Utilizing GRACE data over the Colorado River Basin
Visualizing one of the biggest water losses for the region within ArcGIS Pro.
Biggest Water Loss
From March 2005 to June 2013 , the Colorado River Basin lost 5.7 cubic kilometers (4.6 million acre-feet) of water per year or more than 47 cubic kilometers (3 million acre-feet) over the 100-month study period. 1 The majority of the water loss is attributed to groundwater pumping to irrigate agricultural fields. Castle et al., analyzed both GRACE data and groundwater monitoring wells in the Upper and Lower Colorado River Basins and found that groundwater may compromise a far greater fraction of Basin water use than previously recognized, in particular during drought, and that its disappearance may threaten the long-term ability to meet future allocations to the seven Basin states. 2
What is GRACE?
The Gravity Recovery and Climate Experiment (GRACE) satellites originally launched on March 17, 2002 in collaboration with both NASA and DLR the German Aerospace Center. 3 It is a set of twin satellites that makes detailed measurements of Earth's month-to-month changes of its gravity field and the underlying mass changes related to Earth's water reservoirs over land, ice, and the ocean. GRACE ended its data collection in July 2017, but was succeeded by the GRACE Follow-On (GRACE-FO) mission which launched on May 22, 2018. 4 Just like GRACE did, GRACE-FO measures mass changes in the polar ice sheets, mountain glaciers, total water storage on land, deep ocean currents, as well as glacial isostatic adjustment and impacts from major earthquakes.
Study Site
The Colorado River Basin is an important source of water for many cities in the Southwest region of the United States. The drainage basin is about 246,000 square miles which includes Arizona, Colorado, New Mexico, Nevada, Utah, Wyoming and parts of California. 5 The Colorado River helps supply 1,450 miles of water to major cities, with its headwaters in Colorado and Wyoming and winding down into Mexico. 5
Visualization within ArcGIS Pro
ArcGIS Pro is the latest Geographical Information System (GIS) software offering the capabilities of visualizing geospatial data. GRACE data comes in the file format called Network Common Data Form (NetCDF). NetCDF stores multidimensional scientific data variables such as temperature or time. 6 ArcGIS Pro offers the ability to visualize various scientific data formats with multidimensional analysis and geoprocessing tools.
Tutorial utilizing NetCDF data
The tutorial below is based on a use case of 'Water Storage Anomalies Over The Colorado River Basin' in the GRACE-FO Level-3 Data Product User Handbook, which can be accessed below. Understanding the data and following the use case can be found on pages 29-31 and 43-48.
Step 1: Download Level-3 global mass concentration Terrestrial Water Storage anomalies from GRACE and GRACE-FO and Scaling Factor. 7 Data for GRACE Tellus can be accessed from Earthdata Search . Data for the Scaling Factor file can be downloaded from the GRACE data product page under 'Data Access.'
Step 2: Download the Colorado River Basin boundary shapefile.
Step 3: To open NetCDF data formats within ArcGIS Pro, you must convert the data set into a Raster Layer for this tutorial.
Within the geoprocessing toolbox, navigate to ‘Multidimensional Tools’ under ‘NetCDF’ select ‘Make NetCDF Raster Layer.’ When inputting the NetCDF file, you have the option to select which variable to view.
For the GRACE Tellus dataset, the variable we will be observing is ‘lwe_thickness’ which is Liquid Water Equivalent Thickness. For Scale Factor, the variable defaults to the only variable available in this case ‘scale_factor.’
Resulting visualization of the GRACE Tellus dataset in ArcGIS Pro.
Resulting visualization of the Scale Factor dataset in ArcGIS Pro.
Step 4: The GRACE Tellus dataset provides observations over time. Activating the time dimension is a simple step through the settings.
Right-click the dataset in the ‘contents’ pane and select properties. Select the ‘Time’ section and there will be an option for ‘Layer Time.’ Since we know this dataset has measurements over time it will give the option to activate the time scale. Once activated, you can explore the dataset through a desired time of choice.
Scaling factor, also called gain factors, offers an enhanced spatial resolution of the GRACE dataset. To do this, we have to multiply the GRACE Tellus data with the scaling factor data which results in a new spatial resolution of 0.5 degrees x 0.5 degrees for mascons and 1 degree x 1 degree for harmonic-based grids. 8
Step 5: To make this calculation we will open the ‘Raster Calculator’ tool. It can be found under ‘Image Analyst’ then select ‘Map Algebra.’ The main thing to consider when calculating the scaling factor is the time dimension does not transfer to the newly created layer. Navigating back to the data properties, set a time that you would like to explore before calculating. For this example we will set June 2016.
