Spring 2007
   GEO327G/386G: GIS & GPS Applications in Earth Sciences

 

 

DELINEATION OF COASTLINE CHANGE AFTER DEBRIS FLOWS AND FLASH FLOODS OF 1999 ALONG LA CORDILLERA DE LA COSTA AT VARGAS, VENEZUELA

 

By Rodolfo Hernandez, may 2007

hernandezrta@mail.utexas.edu

 

INTRODUCTION

 

In December 1999, rainstorms induced thousands of landslides along the Cordillera de la Costa, Vargas, northern Venezuela. Rainfall on December 2-3 totaled 200 millimeters (8 inches) and was followed by a major storm (911 millimeters, or 36 inches) on December 14 through 16. Debris flows and flash floods on alluvial fans inundated coastal communities, caused severe property destruction, and resulted in a death toll estimated at 19,000 people. Because most of the coastal zone in Vargas consists of steep mountain fronts that rise abruptly from the Caribbean Sea, the alluvial fans are the only areas where slopes are not too steep to build. (Torres-Sierra et al, 2001).

The main objective of this project is to delineate a new coastline developed during debris flows of 1999 using satellite image before and after that event occurred. Moreover, it is focused on calculating how much area was added to the coastline after 1999 and also on comparing differences between old data a new higher resolution data.

PROBLEM FORMULATION

Due to the rainfall, debris flows and flash floods created a new coastline along areas where alluvial fans were predominant (fig. 1). The purpose of this study consists of:

1.    Delineate coastline of data from year 1990 and compare with another coastline from year 2000 data.

2.    Calculate new areas from the difference between 1990 and 2000 data.

3.    Use higher resolution data to compare with 2000 data in order to see errors made during digitization process.

 

 

Figure 1. Image downloaded from Google Earth representing the study area.

DATA COLLECTION

 

For this project most of the data was downloaded from https://zulu.ssc.nasa.gov/mrsid/bin/show.pl since this data should be useful for comparing the differences between coastlines before and after 1999 to see the changes occurred. We found data of 1990 and also data of 2000. Other images were downloaded from Google Earth as a picture with format TIFF.

 

MrSID Image of 1990: This is 27.0 Mb Image.

Figure 2. MrSID Image of 1990 showing northern part of Venezuela.

 

MrSID Image of 2000: This is 96.7 Mb Image.

Figure 3. MrSID Image of 2000 showing northern part of Venezuela.

 

Google Earth Images/Digital Globe: most of the images used were downloaded as a TIFF format.

 

Figure 4. Google Earth image showing an alluvial fan at Vargas (Zone 3).

 

DATA PREPROCESSING

 

The MrSID images and TIFF format images do not need any kind of preprocessing before starting to work with them in ArcGIS. Before starting work on the data a review was done in order to check the information contained by them.

Spatial Data Description:

·          Zulu90.sid Raster Dataset – MrSID:

Raster dataset information

Raster format: MrSID

SDTS raster type: Pixel

Number of raster bands: 3

Raster properties

Origin location: Upper Left

Has pyramids: TRUE

Has colormap: FALSE

Data compression type: Wavelet

Display type: pixel codes

Cell information

Number of cells on x-axis: 26591

Number of cells on y-axis: 15407

Number of cells on z-axis: 1

Number of bits per cell: 8

Cell Size

X distance: 28.500000

Y distance: 28.500000

 

·          Zulu_2000.sid Raster Dataset – MrSID:

Raster dataset information

Raster format: MrSID

SDTS raster type: Pixel

Number of raster bands: 3

Raster properties

Origin location: Upper Left

Has pyramids: TRUE

Has colormap: FALSE

Data compression type: Wavelet

Display type: pixel codes

Cell information

Number of cells on x-axis: 53184

Number of cells on y-axis: 39013

Number of cells on z-axis: 1

Number of bits per cell: 8

Cell Size

X distance: 14.250000

Y distance: 14.250000

 

·          GE.tif Raster Dataset:

