where the 10th water level recorder has
been established (Chalkidis et al. 2004).
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
0 10 20 30 40 50 60 70
Figure 3.2.4 Cross sections of Agitis River at “Agistas” Railway Station where the
11th water level recorder has been established (Chalkidis et al. 2004).
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 10 20 30 40 50 60 70
Figure 3.2.5 Cross sections at Strymonas River where the 12th water level recorder
has been established (Chalkidis et al. 2004).
Bridge at Skoutari Water level
recorder No10
Agiti’s Bridge Water level recorder No11
Water level recorder No12
Ground Level
32
Tables 3.2.5 and 3.2.6 show the satellite-derived water widths and the estimated
widths for water level recorders No 9, 10, 11 and 12 that are installed at earthen
ditches for scenes 9 and 10.
Table 3.2.5 Satellite-derived water widths W (in meters) for earthen cross sections
where instrument No 9 was installed. The estimated widths (in meters)
for the day of 17th of June 2006 that the 9th SPOT scene was acquired
are included in Column 2. Columns 3 to 10 show the satellite-derived
water widths.
Table 3.2.6 Satellite-derived water widths W (in meters) for earthen cross sections
where instruments No 10 to 12 are installed. The estimated widths (in
meters) for the day of 7th of July 2006 that the 10th SPOT scene was
acquired are included in Column 2. Columns 3 to 10 show the satellite-
derived water widths.
3.3 Deviations between measured and satellite-derived water depths
The SHYLOC-derived water depths were thereafter compared to measured depths.
This resulted in a comparison of each land method’s effectiveness.
Figures 3.3.1 to 3.3.6 indicate the variation of depths derived by the 8 land methods
for cross sections 2, 3, 4, 5, 6 and 7 and their comparison with the depths that were
measured by the water level recorders for the day that SPOT scene 9 was acquired (17
Cross Section
Estimated width (w)
W1 Moving average
5x5 window
W2 Moving average
3x3 window
W3 Fixed
average inside
boundary
W4 Maximum
value inside
boundary
W5 Fixed
average in 1 near
pixel
W6 Maximum value in 1 near pixel
W7 Fixed
average in 2 near pixel
W8 Maximum value in 2
near pixels
9 23 16.61 11.19 24.17 27.18 15.23 26.21 20.03 27.18
Cross Section
Estimated width (w)
W1 Moving average
5x5 window
W2 Moving average
3x3 window
W3 Fixed
average inside
boundary
W4 Maximum
value inside
boundary
W5 Fixed
average in 1 near
pixel
W6 Maximum value in 1 near pixel
W7 Fixed
average in 2 near pixel
W8 Maximum value in 2
near pixels
10 21 10.01 5.85 18.00 28.79 9.51 23.45 13.99 24.56 11 26 0.01 0.01 0.04 0.07 0.01 0.06 0.02 0.06 12 47 20.00 12.23 44.87 53.04 20.87 40.42 28.03 44.87
33
June 2006). D1 to D8 correspond to the depths (in meters) derived by SHYLOC using
each one of the 8 land calculation methods.
Figure 3.3.1 Water Depths for cross section 2 that were derived by SHYLOC using
the 8 automatic land calculation methods. The blue discontinuous line
represents the depth (d=0,51 m), measured by the water level recorder
at June 17, 2006 (date of SPOT Scene 9 overpass).
Figure 3.3.2 Water Depths for cross section 3 that were derived by SHYLOC using
the 8 automatic land calculation methods. The blue discontinuous line
represents the depth (d=0 m), measured by the water level recorder at
June 17, 2006 (date of SPOT Scene 9 overpass).
