Progress 09/15/23 to 09/14/24

Outputs
Target Audience:To date, this project has engaged with both initially proposed audiences and new audiences: Initial Audiences This project initially targeted academic researchers and graduate students in Year 1. This has included ongoing training and mentoring of graduate students while fostering collaboration with researchers in the US and Canada by PI Kendall and PI Winter. Key efforts included formal classroom and lab instruction on AI, data science, and agronomy related to this project. Other informal educational experiences have included idea sharing across aligned researchers and their students, the exploring of potential partnerships, and the refining of methods through these interactions. These target audiences are critical to the project's success, as they directly advance its merit and impact by sharpening its disciplinary focus, identifying evolving knowledge gaps in the field, and situating the project within broader geographic and disciplinary contexts. These comprehensive efforts across these target audiences ensures that the project's outcomes will be at the forefront of the discipline and will have produced at its highest academic level possible. New Audiences In Year 1, this project expanded its target audience to include stakeholders such as land managers, representatives from the alternative energy sector, and state and federal government officials. Key efforts have included leading a multi-day research meeting in New Mexico where PI Smidt discussed the premise, ongoing efforts, and possible outcomes of this project with an experienced audience focused on farmland protection and agricultural viability. Members were particularly engaged by forces driving irrigation migration and what that means for the future of US farming, energy resources, and food supplies. PI Smidt has also presented this project to Google's Regenerative Agriculture Group, a climate lead at Amazon's Whole Foods, and to an audience of global agricultural economists at the ING Future of Food Summit in Texas, among other professionals across private, public, and governmental sectors. Efforts with these added target audiences are particularly valuable for the project because they help to inform the under-utilized nature of irrigation production areas while also focusing outcomes so results are foundationally usable by a wide variety of stakeholders. Changes/Problems:The timetables initially laid out in the proposal were drafted using an earlier anticipated start than this project's start date of September 15, 2023. While this is not expected to have any impact on our ability to complete the project on-time as described by the end of Year 3, some of the timetable tasks are not perfectly aligned with our current activities. For example, the initial proposal had the ANN being constructed by the end of Fall 2024. But because of the slightly later start date, the ANN will be completed in early Spring 2024. Additionally, the proposal initially stated we would use the Land Transformation Model (LTM) to build our Artificial Neural Network (ANN). But since the submission of this project proposal, the LTM has been archived by its creator. We will instead use the Land Change Modeler (LCM) or Modules for Land Use Change Simulations (MOLUSCE) to construct our ANN and create our change forecasts. Both the LCM and MOLUSCE use nearly identical methods to the LTM, so no other meaningful change is expected in the project, its methods, or the initial proposal language. What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest? Nothing Reported What do you plan to do during the next reporting period to accomplish the goals?Over the next year, we will develop the ANN and finalize initial irrigated land use and irrigation enhancement forecasts (Objective #1). We will draft manuscripts for these efforts and continue to construct our GIS data repository for our broader K-12 outreach efforts. These efforts will further contribute to the efforts in Objectives #2-4. We will also focus on completing the irrigation function analysis and drafting a manuscript that describes our findings (Objective #2). In addition, we will start running gridded crop model simulations to assess the potential implications of irrigation changes simulated by the ANN (Objective #1) on agricultural production and water use. We will continue to compile aquifer characteristics related to irrigation pumping depths (Objective #3) that will be fed into the initial estimations for energy demand and ultimately irrigation suitability. Collectively, these efforts will start to uncover our initial under-utilized irrigation zones. As we progress through Objectives #1-3, we will continue to work with our project evaluator to evaluate project impacts, and we will begin to explore how our data products come together to prepare an equity index across our under-utilized irrigation zones. We will continue to train and mentor graduate students while hiring undergraduate students to assist in the data processing, transform the results of this project into course curriculum, and continue to participate in a broad range of activities that share and promote the efforts of this project. We will engage with our broader impact partners (e.g., Scientists in Every Florida School) to leverage our partially processed data products into teaching and learning materials for K-12 teachers and students.

Impacts
What was accomplished under these goals? We have expanded our analysis of crop yield simulations of corn from Partridge et al. (2023) to identify the function of irrigation across the United States, with a focus on the Southeast. Specifically, we are analyzing simulated differences across rainfed and irrigated yields to group areas by three functions: (1) Enabling production where rainfed yields would be zero or extremely low, (2) Dramatically increasing relatively low rainfed yields, (3) Alleviating reductions in typically high rainfed yields during drought. We are exploring both the historical function of irrigation as well as the how the function of irrigation changes by mid- and end-of-century under moderate and high greenhouse gas emissions scenarios. Additionally, we prepped datasets to understand and establish baseline of current spatiotemporal irrigation prediction within the contiguous United States. We collected and collated relevant data products of at least 30-meter resolution in raster format, and the training/testing data used to generate them where available, for both comparison amongst themselves and for use in evaluation of predictions produced from work expected later in this project. The main products collected have included the AIM-HPA product covering the High Plains Aquifer, LanID contiguous United States product, and the IrrMapper product covering the western 17 United States. We further estimated annual crop water need (mm) across CONUS at a 30m resolution. We obtained state- and crop-specific irrigation depths from the USDA Farm and Ranch Irrigation Survey (2013) and the Irrigation and Water Management Survey (2018). Using the Cropland Data Layer (CDL), we masked crop types based on Landsat-Based Irrigation Dataset (LanID) irrigation maps for the corresponding survey years and reclassified CDL crop types according to their reported irrigation depths. This process produced irrigation depth reference maps for the crop years 2013 and 2018. Crop water needs were then estimated as the sum of state- and crop-specific irrigation depths, gridMET precipitation, and POLARIS-derived plant available water (PAW). To extrapolate across years, we applied a linear regression between crop water needs and TerraClimate evapotranspiration (ET) for each crop and state. Lastly, we have continued to refine the development of the Artificial Neural Network (ANN) for irrigated land cover modeling by trialing data attribution from the PAD model to irrigated land use. This has included compiling county-level data to build relationships between estimated irrigated areas from the PAD model to what is captured in other irrigated datasets like LanID. References Partridge, T., Winter, J., Kendall, A. et al. Irrigation benefits outweigh costs in more US croplands by mid-century. Commun Earth Environ 4, 274 (2023). https://doi.org/10.1038/s43247-023-00889-0

Publications