Climate Scenarios for Managing Grasslands in the Great Plains
This summer, our group has utilized Python and Google Colab to analyze historical data from various points in the Great Plains in order to project how annual net primary productivity (ANPP) will be affected by various climate scenarios. We were able to create reproducible code that analyzes data downloaded from Climate Toolbox to create meaningful graphs that can be utilized to project future ANPP scenarios using MACA Future Climate Projections.
We want to acknowledge that the CU Boulder campus resides on the traditional territories of the Ute, Arapaho, and Cheyenne nations and the point location that we gathered our data from resides on traditional Cheyenne and Arapaho nations.
Projected climate variability and change affects the flow of resources from natural systems on which they depend upon. The U.S. Great Plains are valued for a variety of resources, such as agriculture (livestock grazing and crops), wildlife habitats, and other ecosystem services. It is important to project how the changing climate can affect these ecosystems since varying temperatures and precipitation levels may result in food source scarcity for humans as grazing livestock and crops may be unable to adapt to the changing environment.
This summer, our group has utilized Python and Google Colab to analyze historical data from various points in the Great Plains in order to project how annual net primary productivity (ANPP) will be affected by various climate scenarios. We were able to create reproducible code that analyzes data downloaded from Climate Toolbox to create meaningful graphs that can be utilized to project future ANPP scenarios using MACA Future Climate Projections. This project was done through the Earth Data Science Lab Program under the guidance of:
- William Travis: Professor at CU Boulder researching natural disasters and climate adaptation
- Prasad Thota: Masters student in Civil Engineering at CU Boulder who develop the Grasslands Productivity and Climate App that was helpful for data downloading
- Imtiaz Rangwala: Climate scientist with the North Central Climate Adaptation Science Center whose previous work on Species Assessment utilizes Climate Toolbox
The U.S. Great Plains
The U.S. Great Plains region spans more than 180 million acres from Canada to Texas. The ecology of the Great Plains is diverse, mostly due to the large size of this region. The productivity of the grasslands is essential since these ecosystems are a home to a diverse range of organisms including key crops in commercial agriculture as well as endangered species. Over the past century, the overgrazing of domestic stock has led land managers to work towards creating a balance between domestic stock and native species. Projecting the changing climate’s effect on these ecosystems is essential for land management to begin planning for the most efficient ways to keep these grasslands thriving.
For our analysis, we focused on three points located within the U.S. Great Plains. Our initial Python code that we wrote focused on a point selected within the Central Plains Experimental Range located in Weld County, Colorado. We focused most of our time this summer analyzing the results from this location since it is in a central region within the Great Plains. We also selected a northern point in Bismarck, North Dakota and a southern point in Bushland, Texas so that we could not only test the reproducibility of our Python code, but also to see how similar the data projections are in various regions and climates.
Figure 1. Map of the point location chosen through put the Great Plain in the United States. The star indicates our primary location that the analysis was done on.
We began with two different databases that provided us with historical data contained in .csv files. One of the databases we used is called Climate Toolbox. This database allowed us to input any point location across the Great Plains and download a .csv file for the desired climate variable (precipitation, PET, temperature, etc.) called Grid MET. This file could then be read into Colab using the Pandas package.
Another database that is key in our project is the GrassCast tool. This tool provided a massive .csv file (over 400,000 lines of data points!) that contained the ANPP, spring precipitation, and many other values that would provide us with the data needed to create our desired plots.
Once our data was downloaded to our shared drive folder, we utilized Matplotlib to generate graphs that compared how PET, precipitation, and temperature related to ANPP using historical data. The most important graph created from the historical data (Figure 2) was “PET - Precipitation” vs. ANPP. “PET-Precipitation” or the water deficit is important since it considers both temperature (temperature relates directly to evapotranspiration) and precipitation. We then used the NumPy package to generate a linear regression line and wrote code that neatly displayed this equation on the graph. This equation for the line of best fit is important in how we created future projection graphs.
Utilizing the MACA Future Projection files, we then plugged in the projected data points into the historical line equation in order to create future projection graphs. We selected four climate scenarios provided from CMIP 5 (Climate Model Intercomparison Project) which is a project sponsored by the World Climate Research Programme. We decided to use these four scenarios since they each provide a different projection to what the future climate will be in terms of temperature and precipitation. Lastly, we created one time series graph with all four climate models comparing the water deficit (PET - Precip) to the ANPP. This is the graph we used the most in our analysis.