1,38
4,12
8,62
2,3
7,52
3,16
8,62
2,36
0
1
2
3
4
5
6
7
8
9
10
Land calculation methods
Wat
er d
epth
(m)
Series1
D1D2
D3
D4
D5
D6
D7
D8
D cs 2
d=0,51
SPOT Scene 9 (17/6/2006) - Cross section 2
3,11
4,65
11,57
3,43
9,26
5,21
11,57
5,01
-2
0
2
4
6
8
10
12
14
Land calculation methods
Wat
er d
epth
(m)
Series1
D1
D2
D3
D4
D5
D6
D7
D8
D cs 3
d=0
SPOT Scene 9 (17/6/2006) - Cross section 3
34
Figure 3.3.3 Water Depths for cross section 4 that were derived by SHYLOC using
the 8 automatic land calculation methods. The blue discontinuous line
represents the depth (d=1,048 m), measured by the water level recorder
at June 17, 2006 (date of SPOT Scene 9 overpass).
Figure 3.3.4 Water Depths for cross section 5 that were derived by SHYLOC using
the 8 automatic land calculation methods. The blue discontinuous line
represents the depth (d=0,776 m), measured by the water level recorder
at June 17, 2006 (date of SPOT Scene 9 overpass).
1,98
6,34
9,65
3,75
9,65
4,67
9,65
3,72
0
2
4
6
8
10
12
Land Calculation methods
Wat
er d
epth
(m)
Series1
D1
D2
D3
D4
D5
D6
D7
D8
D cs 4
d=1,048
SPOT Scene 9 (17/6/2006) - Cross section 4
1,16
8,3
0,35
5,71
1,49
8,3
-0,050,97
-1
1
3
5
7
9
Land calculation methods
Wat
er d
epth
(m)
Series1
D1
D2
D3
D4
D5
D6
D7
D8
D cs 5
d=0,776
SPOT Scene 9 (17/6/2006) - Cross section 5
35
Figure 3.3.5 Water Depths for cross section 6 that were derived by SHYLOC using
the 8 automatic land calculation methods. The blue discontinuous line
represents the depth (d=1,789 m), measured by the water level recorder
at June 17, 2006 (date of SPOT Scene 9 overpass).
Figure 3.3.6 Water Depths for cross section 7 that were derived by SHYLOC using
the 8 automatic land calculation methods. The blue discontinuous line
represents the depth (d=1,81 m), measured by the water level recorder
at June 17, 2006 (date of SPOT Scene 9 overpass).
30,9234,47
38,08
42,15
36,22
43,35
41,3
42,6
0
5
10
15
20
25
30
35
40
45
50
Land calculation methods
Wat
er d
epth
(m
)
Series1
D1
D2
D3D4
D5
D6
D7
D8
D cs 6
d=1,789
SPOT Scene 9 (17/6/2006) - Cross section 6
3,46
15,8
2,95
12,16
5,27
14,25
7,54
0,90
2
4
6
8
10
12
14
16
18
Land calculation methods
Wat
er d
epth
(m)
Series1
D1
D2
D3
D4
D5
D6
D7
D8
D cs 7
d=1,81
SPOT Scene 9 (17/6/2006) - Cross section 7
36
As indicated by the above Figures (3.3.1 to 3.3.6), the correlations between the
SHYLOC-derived water depths and those depths measured by water level recorders
are not good. The differences appear to be independent of the land calculation method
that was used. Apart from that, at cross section 5 the estimated depth, computed using
the ‘moving average 3x3 window’ land calculation method, was found to be negative.
This depth is not realistic and indicates the ineffectiveness of this method.
Table 3.3.1 that follows, shows which land calculation methods were found to be
most appropriate for calculating water depths at cross sections 2 to 7 for SPOT scene
9. It should be mentioned though, that water depths that were computed using some
land reference methods resulted in significant deviations from measured depths. These
methods were the “maximum value inside boundary”, the “maximum value in 1
nearest pixel” and the “maximum value in 2 nearest pixels”. On the other hand, the
“moving average 3x3 window” method and the “fixed average in 1 nearest pixel”
method, resulted to better correlations between SHYLOC-derived depths and
measured depths. Nevertheless, at some cases (cross sections 3 and 6) none of the
land calculation methods was proved effective.
Table 3.3.1 Effectiveness of land calculation methods on calculating water depths at
cross sections 2 to 7 for SPOT scene 9. Symbol “*” indicates most
appropriate methods and void cells indicate methods that resulted to bad
correlations between SHYLOC-derived and measured depths.
