Lab 5

Goal:

The goal of this lab was to gather and interpret spectral reflectance signatures of 12 surface materials using a satellite image of Eau Claire. These reflectance signatures graphed, compared and analyzed. Also different band ratio techniques were used to analyze and map vegetation and ferrous soil in the Eau Claire and Chippewa counties.

Methods:

Part 1: Spectral Signature Analysis

The first task involved plotting the spectral reflectance of twelve common surfaces using a Landsat ETM+ image of the Eau Claire area. The image was brought into Erdas Imagine and a polygon was created over each of the following surfaces:

1. Standing Water

2. Moving water

3. Forest  

4. Riparian vegetation. 

5. Crops

6. Urban Grass

7. Dry soil (uncultivated)  

8. Moist soil (uncultivated)

9. Rock

10. Asphalt highway

11. Airport runway

12. Concrete surface

After each polygon, the selection was brought into Signature Editor where it was labeled and then plotted so that the graph displayed the amount of reflectance there was at each of the six bands. Many observations were made in each individual graph of the different surface. For example, vegetation displayed high reflectance in the infrared bands, but not so much in the visible bands due to the fact the vegetation uses the visible band in photosynthesis and the infrared band is harmful to important proteins in the plant. The variance between dry soil and moist soil was also observed and it was noted that the moist soil displayed less reflectance due to the presence of water in the soil (Figure 1.) All the signatures were graphed together as well (Figure 2.)

Part 2: Resource Monitoring

In section one the vegetation of Eau Claire and Chippewa county were analyzed and mapped by using the normalized difference vegetation index (NDVI). To bring out the vegetation in an image, a formula is used:

NDVI=NIR-Red/NIR+Red

NIR and Red refer to the near infrared and red bands of the image. The NDVI tool in Erdas Imagine was used on a satellite image of the Eau Claire/Chippewa county area. The resulting NDVI map can be found in the results area (Figure 3.)

In the second section, the same process as in the first section was repeated, except the Indices tool was used to observe ferrous soils in the Eau Claire/Chippewa area. This process uses a different formula to highlight the ferrous soils in an image:

Ferrous mineral=MIR/NIR

A map of ferrous soils (Figure 4.) can be found in the results section.

Results:

Figure 1. shows the reflectance of dry soil and moist soil. The moist soil has less reflectance, because it contains water, which readily absorbs electromagnetic energy.

Figure 1.

Figure 2. displays the reflectance of all twelve surfaces observed in the lab.

Figure 2.

Figure 3. is a map of the vegetation in the Eau Claire/Chippewa area. The more intense the color green is, the more dense the vegetation in that area.

Figure 3.

Figure 4. is a map of the distribution of ferrous soil in the Eau Claire/Chippewa area. The lighter the gray, the greater the presence of ferrous soils.

Figure 4.

Sources:

Satellite image is from Earth Resources Observation and Science Center, United States Geological Survey.

Lab 7

Goal:

The objective of this lab was to calculate photographic scales, measure the area and perimeter of a feature and calculate relief displacement, as well as understand stereoscopy and perform orthorectification on satellite images.

Method:

Part One:

In the first section the scale of an aerial photograph of a section of the city of Eau Claire was calculated from given information and using a ruler to manually measure features on the screen. Every variable was converted to inches. The work and results of this section can be found in the “results” section of this post. Another aerial photograph of the city of Eau Claire was calculated, but this photograph was taken from a higher altitude. Every variable was given and converted to feet and the formula: S=f/(H-h) was used to find the scale.

In the second section the “measure” tool in Erdas Imagine was used to find the area and perimeter of a lagoon in Eau Claire. The polygon tool was used to measure area by tracing the perimeter of the lagoon. The area was 35.5957 hectares (92.9 acres). The polyline tool was used to trace the perimeter. The perimeter amounted to 4,100.25 meters (2.55 miles).

In the third section an image of the upper campus of the University of Wisconsin – Eau Claire was analyzed to determine relief displacement. Using a smoke stack that was altered due to the photogrammetric error the relief displacement was calculated in inches. A ruler was used to find the height of the smoke stack and its distance from the principal point. The rest of the variables were given. The answers can be found in the results section of this post.

Part Two:

In this part of the lab a 3D image was created from an elevation model using the Anaglyph tool in Erdas Imagine. A 1-meter spatial resolution image of the City of Eau Claire was used along with a DEM version of the same image to create an anaglyph version of the image, which was saved in a personal stereoscopy folder. Next a digital surface model was combined with the original City of Eau Claire to create another 3D image using the anaglyph tool, but this anaglyph proved to better represent surface features than the DEM anaglyph.