Raster dataset information

SDTS raster type: Pixel

Number of raster bands: 3

Raster properties

Origin location: Upper Left

Has pyramids: TRUE

Has colormap: FALSE

Data compression type: Run-Length Encoding (ESRI)

Display type: pixel codes

Cell information

Number of cells on x-axis: 961

Number of cells on y-axis: 719

Number of cells on z-axis: 1

Number of bits per cell: 8

Cell Size

X distance: 3.115522

Y distance: 3.115522

 

ARCGIS PROCESSING

 

The following processing was applied within three different zones were the alluvial fans are located. However, we are going to present just an example from one zone processing.

 

1.    Spatial References: System Coordinate used for this project was NAD_1983_UTM_Zone_19N. The steps to do this were as follow:

·        Add the data to ArcCatalog.

  • Right-click on the file to be referenced and choose Properties (Figure 5).

 

 

Figure 5. Selecting Properties of Raster Dataset.

 

·        Edit Spatial Reference (Figure 6).

Figure 6. Raster Dataset Properties.

 

·        Assign Coordinate System to the data (figure 7).

 

Figure 7. Spatial Reference NAD_1983_UTM_Zone_19N.

 

 

2.    Clipping Raster: There many ways to clip a raster but to work over the study area was necessary to create a shapefile to determine maximum and minimum X and Y coordinates. This area was used to clip raster data as follow:

  • Create a new shapefile in ArcCatalog (Figure 8).

 

Figure 8. Creating a Shapefile.

 

  • Open new shapefile in ArcMap, add raster data and create a rectangle for the study area (figure 9).
  • Select maximum and minimum of X and Y coordinates from shapefile (figure 10).

 

Figure 9. ArcMap view of shapefile rectangle area and raster data.

 

Figure 10. Shapefile maximum and minimum of X and Y coordinate.

 

  • Open Data Management Tools/Raster/Clip. Select raster to be clipped and put the maximums and minimums in the rectangle area (figure 11).

 

 

Figure 11. Clipping raster data.

Figure 12. Raster data 1990 clipped.

Figure 13. Raster data 2000 clipped.

 

 

3.    Creating a Feature Class: Digitizing coastlines implies creating a feature class. Thus a coastline polyline feature was created to delineate coastline for 5 alluvial fans within the study area (see figure 13) and 3 new polygon areas were created to have better limited area of the alluvial fans.

·        Right-click on the My Geodatabase, select "New", then "Feature Class..." (figure 14).

·        Select Feature Class Type (figure 15).

 

 

 

Figure 14. Creating a Feature Class.

 

Figure 15. Selecting a Feature Class Type.

4.    Delineating Coastline: Once created feature class objects we started to create coastline within three major areas where alluvial fans are. First of all, three areas were selected then a coastline was digitized for data of 1990 (figure 16). Later the same coastline was edited for 2000 data (figure 17).

 

Figure 16. Coastline within AREA 1 for 1990 data.

 

Figure 16. Coastline within AREA 1 for 2000 data keeping old coastline.

 

5.    Creating a Topology for Coastline: Before creating coastline area polygons, it is useful to clean the lines of errors that will corrupt polygon creation.

·        In ArcCatalog, right-click on the Coastline feature dataset, select "New", then "Topology" (figure 17).

Figure 17. Creating a New Topology.

 

·        If after validate the feature we get something as figure 18, we do not need to fix errors.

Figure 18. Topology Validated.

 

 

6.    Making Coastline Polygons: Once all topology errors are fixed, Coastline polygons can be generated.

·        In ArcCatalog, right-click on the Coastline feature dataset and select "New", then select "Polygon Feature Class From Lines..."; the Polygon Feature Class From Lines window opens (figure 19). After doing this we are going to obtain polygons as figure 20. From figure 20 we can see area values of the polygons as figure 21 showed just clicking identify buttom.

Figure 19. Making Polygon Feature Class from Lines.

 

Figure 20. Polygons of Coastline within Zone 1.

 

Figure 21. Identify Results for calculated shape area.

