Lakes – Master of Spatial Analysis

Monitoring Water Level Changes Using High Spatial and High Temporal Resolution Satellite Imagery

Author: Menglu Wang

Geovis Project Assignment @RyersonGeo, SA8905, Fall 2019

Introduction

The disappearing of the once world’s fourth largest lake, Aral Sea, was a shocking tragedy to the world, not only just the shrinkage of lake volume from 1,093.0 km3 in 1960 to 98.1 km3 in 2010 ( Gaybullaev et al., 2012), but also, the rate of shrinkage. Impacts on environment, local climate, citizen’s health, and agriculture are irreversible. This human made disaster could have been prevented in some degree if close monitoring of the lake was made and people are more educated about the importance of ecosystem. One efficient approach to monitor lake water level changes is the utilizing of satellite imagery .The spreading of free high spatial and high temporal resolution satellite imagery provides excellent opportunity to study water level changes through time. In this study, spatial resolution in 3 and 5 meters and temporal resolution as high as 3 days per visit PlanetScope Scene Satellite Imagery are obtained from Planet website. Iso-Cluster Unsupervised Classification in ArcGIS Desktop and Animation Timeline in ArcGIS Pro are used. Study area is set to Claireville Reservoir and 10 dates of imagery starting from April to late June are used to study water level changes.

Data Acquisition

To download the satellite imagery, a statement of research interest needed to be submitted to Planet Sales personal on their website (https://www.planet.com/). After getting access, go on typing in the study area and select a drawing tool to determine an area of interest. All available imagery will load up after setting a time range, cloud cover percentage, area coverage, and imagery source. To download a imagery, go select a imagery and click “ORDER ITEM” and items will be ready to download on the “Orders” tab when you click on your account profile. When downloading a item, noticing that there is a option to select between “Analytic”, “Visual”, and “Basic”. Always select “Analytic” if analysis will be made on the data. “Analytic” indicating geometric and radiometric calibration are already been made to imagery.

Methodology

ArcGIS desktop is used to implement classification and data conversion. Following after, ArcGIS Pro is used to create a animated time slide. Steps are list below:

After creating a file geodatabase and opening a map, drag imagery labeled letter ending with “SR” (surface reflectance) into the map .
Find or search “Mosaic To New Raster” and use it to merge multiple raster into one to get a full study area (if needed).
Create a new polygon feature class and use it to cut the imagery into much smaller dataset by using “Clip”. This will speed up processing of the software.
Grab “Image Classification” tool from Customize tab on top after selecting “Toolbars”.
On “Image Classification” toolbar, select desired raster layer and click on “Classification”. Choose Iso Cluster Unsupervised classification. Please see Figure 1. for classified result.
Identify classes that belong to water body. Search and use “Reclassify” tool to set a new value (for example: 1) for classes belong to water body, leave new value fields empty for all the rest of classes. Check “ Change missing values to NoData” and run the tool. You will get a new raster layer contain only 1 class: water body as result (Figure 2. and Figure 3.).
Use “Raster to Polygon” tool to convert resulted raster layer to polygons and clean up misclassified data by utilizing editor tool bar. After select “Start editing” from Editor drop down menu, select and delete unwanted polygons (noises).
Use resulted polygons to cut imagery in order to get a imagery contain water bodies only.
Do the above process for all the dates.
Open ArcGIS Pro and connect to the geodatabase that has been using in ArcGIS Desktop.
Search and use “Create Mosaic Dataset” tool to combine all water body raster into one dataset. Notes: Select “Build Raster Pyramids” and “Calculate Statistics” in Advanced Options.
After creating a mosaic dataset, find “Footprint” under the created layer and right click to open attribute table.
Add a new field, set data type as “text” and type in dates for these water body entries. Save edited table.
Right click on the layer and go to properties. Under time tab, select “each feature has a single time field” for “Layer Time”, select the time field that just has been created for “Time Field”, and specify the time format same as the time field format.
A new tab named “Time” will show up on first line of tabs in the software interface.
Click on the “Time” tab and specify “Span”. In my case, the highest temporal resolution for my dataset is 3 days, so I used 3 days as my “Span”.
Click the Run symbol in the “Play Back” section of tabs and one should see animated maps.
If editing each frame is needed, go to “Animation” tab on the top and select “Import” from tabs choose “Time Slider Step”. A series will be added to the bottom and waiting to be edited.
To export animated maps as videos, go to “Movie” in “Export” section of Animation tabs. Choose desired output format and resolution.