Part Three:

This part of the lab utilized Erdas Imagine Lecia Photogrammetric Suite (LPS) to orthorectify images and create a planimetrically accurate orthoimage of a part of Palm Springs, California. In LPS a new block file was created for this project and it was set up using a polynomial-based pushbroom geometric model and the image spot_pan.img of Palm Springs, CA was brought in. The horizontal reference coordinate system used was UTM Zone 11 with using NAD27 (CONUS) datum and the Clarke 1866 spheroid. The SPOT PAN was filled out in the Sensor Information field to indicate that an image taken by the SPOT satellite was used. Next, the point measurement tool was opened, which is where all of the ground control points (GCP) were placed.

Before placing any GCPS on the spot_pan image, a reference image (xs_ortho) was brought in. Below the two images was a box where cells containing information about the GCPs would appear. Finally, using the Create Point tool, a GCP was placed in the reference image at certain point and then another one in the spot_pan image at the exact same spot. In total 9 GCPs were placed using xs_ortho as the reference image. The exact coordinates of where the GCPs needed to be in both images were provided in the lab instructions and were manually entered into the box where the GCP information cells were. For the last two GCPs NAPP_2m-ortho.img was the reference image. Once all the GCPs were placed the vertical reference source had to be set up. For this, the palm_springs_dem image was used to supply the height information of this orthorectified image. After that, the Type column for all of the GCPs was set as full and the usage column was set as Control.

Next, a second image was added to the block (spot_panb). In LPS seven GCPs were added to spot_panb in the exact same spot as the corresponding GCPs in spot_pan. Looking at the home screen of LPS the two images were overlayed and the shared GCPs were shown. The automatic tie point tool places tie points to create a more accurate placement of the overlayed image (spot_panb) onto spot_pan. 40 tie points were used when running the tool. After the tie points were created triangulation had to be performed to establish mathematical relationships between images, the sensor and the ground. For this the ground point type was set to same weighted values and the X, Y, Z coordinates for set to 15 to ensure that the GCPs were accurate to about 15 meters. The triangulation summary can be found in the results portion of the post.

Finally, the ortho resampling process was started. The resampling method that was used was bilinear interpolation and the output was labeled orthospot_panb. The ortho resampling process was run and the result can be found in the results portion of the post.

Results:

Part One:

This is the work for finding the scale of the first image of Eau Claire.

1.       Actual measurement: 8,822.47 ft.

Ruler Measurement: 2.75 inches.

8,822.47 ft*12=105,869.64 inches.

2.75 inches on the screen represents 105,869.64 inches.

So 105,869.64/2.75= 38,498.05

Scale is 1:38,498

This is the work for the second image of Eau Claire where the formula used was S=f/(H-h) and all of the variables were given.

1.       Formula: S= f/(H-h)

S=scale

f=152mm=.498688ft

H=20,000ft

h=796ft

S=.498688/(20,000-796)

S=.498688/19,204

S= 1:38,509

This is the work for finding the relief displacement using the smoke stack on upper campus of UW - Eau Claire.

1.       Relief displacement=d=hxr/H

Height of object in image= .5 inches

Scale=1:3,209

Height of object=1,604.5 inches

d=1,604.5inchesx11.5inches/47,760 inches

d=.386 inches

The top of the tower should be moved .386 inches towards the principal point.

Part Three:

Below is the final product of placing all the GCPs, adding tie points, triangulation and ortho processing.

Sources:

 National Agriculture Imagery Program (NAIP) images are from United States Department of Agriculture, 2005. 

Digital Elevation Model (DEM) for Eau Claire, WI is from United States Department of Agriculture Natural Resources Conservation Service, 2010.

Lidar-derived surface model (DSM) for sections of Eau Claire and Chippewa are from Eau Claire County and Chippewa County governments respectively. 

Spot satellite images are from Erdas Imagine, 2009. 

Digital elevation model (DEM) for Palm Spring, CA is from Erdas Imagine, 2009.

National Aerial Photography Program (NAPP) 2 meter images are from Erdas Imagine, 2009

Lab 4

Goal:

The objective of this lab was to calculate photographic scales, measure the area and perimeter of a feature and calculate relief displacement, as well as understand stereoscopy and perform orthorectification on satellite images.

Method:

Part One:

In the first section the scale of an aerial photograph of a section of the city of Eau Claire was calculated from given information and using a ruler to manually measure features on the screen. Every variable was converted to inches. The work and results of this section can be found in the “results” section of this post. Another aerial photograph of the city of Eau Claire was calculated, but this photograph was taken from a higher altitude. Every variable was given and converted to feet and the formula: S=f/(H-h) was used to find the scale.