 

 

7.    Inserting TIFF image: To have a better constrained with data digitized we added some tiff images downloaded from Google Earth to do more accurate new coastline after 1999. The steps to do this are as follow:

·        Georeferencing image:

-         Copy Image in the Folder.

-         Open Georeferencing toolbar.

-         Fit image to Display.

-         Georeference image.

-         Rectify georeferenced image.

After doing all this we are going to have an image georeferenced as showed figure 22.

Figure 22. High resolution image georeferenced.

 

8.    Delineating Coastline from a better resolution image: We can correct differences between coastline for image of 2000 whose cell size is 14.25 m and compare with a new coastline made from DigitalGlobe image whose cell size is 3.11 m. To do this we just repeated step 4 and we can see a new and more accurate coastline (figure 23).

 

Figure 23. Coastline using a higher resolution image.

 

RESULTS

From data interpretation and analysis we get the following results:

 

  • Map 1: New alluvial fan areas at the coastline after debris flows of 1999 at Vargas, Venezuela. Zone 1.(figure 24)
  • Map 2: New alluvial fan areas at the coastline after debris flows of 1999 at Vargas, Venezuela. Zone 2.(figure 25)
  • Map 3: New alluvial fan areas at the coastline after debris flows of 1999 at Vargas, Venezuela. Zone 3.(figure 26)
  • Map 4: Coastlines superposed within a high resolution data. Zone1.(figure 27)
  • Map 5: Coastlines superposed within a high resolution data. Zone2.(figure 28)
  • Map 6: Coastlines superposed within a high resolution data. Zone1.(figure 29)
  • Table with values of areas calculated in three different alluvial fan zones along coastline at Vargas, Venezuela.

 

 

Figure 24. Map 1: New alluvial fan areas at the coastline after debris flows of 1999 at Vargas, Venezuela. Zone 1.

 

Figure 25. Map 2: New alluvial fan areas at the coastline after debris flows of 1999 at Vargas, Venezuela. Zone 2.

 

Figure 26. Map 3: New alluvial fan areas at the coastline after debris flows of 1999 at Vargas, Venezuela. Zone 3.

 

Figure 27. Map 4: Coastlines superposed within a high resolution data. Zone1.

 

Figure 28. Map 5: Coastlines superposed within a high resolution data. Zone2.

Figure 29. Map 6: Coastlines superposed within a high resolution data. Zone3.

 

Zone

Polygon

Area (m2)

Area (Km2)

1

1

22225.16

0.02222516

2

19634.56

0.01963456

3

14778.74

0.01477874

4

8426.78

0.00842678

5

67003.95

0.06700395

6

43673.39

0.04367339

2

1

164422.11

0.16442211

2

87988.67

0.08798867

3

31066.74

0.03106674

4

22616.11

0.02261611

3

1

55825.02

0.05582502

Total

 

537661.23

0.53766123

 

Table 1. Values calculated from polygons in ArcCatalog for new areas added after debris flows of 1999 in three different zones at Vargas, Venezuela.

 

Figure 30. Polygon areas from ArcCatalog for Zone1.

 

CONCLUSIONS

 

Summarizing from the final results we can conclude the following:

 

  • Maps show a notable difference between coastlines from data of 19990 to coastline from data of 2000.
  • Resolution differences between 1990 and 2000 data do not allow digitizing a better coastline due to the fact that every pixel of 1990 data represents four time pixel of 2000 data.
  • Digitizing coastline within low resolution images can cause human errors as a consequence of cell size, sometimes it is difficult to difference if a cell is part of the sea of it is part of land.
  • New area added to the coastline after debris flows of 1999 in the 3 zones is about 0.53 Km2.
  • Using a higher resolution image allows digitizing an accurate coastline. This is evident when we compare data of DigitalGlobe with data of 2000.

 

REFERENCES

Torres-Sierra et al.(2001). “Natural Hazards on Alluvial Fans: The Venezuela Debris Flow and Flash Food Disaster”. USGS.