SPOT SCENE 9
Cross section
Moving average
5x5 window
Moving average
3x3 window
Fixed average inside
boundary
Maximum value inside
boundary
Fixed average
in 1 near pixel
Maximum value in 1 near pixel
Fixed average
in 2 near pixel
Maximum value in 2
near pixels
2 * 3 4 * 5 * * * 6 7 * *
37
3.4 Deviations between measured and satellite-derived water widths
The water widths that were computed by SHYLOC for cross sections 9, 10, 11 and 12
were compared with those water widths that were estimated by the measured depths.
The effectiveness of the land calculation methods that were used at SHYLOC was
then examined.
Figures 3.4.1 to 3.4.4 indicate the variance of water widths derived by the 8 land
methods for cross sections 9, 10, 11 and 12 and their comparison with the widths that
were measured by the water level recorders for the days that SPOT scenes 9 and 10
were acquired. W1 to W8 correspond to the widths (in meters) computed by
SHYLOC using each one of the 8 land calculation methods. It is noticeable that bad
correlations exist between SHYLOC-derived water widths and those widths estimated
by measured depths and differences appear to be independent of the land calculation
method.
Figure 3.4.1 Water widths (W) for cross section 9 that were derived by SHYLOC
using 8 automatic land calculation methods. The blue discontinuous
line represents the width (w=23 m), estimated by the water level
recorder measurements at June 17, 2006 (date of SPOT Scene 9
overpass).
11,19
24,17
27,18
15,23
26,21
16,61 20,03
27,18
0
5
10
15
20
25
30
35
Land calculation methods
Wat
er w
idth
(m)
Series1
W1
W2
W3
W4
W5
W6
W7
W8
W cs9
w=23
SPOT Scene 9 (17/6/2006) - Cross section 9
38
Figure 3.4.2 Water widths (W) for cross section 10 that were derived by SHYLOC
using 8 automatic land calculation methods. The blue discontinuous
line represents the width (w=21 m), estimated by the water level
recorder measurements at July 7, 2006 (date of SPOT Scene 10
overpass).
Figure 3.4.3 Water widths (W) for cross section 11 that were derived by SHYLOC
using 8 automatic land calculation methods. The blue discontinuous
line represents the width (w=26 m), estimated by the water level
recorder measurements at July 7, 2006 (date of SPOT Scene 10
overpass).
5,85
18
28,79
9,51
23,45
10,01 13,99
24,56
0
5
10
15
20
25
30
35
Land calculation methods
Wat
er w
idth
(m
)
Series1
W1
W2
W3
W4
W5
W6
W7
W8
W cs10w=21
SPOT Scene 10 (7/7/2006) - Cross section 10
0,060,010,070,040,010,01 0,02 0,06
-5
0
5
10
15
20
25
30
Land calculation methods
Wat
er w
idth
(m
)
Series1
W1 W2 W3 W4 W5 W6 W7 W8
W cs11
w=26
SPOT Scene 10 (7/7/2006) - Cross section 11
39
Figure 3.4.4 Water widths (W) for cross section 12 that were derived by SHYLOC
using 8 automatic land calculation methods. The blue discontinuous
line represents the width (w=47 m), estimated by the water level
recorder measurements at July 7, 2006 (date of SPOT Scene 10
overpass).
Table 3.4.1 shows which land calculation methods were found to be most effective to
calculate water widths at cross sections 9 to 12 for SPOT scenes 9 and 10.
Water depths that were computed using some particular land reference methods
resulted in significant deviations from measured depths. These methods were the
“moving average 5x5 window”, the “moving average 3x3 window”, the “fixed
average in 1 nearest pixel” and the “fixed average in 2 nearest pixels” methods. On
the other hand, the “fixed average inside boundary” method resulted in better
correlations between the SHYLOC-derived depths and the measured depths. For cross
section 11 none of the land calculation methods was proved effective.