In the second section the “measure” tool in Erdas Imagine was used to find the area and perimeter of a lagoon in Eau Claire. The polygon tool was used to measure area by tracing the perimeter of the lagoon. The area was 35.5957 hectares (92.9 acres). The polyline tool was used to trace the perimeter. The perimeter amounted to 4,100.25 meters (2.55 miles).

In the third section an image of the upper campus of the University of Wisconsin – Eau Claire was analyzed to determine relief displacement. Using a smoke stack that was altered due to the photogrammetric error the relief displacement was calculated in inches. A ruler was used to find the height of the smoke stack and its distance from the principal point. The rest of the variables were given. The answers can be found in the results section of this post.

Part Two:

In this part of the lab a 3D image was created from an elevation model using the Anaglyph tool in Erdas Imagine. A 1-meter spatial resolution image of the City of Eau Claire was used along with a DEM version of the same image to create an anaglyph version of the image, which was saved in a personal stereoscopy folder. Next a digital surface model was combined with the original City of Eau Claire to create another 3D image using the anaglyph tool, but this anaglyph proved to better represent surface features than the DEM anaglyph.

Part Three:

This part of the lab utilized Erdas Imagine Lecia Photogrammetric Suite (LPS) to orthorectify images and create a planimetrically accurate orthoimage of a part of Palm Springs, California. In LPS a new block file was created for this project and it was set up using a polynomial-based pushbroom geometric model and the image spot_pan.img of Palm Springs, CA was brought in. The horizontal reference coordinate system used was UTM Zone 11 with using NAD27 (CONUS) datum and the Clarke 1866 spheroid. The SPOT PAN was filled out in the Sensor Information field to indicate that an image taken by the SPOT satellite was used. Next, the point measurement tool was opened, which is where all of the ground control points (GCP) were placed.

Before placing any GCPS on the spot_pan image, a reference image (xs_ortho) was brought in. Below the two images was a box where cells containing information about the GCPs would appear. Finally, using the Create Point tool, a GCP was placed in the reference image at certain point and then another one in the spot_pan image at the exact same spot. In total 9 GCPs were placed using xs_ortho as the reference image. The exact coordinates of where the GCPs needed to be in both images were provided in the lab instructions and were manually entered into the box where the GCP information cells were. For the last two GCPs NAPP_2m-ortho.img was the reference image. Once all the GCPs were placed the vertical reference source had to be set up. For this, the palm_springs_dem image was used to supply the height information of this orthorectified image. After that, the Type column for all of the GCPs was set as full and the usage column was set as Control.

Next, a second image was added to the block (spot_panb). In LPS seven GCPs were added to spot_panb in the exact same spot as the corresponding GCPs in spot_pan. Looking at the home screen of LPS the two images were overlayed and the shared GCPs were shown. The automatic tie point tool places tie points to create a more accurate placement of the overlayed image (spot_panb) onto spot_pan. 40 tie points were used when running the tool. After the tie points were created triangulation had to be performed to establish mathematical relationships between images, the sensor and the ground. For this the ground point type was set to same weighted values and the X, Y, Z coordinates for set to 15 to ensure that the GCPs were accurate to about 15 meters. The triangulation summary can be found in the results portion of the post.

Finally, the ortho resampling process was started. The resampling method that was used was bilinear interpolation and the output was labeled orthospot_panb. The ortho resampling process was run and the result can be found in the results portion of the post.

Results:

Part One:

This is the work for finding the scale of the first image of Eau Claire.

1.       Actual measurement: 8,822.47 ft.

Ruler Measurement: 2.75 inches.

8,822.47 ft*12=105,869.64 inches.

2.75 inches on the screen represents 105,869.64 inches.

So 105,869.64/2.75= 38,498.05

Scale is 1:38,498

This is the work for the second image of Eau Claire where the formula used was S=f/(H-h) and all of the variables were given.

1.       Formula: S= f/(H-h)

S=scale

f=152mm=.498688ft

H=20,000ft

h=796ft

S=.498688/(20,000-796)

S=.498688/19,204

S= 1:38,509

This is the work for finding the relief displacement using the smoke stack on upper campus of UW - Eau Claire.

1.       Relief displacement=d=hxr/H

Height of object in image= .5 inches

Scale=1:3,209

Height of object=1,604.5 inches

d=1,604.5inchesx11.5inches/47,760 inches

d=.386 inches

The top of the tower should be moved .386 inches towards the principal point.

Part Three:

Below is the final product of placing all the GCPs, adding tie points, triangulation and ortho processing.