Figure 3. Reclassified satellite imagery

Conclusion

A set of high temporal and high spatial resolution imagery can effectively capture the water level changes for Claireville Reservoir. The time range is 10 dates from April to June, and as expected, water level changes as time pass by. This is possibly due to heavy rains and flood event which normally happens during summer time. Please see below for animated map .

Reference

Gaybullaev, B., Chen, S., & Gaybullaev, D. (2012). Changes in water volume of the Aral Sea after 1960. Applied Water Science, 2(4), 285–291. doi: 10.1007/s13201-012-0048-z

Visualizing Freshwater Resources: A Laser Cut Model of Lake Erie with Water Volume Representations

Author: Anna Brooker

Geovisualization Project Assignment @RyersonGeo SA8905, Fall 2018

Freshwater is a limited resource that is essential to the sustenance of all life forms. Only 3% of the water on earth is freshwater, and only 0.03% is accessible on the surface in the form of lakes, streams, and rivers. The Great Lakes, located in Southern Ontario and along the US border, contain one fifth of the surface freshwater. I wanted to visualize this scarcity of freshwater by modelling Lake Erie, the smallest of the Great Lakes. Lake Erie is 6th largest freshwater lake in the world, but is has the smallest water volume out of the Great Lakes. I decided to create a laser cut model of the lake and use water spheres to represent its proportion of the world’s surface freshwater resources. I used the infographic from Canadian Geographic for reference.

Process:

Retrieve bathymetric imagery and import into ArcScene
Generate contours lines for every 20m of depth and export them each into individual CAD files
Prepare the CAD files in an Adobe Illustrator layout file to optimize them for laser printing
Paint and assemble the laser cut layers
Create spheres out of clay to scale with the model

The following images show the import of the bathymetric imaging and contour retrieval:

The bathymetry data used was collected in 1999 by the National Oceanic and Atmospheric Association and comes in a raster file format. They were retrieved from Scholar’s Geoportal. I used a shapefile of the Lake Erie shoreline from Michigan’s GIS Open Data as a mask to clip the raster imaging to only the extent of the lake surface. I then created 20m contours from the raster surface. I exported each of the 3 contour vectors into individual shapefiles. These were added to the scene and exported again as CAD files to be able to manipulate them in Adobe Illustrator and prepare them on a template for laser cutting.

The screenshots above show the template used for laser cutting. The template was downloaded from the Hot Pop Factory homepage. Hot Pop Factory is the service I used for laser cutting the plywood layers. I used their templates and arranged my vector files to reflect the size I want the model to be, 18″x7″. I added the rectangles around each contour to ensure a final product of a rectangular stacked model. I then sent this to the Factory for cutting. The photos below show what I received from Hot Pop.

Lake Erie is incredibly shallow with maximum depth of 64m. In order to show the contours of the lake I needed to exaggerate the depth. Limited by the thickness of the materials available to me, the final model had an exaggerated depth of approximately 130% at its deepest point. The final result of this exaggeration allowed me to create three layers of depth to Lake Erie and make it more visually engaging. I included as a part of my model a flat cut out of Lake Erie, which is what the model would have looked like if I had not exaggerated it.

The water volume spheres were created using a material called porcelain clay. This air dry medium has a slightly translucent finish. I stained the clay with blue oil paint so that it would intuitively represent water. The size of the spheres is based on the information in the Canadian Geographic infographic linked in the introduction to this tutorial. The diameter of the spheres was made to scale with the scale bar on the models. A limitation with this model is that the scale bar only refers to the lateral size of the lake and spheres, and does not refer at all to the depth of the model.

The photos above show the final product. The photo on the right shows the scale bar that is included on both parts of the model. I painted the interior layers in blue, the top two layers in the same shade. The third layer was slightly darker, and the deepest layer was the darkest shade of blue. I chose to paint the layers in this way to draw attention to the deepest part of the lake, which is very small area. I attached the layers together using wood glue and laid them beside each other for display. I painted the 3D and 2D models in slightly different hues of blue. The 2D model was made to better match the hue of the water spheres to visually coordinate them. I wanted the spheres to be distinct from the 3D model so that they would not be interpreted as being representative of the water volume of an exaggerated model.

3D Printing Canadian Topographies

by Scott Mackey, Geovis Project Assignment @RyersonGeo, SA8905, Fall 2016

Since its first iteration in 1984 with Charles Hull’s Stereo Lithography, the process of additive manufacturing has made substantial technological bounds (Ishengoma, 2014). With advances in both capability and cost effectiveness, 3D printing has recently grown immensely in popularity and practicality. Sites like Thingiverse and Tinkercad allow anyone with access to a 3D printer (which are becoming more and more affordable) to create tangible models of anything and everything.