12,23
44,87
53,04
20,87
40,42
20
28,03
44,87
10
15
20
25
30
35
40
45
50
55
Land calculation methods
Wat
er w
idth
(m
)
Series1
W1
W2
W3
W4
W5
W6
W7
W8
W cs12
w=47
SPOT Scene 10 (7/7/2006) - Cross section 12
40
Table 3.4.1 Effectiveness of land calculation methods appropriate for calculating
water widths at cross sections 9 to 12 for SPOT scenes 9 and 10. Symbol
“*” indicates the most appropriate methods, whereas void cells indicate
methods that resulted in bad correlations between SHYLOC-derived
widths and estimated widths.
SPOT SCENE 9
Cross section
Moving average
5x5 window
Moving average
3x3 window
Fixed average inside
boundary
Maximum value inside
boundary
Fixed average
in 1 near pixel
Maximum value in 1 near pixel
Fixed average
in 2 near pixel
Maximum value in 2
near pixels
9 * SPOT SCENE 10
Cross section
Moving average
5x5 window
Moving average
3x3 window
Fixed average inside
boundary
Maximum value inside
boundary
Fixed average
in 1 near pixel
Maximum value in 1 near pixel
Fixed average
in 2 near pixel
Maximum value in 2
near pixels
10 * * 11 12 * *
41
CHAPTER 4: DISCUSSION AND CONCLUSIONS
The calculation of the water depth with SHYLOC demonstrates the importance of
using remote sensing data to determine surface water stored in a network of ditches.
Main obstacles that were met during the third application of SHYLOC in Strymonas
River basin were related to digitization inaccuracies that resulted to bad estimation of
some water carrying pixels. The software interface was proved week to deal with the
problem, as it required a lot of manual and time-consuming editing processes.
Most of the land calculation methods that evaluate the digital land reference value
produced week statistical correlations between the satellite-derived depths and the
measured ones. It is possibly caused on the land pixels that correspond to the
vegetation that surrounds those ditches were instrumentation is installed. It is possible
that either dense vegetation may dominate some of the ditches, or deciduous trees
(e.g. aspens) may cover part of them and prevent water detection by the satellite.
At the upcoming time period, all 3 SHYLOC applications to satellite images of the
years 2004, 2005 and 2006 will be compared and their results will be summarized.
The best fit linear equations between satellite-derived water widths and measured
depths will then be selected. The examination of the temporal variation of the pure
water and land reference values and the comparison of the image histograms of the
defined areas at different years or at different months of the same year will enable an
integrated multi-temporal analysis.
Based on the results, the possibility of using SHYLOC for providing water levels at
certain positions in surface waters for the calibration of MIKE SHE/MIKE 11 will be
decided.
42
REFERENCES
Chalkidis, I., D. Papadimos and Ch. Mertzianis. 2004. Water quality and hydrological
regime monitoring network. Greek Biotope/Wetland Centre (EKBY). Thermi, Greece.
21 p.
Hatziiordanou, Eleni. 2005. SHYLOC implementation in Strymonas basin - Volume
2. Greek Biotope/Wetland Centre (EKBY). Thermi, Greece. 88 p.
Hatziiordanou, Lena and D. Papadimos. 2004. SHYLOC implementation in
Strymonas basin - Volume 1. Greek Biotope/Wetland Centre (EKBY). Thermi,
Greece. 52 p.
Khudhairy Al D.H.A, V. Hoffmann and C. Leemhuis. 2001. SHYLOC user manual,
EUR 19745 EN, European Commission.
Khudhairy Al (D.H.A.), C. Leemhuis, V. Hoffmann & I.M. Shepherd (JRC), J.R.
Thompson, H. Gavin & D. Gasca Tucker (UCL), G. Zalidis & G. Bilas (AUT), H.
Refstrup Sørenson & A. Refsgaard (DHI), D. Papadimos (EKBY) and A. Argentieri
(ESA-ESRIN). 2001. SHYLOC final report, EUR 19755 EN, European Commission.
Shepherd, I., G. Wilkinson and J. Thompson. 2000. Monitoring surface water storage
in the north Kent marshes using Landsat TM images. International Journal of Remote
Sensing. Volume 21, no 9: 1843–1865.