Sources:
 National Agriculture Imagery Program (NAIP) images are from United States Department of Agriculture, 2005.
Digital Elevation Model (DEM) for Eau Claire, WI is from United States Department of Agriculture Natural Resources Conservation Service, 2010.
Lidar-derived surface model (DSM) for sections of Eau Claire and Chippewa are from Eau Claire County and Chippewa County governments respectively.
Spot satellite images are from Erdas Imagine, 2009.
Digital elevation model (DEM) for Palm Spring, CA is from Erdas Imagine, 2009.
National Aerial Photography Program (NAPP) 2 meter images are from Erdas Imagine, 2009

Lab 3

Lab 6

Goal:

The goal of this lab was to demonstrate geometric correction, which is a very important preprocessing activity so that the extracted information of that image is reliable. Geometric correcting eliminates most geometric distortion in an image so that individual pixels are in their proper planimetric position. 

Methods:

Part One:

Part one consisted of geocorrecting an image of the Chicago region to a contour map of the same area in Erdas Imagine by placing ground control points (GCPs) in the same places on the image and the map. A first order polynomial equation was used, so four GCPs were placed (an extra was added for better accuracy). When placing the GCPS, one is first placed on the image and another is placed on the same exact spot on the map. In order to make sure the GCPs are in the exact same spot, one must observe the Root Mean Square (RMS) error located to the right of each GCP on the bottom panel and the control point error (CPE) in the bottom right corner of the screen. These errors shows how close the GCPs are to matching exactly. An ideal RMS error is 0.5 or lower, but in this section 2.0 or lower was acceptable. By zooming into each GCP they could each be altered until the CPE was below 2.0. The result is a geometrically corrected image.

Part Two:

In this part a slightly distorted image of Sierra Leone was rectified with another reference image of Sierra Leone. This image was geocorrected using the same methods as in part one, but it used a 3^rd order polynomial instead, so 9 GCPs are required, but 12 total were added to prevent wrap tool error. In completion of the georectification the bilinear interpolation resampling method was used. The result is a nearly identifcal geometrically corrected image of Sierra Leone to that of the reference image of Sierra Leone.

Results:

Below is the geometrically corrected image of the Chicago region in part one. For unknown reasons the control point error would not appear in the lower right-hand corner of the screen, but the individual RMS error of each GCP is displayed and each one is below 2.0.

Below is the geometric correction of the Sierra Leone region. The corrected image is on the left and the reference image is on the right. By looking at the bottom right corner of the screen the CPE is displayed and is less .5, meaning that the image is geometrically accurate.

Lab 2

Lab 5

Goal:

LiDAR is an extremely using way of remotely sensing features. In this lab LiDAR data was used to process various surface and terrain models and create intensity images. All the LiDAR point clouds are formatted in LAS files.

Methods:

Part One of the lab consisted of copying the LAS files to a personal folder.

Part Two: Generate a LAS dataset and explore lidar point clouds with ArcGIS

In this part, as a GIS manager, I am working on a project for the city of Eau Claire, WI using LAS files acquired from the city. First, the LAS files were brought into ArcGIS, which is the program that is used throughout this lab. They were arranged into an LAS dataset. Once they are brought in, the statistics of each file can be observed. To make sure the data were accurate, the elevation of certain points can be compared to Eau Claire’s actual elevation. To ensure accuracy, the XY coordinate system was set to NAD 1983 HARN Wisconsin CRS Eau Claire (US Feet) and the vertical coordinate system was set to NAVD 1988 US Feet. A shapefile of Eau Claire was added to confirm that the data was spatially located correctly. 

Once the data were brought in, the LiDAR data was observed using different LAS data surfaces. Elevation, slope and contour were closely looked at. Looking at these surfaces with all the different filters using returns (All, ground and first) gave different perspectives also. Another useful tool is the Profile View tool, which allows the user to select an area of the image to view separately in either a 3D or a 2D view.

Part Three: Generating LiDAR derivative products.

Section One:

In this section a digital surface model with first return (DSM), digital terrain model (DTM) and hillshades of both were created. To determine spatial resolution the average nominal pulse spacing was determined. The spatial resolution was determined to be 2 meters. The LAS to Raster tool is used to create the models and natural neighbor was used to filled in any voids. A hillshade tool was used to create a hillshade which enhanced the DSM and DTM models. The images are revealed in the results portion. The difference between the DSM and the DTM models is that DSM uses the first return LiDAR points to display surface features, such as houses and trees, as well as the ground. The DTM model uses ground return points which better display elevation and terrain of the bare Earth.

Section Two:

In this section a LiDAR intensity image was created using almost the exact same method as DTM and DSM. Intensity imagery is created using first return points and the natural neighbor method was used again. Intensity images help interpret and classify LiDAR masspoints.