When I discovered the 3D printers at Ryerson’s Digital Media Experience (DME) lab, I decided to 3D print models of interesting Canadian topographies, selecting study areas from the east coast (Nova Scotia), west coast (Alberta), and central Canada (southern Ontario). These locations show the range of topographies and land types strewn across Canada, and the models can provide practical use alongside their aesthetic allure by identifying key features throughout the different elevations of the scene.

The first step in this process was to learn how to 3D print. The DME has three different 3D printers, all of which use an additive layering process. An additive process melts materials and applies them thin layer by thin layer to create a final physical product. A variety of materials can be used in additive layers, including plastic filaments such as polylactic acid (PLA) (plastic filament) and Acrylonitrile Butadiene Styrene (ABS), or nylon filaments. After a brief tutorial at the DME on the 3D printing process, I chose to use their Lulzbot TAZ, the 3D printer offering the largest surface area. The TAZ is compatible with ABS or PLA filament of a 1.75 mm diameter. I decided on white PLA filament as it offers a smooth finish and melts at a lower temperature, with the white colour being easy to paint over.

The next step was to acquire the data in the necessary format. The TAZ requires the digital 3D model to be in an STL (STereoLithography) format. Two websites were paramount in the creation of my STL files. The first was GeoGratis Geospatial Data Extraction. This National Resources Canada site provides free geospatial data extraction, allowing the user to select elevation (DSM or DEM) and land use attribute data in an area of Canada. The process of downloading the data was quick and painless, and soon I had detailed geospatial information on the sites I was modelling.

geogratis — GeoGratis Geospatial Data Extraction

One challenge still remained despite having elevation and land use data – creating an STL file. While researching how to do this, I came across the open source web tool called Terrain2STL on a visualization website called jthatch.com. This tool allows the user to select an area on a Google basemap, and then extracts the elevation data of that area from the Consortium for Spatial Information’s SRTM 90m Digital Elevation Database, originally produced by NASA. Terrain2STL allows the users to increase the vertical scaling (up to four times) in order to exaggerate elevation, lower the height of sea level for emphasis, and raise the base height of the produced model in a selected area ranging in size from a few city blocks to an entire national park.

The first area I selected was Charleston Lake in southern Ontario. Being a southern part of the Canadian Shield, this lake was created by glaciers scarring the Earth’s surface. The vertical scaling was set to four, as the scene does not have much elevation change.

Once I downloaded the STL, I brought the file into Windows 10’s 3D Builder application to slim down the base of the model. The 3D modelling program Cura was then used to further exaggerated the vertical scaling to 6 times, and to upload the model to the TAZ. Once the filament was loaded and the printer heated, it was ready to print. This first model took around 5 hours, and fortunately went flawlessly.

Cape Breton, Nova Scotia was selected for the east coast model. While this site has a bit more elevation change than Charleston Lake, it still needed to have 4 times vertical exaggeration to show the site’s elevations. This print took roughly 4 and a half hours.

Finally, I selected Banff, Alberta as my final scene. This area shows the entrance to Banff National Park from Calgary. No vertical scaling was needed for this area. This print took roughly 5 and half hours.

Once all the models were successfully printed, it was time to add some visual emphasis. This was done by painting each model with acrylic paint, using lighter green shades for high areas to darker green shades for areas of low elevation, and blue for water. The data extracted from GeoGratis was used as a reference in is process. Although I explored the idea of including delineations of trails, trail heads, roads, railways, and other features, I decided they would make the models too busy. However, future iterations of such 3D models could be designed to show specific land uses and features for more practical purposes.

3D models are a fun and appealing way to visual topographies. There is something inexplicably satisfying about holding a tangible representation of the Earth, and the applicability of 3D geographic models for analysis should not be overlooked.

Sources:

GeoGratis Geospatial Data Extraction. (n.d.). Retrieved November 28, 2016, from http://www.geogratis.gc.ca/site/eng/extraction

Ishengoma, F. R., & Mtaho, A. B. (2014). 3D Printing: Developing Countries Perspectives. International Journal of Computer Applications, 104(11), 30-34. doi:10.5120/18249-9329

Terrain2STL Create STL models of the surface of Earth. (n.d.). Retrieved November 28, 2016, from http://jthatch.com/Terrain2STL/