Resulting visualization from Raster Calculator LWE Thickness and Scale Factor calculation.
Now we will clip the area to the Colorado River Basin area.
Step 6: Under the ‘Spatial Analyst’ tools select ‘Extraction’ then we will use the ‘Extract by Mask’ tool. This tool clips an area of interest within the raster cells. In this case, we will use the Colorado River Basin shapefile to clip the data to our region of interest.
Tutorial on the Biggest Water Loss
To further expand the utilization of GRACE data and ArcGIS Pro, let's replicate the workflow using an automated model to explore the extensive time series from the 'Biggest Water Loss' , March 2005 - June 2013.
As previously mentioned, when running the 'Raster Calculator' tool, time does not transfer over from the Tellus dataset to the new layer. This can get tedious, especially when exploring a larger time series.
One workaround is to utilize the 'ModelBuilder' which allows the building of geoprocessing workflows. Instead of inputting each variable one by one and running the tool individually for each process, ModelBuilder allows you to connect geoprocessing tools and variables in order of processes and outputs you desire. 9
Step 1: In this instance, we will run the two geoprocessing tools mentioned in the tutorial above. Inputting the 'Raster Calculator' tool, multiplying layers 'LWE Thickness' and 'Scale Factor.' Next, connect the output layer to the 'Extract by Mask' tool to clip to the Colorado River Basin shapefile.
The main thing to consider before running this workflow is selecting the desired time of the 'LWE Thickness' before executing. In this example we will focus on a one-month average for each layer. With your selected time frame, the output layer must be manually inputted with the desired layer name to keep things organized. I opted for naming each new layer by its selected time frame.
With the new layers created, stitching them together is the next step by creating a mosaic dataset. This will help in managing and displaying the newly created raster layers.
Step 2: 'Create Mosaic Dataset' tool will create an empty mosaic dataset to a file geodatabase. Next, you will need to add the new layers with the 'Add Rasters To Mosaic Dataset' tool. These two tools will help in creating a seamless raster dataset of the new layers created.
With our newly created dataset, we would like to access the time series to visualize the rate of change of groundwater in the Colorado River Basin.
As previously mentioned, you can go to the properties of the new layer within the contents pane to activate the time series. When activated you notice the time series does not visualize any of the layers created based on the specific time frames. We will need to manually set the time frame.
When the mosaic dataset is created, a 'Boundary' and 'Footprint' layer is also created. Since raster layers do not have attribute tables, the 'Footprint' layer compliments as an attribute table to that raster layer. The attribute table within 'Footprint' lists the raster layers that were added to the mosaic dataset. Since each layer's output name corresponds to a desired researched time frame this will be easier to recognize.
Step 3: Open the attribute table for the 'Footprint' layer, select 'Add' in the 'Fields' section. We will create a new field called 'TimeStamp' and the data type associated with the field will be 'Date'. This will allow us to input the date that is associated with the layer.
Step 4: Once the 'TimeStamp' field is created, you can select the calendar as shown in the image and pick the date that is associated with the layer.
Once all dates are selected with their associated layer, don't forget to click save in the 'Edit' tab.
Step 5: Next, select the layer's properties of the Colorado mosaic dataset. Go to Time and set Layer Time to 'Each feature has a single time field.' Then select the Time Field as the newly created field 'Time Stamp.' From there you can specify the time extent of the layers.
Once the data is set we can visualize it with many symbology options. Since we are working with raster data you have the option to resample pixel size and use the stretch function.
The 'Resample Function' changes the raster pixel size from Nearest Neighbor, Bilinear Interpolation, and Cubic Convolution. 10 This must be done with caution as it can alter the interpretation of the data even though certain resampling types might be visually appealing.
The 'Stretch Type' function uses the statistics from the raster to enhance the image. 11
You also have many color ramp options to choose from that will fit your research needs.
Conclusion
As you can see above, the results of the 'Biggest Water Loss' speaks for itself. The GIF created loops from March 2005 - June 2013 showcasing water loss of each month.
This tutorial can be replicated using any other area of interest. Also, most tools mentioned can be used to visualize any NetCDF data. ArcGIS Pro also offers the ability to create 3D Voxel layers or Multidimensional Raster layers with many different scientific data formats.
For more information please visit PO.DAAC and the PO.DAAC StoryMap Collection page to learn about other datasets.