Results:

This image is slope LiDAR data.

This is contoured LiDAR data.

This is a 2D profile of a bridge using the LiDAR data.

Using the 3D profile tool one can view the bridge from different angles.

The following is the DSM, which shows suface features such as houses and trees.

 The following is the DTM, which shows ground features and reveals more information about the terrain.

Lab 1

Lab 4

Goal:

The goal of this lab is to use Erdas Imagine to properly subset an area of interest, create a higher spatial resolution for an image, reduce haze in an image, link an image viewer to Google Earth, mosaic images and detect binary changes.

Methods:

Part 1: Subsetting with the use of an inquire box

In this section two methods are used to delineate an area of interest (AOI). Image eau_claire_2011.img was opened in Erdas Imagine. Next an Inquire Box was opened under raster tools. This inquire box was placed over the Eau Claire/ Chippewa area and a subset image was created and saved in my personal lab 4 folder.

Above is the subsetted area of the Eau Claire/Chippewa area. This next method is more commonly used for areas that are not perfectly square or rectangular.

 Result: 

By adding a shape file of the AOI it provides a more exact spot. The shape file is georeferenced so that it covers the exact intended AOI and because the image is subsetted, it is in the same shape as the Chippewa and Eau Claire counties.

Part 2: Image fusion

In this section, a higher spatial resolution image will be created from a course resolution image by using a panchromatic version of that image. By merging the resolutions, a clearer picture can be made of the first image.

Result:

The Panchromatic image is of 15 meter resolution and the Reflective image is 30 meter resolution. When merging the images, the Multiplicative method was used and the Nearest Neighbor resampling technique was used. Although it is not immediately clear from the image above, the image on the left has a higher spatial resolution than the image on the right. It is more obvious when zoomed in.

Part 3: Simple radiometric enhancement techniques

This part of the lab dealt with reducing haze in one image.

Result:

The process of reducing the haze of the image (left) was simply to use the haze reduction tool in Erdas Imagine. The output image is on the right and is much clearer than the original image.

Part 4: Linking image viewer to Google Earth

Linking a viewer to Google Earth can be advantageous for collecting training data for image classification for recent images. Google Earth uses a high resolution satellite image that can be useful when interpreting one’s own image.

Results:

By simply uploading an image of Eau Claire and connecting to Google Earth it is easy to find the same place on Google Earth. Clicking Match GE to View will find the exact location in the image. From there it is easy to analyze and make comparisons by linking and syncing the two views.

Part 5: Resampling

In this part an image was resampled, which means that its pixel size was changed so that the pixel size was reduced (resampled up). Two methods of resampling were used; nearest neighbor and bilinear interpolation.

Results:

The above image shows bilinear interpolation (right). This image has a resolution of 15x15m, while the original image (left) has a resolution of 30x30m.

Part Six: Image Mosaicking

Image mosaicking is used when a study area is larger than a single satellite image. When there are more than one images that intersect, but cover the study area they can be combined together to form a single satellite image. There are a few was this process can be carried out in Erdas Imagine.

Section One: Mosaic Express

Mosaic Express is an easy way to mosaic images, but the outcome may not be as good.

Section Two: MosaicPro

MosaicPro is a more sophisticated way to mosaic images. In this mosaicked image the color flows better between the images.

Results:

MosaicPro takes longer to create a mosaicked image, but it produces a better result.

Part 7: Binary change detection

In this section, two different images of the same area of Eau Claire County are looked at, but the images were taken at different times. By looking at the change in brightness values of pixels, one can detect changes in the landscape.

Section 1: Creating a difference image

Image differencing is one way to detect binary change. In this section only one pair of layers (4) is going to be processed. After combining the images, looking at the histogram is the best way to detect change. Since the areas of change are normally located at the tail of the histogram the tails need to be cut off. In determining the cutoff point for this histogram the rule of thumb threshold of mean+1.5 standard deviation was used.

The cutoff value turned out to be 72.3766.

Section Two: Mapping pixel change using a spatial modeler

Using a spatial modeler, two images the 1991 image was subtracted from the 2011 image to create a new image. In this new image a threshold of mean+(3xstandard deviation) was used to determine the change/no change threshold. Another spatial model was created in order to create an image that shows all the pixels above the change/no change threshold that was just calculated. The image that is created from this model shows only the areas in Eau Claire county that were changed, which makes it difficult to read.

By overlaying the ec_91-11bvis image over the original 1991 image the change in pixel can be detected. The portions in green in the image below shows the change in pixels between the images.

Results: