Source: CALIFORNIA STATE POLYTECHNIC UNIV submitted to NRP
IMPROVING SUSTAINABILITY OF SPECIALTY CROP AGRICULTURAL SYSTEMS THROUGH GEOSPATIAL DIGITAL AGRICULTURE IN CALIFORNIA
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
ACTIVE
Funding Source
AFRI COMPETITIVE GRANT
Reporting Frequency
Annual
Accession No.
1027847
Grant No.
2022-67023-36150
Cumulative Award Amt.
$299,911.00
Proposal No.
2021-10179
Multistate No.
(N/A)
Project Start Date
Jan 1, 2022
Project End Date
Dec 31, 2025
Grant Year
2022
Program Code
[A1601]- Agriculture Economics and Rural Communities: Small and Medium-Sized Farms
Recipient Organization
CALIFORNIA STATE POLYTECHNIC UNIV
3801 WEST TEMPLE AVENUE
POMONA,CA 91768
Performing Department
Geography & Anthropology
Non Technical Summary
There is a critical need for research that will contribute to enhancing farming sustainability through improving farmers' access to information and decision-making under future climate conditions. The proposed research brings the benefits of climate change research to specialty crop farmers. The study of agricultural suitability incorporates multiple environmental data at diverse spatial scale. Deployment of new sensors and platforms hold the potential to transform the use of data in agricultural operations. However, methods developed for a type of data input needs to be adapted to benefit from new data. For agricultural suitability this implies the adaptation of using satellite data to unmanned aerial vehicles (UAVs). The overall objective of the proposed research is to develop a pipeline to model suitability using UAV and satellite data under current and future climate conditions.This project will make use of satellite and UAVs data and machine-learning algorithms to identify agricultural suitability for three specialty crops. We will examine grapes, strawberries, and citrus production in California. After identifying agricultural suitability at current climate conditions, we will examine the potential effects of future climate conditions in the agricultural suitability. The analysis of future agricultural suitability will allow the estimation of suitability gain or loss due to environmental change.The current scientific understanding predicts that severe and prolonged drought will become more frequent as a consequence of the changing climate. Climate change will impact food production worldwide and California's specialty crops are not exceptions. This research project aims tocontributingto farmers and industry stakeholders with suitability information that is actionable at the level. The ultimate goals of the project are to increase sustainability for small and medium-sized specialty crop farms (SMSCF) by improving preparedness for future climate conditions by mapping land suitability using unmanned-aerial-vehicles (UAV) and satellite-based remote sensing data and by training a new generation of agriculturalists that are ready to apply data-intensive methods to agricultural production.
Animal Health Component
25%
Research Effort Categories
Basic
75%
Applied
25%
Developmental
(N/A)
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
2057210206050%
2057210202050%
Knowledge Area
205 - Plant Management Systems;

Subject Of Investigation
7210 - Remote sensing equipment and technology;

Field Of Science
2020 - Engineering; 2060 - Geography;
Goals / Objectives
The major goal of the research is to develop a pipeline to model suitability using UAV and satellite data under current and future climate conditions. We will develop and apply the pipeline for three specialty crops, grapes, strawberries and citrus under current climate and 2021-2040, 2041-2060, 2061-2080, 2081-2100 conditions. In this project, parallel and distributed algorithms will be used to supply coarse, medium and fine-grained analysis to the suitability model.The specific objectives are:1. Estimate current and future agricultural suitability using satellite data.2. Estimate current and future plant-health suitability using UAV data.3. Geospatial data fusion for agricultural suitability.
Project Methods
Objective 1: Method: Our goal is to estimate agricultural suitability using near-present data and future climate data. Species Distribution Model (SDM) is a modeling approach developed to estimate the habitat range of species. SDM has been used by economic geographers for modeling land use suitability for perennial crops and establishing potential of new producing regions.We will test three algorithms: (i) Maximum Entropy (Maxent); (ii) Random Forest; and (iii) Support Vector Machine. The results of each algorithm will be evaluated and only the algorithm that produces the best-fitting prediction will be used on the next steps of this research. To identify which algorithm generates the SDM that best represents the set of conditions for each crop, we will perform the True Skilled Statistics (TSS) as the assessment tool. The TSS varies from -1 to +1, where negative values and <0.5 are considered no better than random and where a value closer to +1 is considered excellent. Additionally, our evaluation uses a repeated measures ANOVA to assess the equality of the TSS means for each algorithm regarding each crop. Thus, the best algorithm is the one with the highest TSS over all samples for each crop.To estimate the SDM, we will need a set of input locations, i.e., the current crop producing areas for grapes, citrus, and strawberries in California. Our data source is the US Department of Agriculture National Statistics Service (USDA-NASS) Cropland Data Layer (CDL) from 2008 - 2020. For the environmental layers, we will use temperature, precipitation, bioclimatic variables, and slope from the WorldClim v2.1 database. Soil is from the Soil Survey Geographic Database (SSURGO). We will utilize future climate bioclimatic variables generated from eight global climate models (GCMs) and for four Shared Socio-Economic Pathways (SSPs): 126, 245, 370, and 585. Our future conditions SDM will be estimated based on four time periods: 2021-2040, 2041-2060, 2061-2080, 2081-2100. For each time period, we will estimate 32 SDMs for each crop (which will be aggregated into an ensemble layer) for a total of 96 SDMs (3 crops x 8 GCMs x 4 SSPs). The SDM will be estimated at a spatial resolution of 2.5 minutes or approximately 4.5 km. Areas predicted as suitable are those with high congruence of SDM' prediction by 70% or more of the models.Effort: Experiential learning opportunities for Cal Poly Pomona students (undergraduates and graduates). Informal educational program related to interpretation of the results as suitability maps.Evaluation: The key milestones are: 1) data collection; 2) calibration of SDM; 3) projection of SDM into several SSPs and time periods. Additional milestones: 1) recruit undergraduate and graduate students; 2) training students to perform SDM analysis; 3) creation of self-guided interpretation for the suitability maps.Objective 2: Method: The rapid advance on UAVs technology has opened up several applications of UAVs in agriculture. The proposed suitability using UAV remote sensing data is a Suitability Plant Health (SPH). The SPH is a modified version of the SDM described on Objective 1, in which we will estimate the relationship between plant health and local environmental conditions and field indicators such as chlorophyll and water potential. The SPH considers each plant as an observation unit. Our analysis has three components: (1) Data collection and remote sensing; (2) SPH estimation; (3) SPH under future climate.In (1), we will collect and process the UAV data and calculate the plant health index (i.e., Normalized Difference Vegetation Index - NDVI) and other metrics (water band index (WBI), green NDVI or GNDVI and ENDVI (variations of NDVI), modified chlorophyll absorption ratio index (MCARI), photochemical reflectance index (PRI), and red edge ratio). Additionally, we will collect weather data from the Cal Poly Pomona Weather Station. These datasets will be used as inputs for the SPH estimation. In (2), we will test three machine-learning algorithms to estimate the best fit between the plant's NDVI and its local environment. The selection of the best algorithm will be based on statistical analysis of the estimation results, including the TSS at the ROC threshold, and repeated measures ANOVA. This estimation process will generate a probability layer indicating where the healthy plants are located and which combination of factors are more relevant to SPH. In (3), the SPH model will be used will the same set of future climate projections described in Objective 1, to know: eight GCM, under four SSP, for the decades of 2021-2040, 2041-2060, 2061-2080, 2081-2100.Effort: Experiential learning opportunities for Cal Poly Pomona students (undergraduates and graduates). Informal educational program related to interpretation of the results as suitability maps.Evaluation: The key milestones are: 1) UAV data collection; 2) calibration of SPH; 3) projection of SPH into several SSPs and time periods. Additional milestones: 1) recruit undergraduate and graduate students; 2) training students to perform SPH analysis; 3) creation of self-guided interpretation for the suitability maps.Objective 3: Method: Currently, the extraction of required information from UAV remote sensing data is a time-consuming procedure. Our goal is to reduce this time thus making data immediately available for plant health assessment and precision agriculture. We will focus on parallelizing the data fusion of UAV and satellite data and the estimation of suitability components. Our particular focus will be on formulating parallel/distributed algorithms for heterogeneous environments of shared memory (e.g. GPUs) and distributed memory architectures (e.g. clusters). As noted above, we will focus on the following two components: (i) Data Fusion: Data fusion is conducted at three levels, namely pixel level, feature level and decision level. We will develop both coarse grained and fine grained algorithms for fusion at the pixel and feature levels. It is expected that these designs will also support the future development of real time and near real time fusion of UAV and satellite data; and (ii) Machine Learning based suitability estimation: Classification based predictive analytics will be utilized as part of the crop suitability analysis. We will specifically explore supervised learning-based techniques which utilizes Maxent, Support Vector Machine, and neural networks based classification. We will focus on parallelizing the training of these classifiers. Our parallel designs will be based on both data parallelism and model parallelism. The designs with data parallelism partitions the training data and a separate model is trained using a partition. These models are combined to form the final classifier. Similarly, in model parallelism based designs, the entire data set is used to train multiple partial models which are then combined to form the final classifier. As mentioned before, our work will also include performance comparison of algorithm designs deployed on heterogenous computing environments. Specifically, we will analyze performance metric such as speedup and communication costs for executing the algorithms on GPU and multi-core architectures in both cluster and cloud computing environments.Effort: Research opportunities for Cal Poly Pomona students (undergraduates and graduates). Informal educational program related to interpretation of the results as suitability maps.Evaluation: The key milestones are: 1) data fusion at pixel and feature level; 2) data fusion at decision level; 3) machine-learning algorithm development. Additional milestones: 1) recruit undergraduate and graduate students; 2) training students on data fusion; 3) training students on machine-learning suitability estimation

Progress 01/01/24 to 12/31/24

Outputs
Target Audience:During this reporting period, our project reached a diverse target audience across academia and industry. We engaged with California-based growers of specialty crops, providing them with practical insights to improve their agricultural practices. We also targeted researchers in the scientific community focused on agricultural suitability modeling, fostering collaboration and knowledge sharing among experts. Additionally, we reached out to experts exploring sustainability through interdisciplinary approaches, facilitating discussions on innovative solutions for a more sustainable future. Furthermore, our project had an impact on students at higher education in disciplines not traditionally associated with agriculture such as geography, computer science, and aerospace engineering. We introduced undergraduate students to cutting-edge research methods and applications in the field of agricultural sustainability, while providing graduate students with opportunities for mentorship and professional development in this area. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?The project has provided numerous opportunities for training and professional development, particularly in the area of agricultural suitability analysis. To support this goal, we supported the hiring of undergraduate and graduate students to work on estimating current and future agricultural suitability using satellite data. These students were trained on suitability analysis and database management, gaining valuable skills that will benefit them throughout their careers. Students were first authors in several conference posters and they are leading the efforts for conference papers targeting next year conference cycle. In addition to our work on agricultural suitability, the project has also provided opportunities for training in precision agriculture and geospatial data fusion. Students learned about various platforms, including Unmanned Aerial Vehicles (UAVs), that can be used to collect high-resolution data. This knowledge will enable them to contribute meaningfully to projects focused on plant health assessment and management. Furthermore, the project has provided opportunities for graduate students to gain advanced training in computer science applications relevant to agriculture and natural resources. Our team's work on geospatial data fusion has enabled these students to develop expertise in this critical area, preparing them for careers that require a deep understanding of both computational methods and agricultural systems. How have the results been disseminated to communities of interest?Dissemination has taken the shape of academic posters and presentations at conferences and showcases. What do you plan to do during the next reporting period to accomplish the goals?For the next year, we will focus on refining the plant-health analysis using UAV data. The preliminary results are interesting, however, they are limited to one crop and one growing season. We expect that additional data will enable us to perform further model improvements.

Impacts
What was accomplished under these goals? Specific Objective 1: One of our major accomplishments during this reporting period was to estimate current and future agricultural suitability using satellite data. To achieve this, we gathered relevant data for suitability analysis, which laid the foundation for our subsequent work. We then performed a comprehensive analysis at 2.5 minutes spatial resolution, focusing on two time periods: 2041-2060 and 2081-2100. This effort enabled us to provide valuable insights into the current and future agricultural potential of various regions. The new stage of research is to improve on the spatial resolution, to use environmental layers at 30 seconds spatial resolution. The improved resolution will make our findings more accessible to farmers and other decision-makers. Furthermore, we explored land use change and movement due to environmental changes, recognizing the critical importance of understanding these dynamics in the context of agricultural sustainability. Our analysis revealed the pattern of crop competition in California and how this competition is influenced by climatic conditions. Specific Objective 2: We estimated current and future plant-health suitability using Unmanned Aerial Vehicle (UAV) data. To achieve this, our team collected high-resolution imagery during the growing season, developed a protocol for data collection and sharing within our group, and fused geospatial data to generate valuable insights into agricultural suitability. The initial findings indicate that UAV data has a limited role in the estimation of future plant-health. Our team is working on the correlation between the UAV and other high-resolution environmental data. Specific Objective 3: In support of these efforts, we identified current algorithms for satellite image fusion and optimized their performance on big data. This work enabled us to leverage the strengths of various approaches, ultimately improving the accuracy and efficiency of our analyses. Finally, we explored the potential of convolutional neural networks (CNNs) as classifiers for high-resolution UAV data while comparing it to the establish algorithm Maxent and other machine-learning methods. Our examination revealed promising results, highlighting the potential of this approach to enhance plant-health suitability assessments in agricultural settings. On the other hand, CNNs are not required as classifier for coarse resolution data. In one study comparing CNNs and Maxent using the Cropland Data Layer as input, Maxent outperformed CNNs.

Publications

  • Type: Conference Papers and Presentations Status: Accepted Year Published: 2024 Citation: 2024. Synder, S., Pool, J., G. Granco. Sustainability and Land Change PolyX. Poster presentation at the PolyX Showcase, Cal Poly Pomona, Pomona, CA
  • Type: Conference Papers and Presentations Status: Accepted Year Published: 2024 Citation: 2024. Synder, S., Pool, J., G. Granco. Climate-Driven Shifts in Specialty Crop Production: A Spatial Analysis of California's Agricultural Landscape (2008-2020). Poster presentation at the Annual Meeting of the Association of Pacific Coast Geographers, Cal Poly Humbolt, Arcata, CA.
  • Type: Theses/Dissertations Status: Published Year Published: 2023 Citation: Siddharth Kekre. Gen-AI Based Spatio-Temporal Fusion. Proposed Approach for Image Data Enhancement. California State Polytechnic University, Pomona. Print.
  • Type: Theses/Dissertations Status: Published Year Published: 2023 Citation: Shubhrose Singh. A Hybrid Edge-Cloud Computing Framework To Fuse Multi-Resolution Images For Agricultural Applications. California State Polytechnic University, Pomona, 2023. Print.
  • Type: Theses/Dissertations Status: Published Year Published: 2024 Citation: Kalin Zaluzec. An Edge Computing Framework for Fusing Geospatial Data Using Laplacian Super Resolution Networks. California State Polytechnic University, Pomona Print.
  • Type: Theses/Dissertations Status: Published Year Published: 2023 Citation: Nilay Nagar. Crop Suitability Modeling Techniques Using CNNs in Context of Species Distribution Modeling: A Case Study on Distribution of Strawberries and Grapes in California. California State Polytechnic University, Pomona, 2023. Print.


Progress 01/01/23 to 12/31/23

Outputs
Target Audience:During the second year of this project we reached a target audience of: California growers of almonds, citrus, pistachios, and walnuts. Scientific community working on agricultural suitability modeling. Scientific community working on sustainability using an interdisciplinary approach. Undergraduate students at a minority serving institution. Graduate students at a minority serving institution. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided? Estimate current and future agricultural suitability using satellite data. Supported hiring of undergraduate and graduate students to work on this goal. Students were trained on suitability analysis and database management Estimate current and future plant-health suitability using UAV data. Students learned about UAV and other platforms to collect data Training on precision agriculture Geospatial data fusion for agricultural suitability. Graduate students were trained on applications of advanced computer science to agriculture and natural resources. How have the results been disseminated to communities of interest?Results have been disseminated through articles publication and conference presentations. At the university level, results have been disseminated to attract more students into this research program. What do you plan to do during the next reporting period to accomplish the goals?We plan to: 1) Conduct suitability analysis using newer climate change data at higher resolution; 2) Create public facing platform to share results with broader audience.

Impacts
What was accomplished under these goals? Estimate current and future agricultural suitability using satellite data. Gathered data for suitability analysis. Performed analysis at 2.5 minutes spatial resolution, for period of 2041-2060 and 2081-2100 Preparation of manuscript on agricultural suitability Analysis of land use change and land use movement due to environmental changes Estimate current and future plant-health suitability using UAV data. Collected UAV imagery during growing season Created protocol for data collection and data sharing for group Geospatial data fusion for agricultural suitability. Identified current data fusion algorithms for satellite imagery Optimizing algorithm performance on big data Examination of convolutional neural network as a classifier for high-resolution data.

Publications

  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Martha E Mather, Gabriel Granco, Jason S Bergtold, Marcellus M Caldas, Jessica L Heier Stamm, Aleksey Y Sheshukov, Matthew R Sanderson, Melinda D Daniels, Achieving success with RISE: A widely implementable, iterative, structured process for mastering interdisciplinary team science collaborations, BioScience, Volume 73, Issue 12, December 2023, Pages 891905, https://doi.org/10.1093/biosci/biad097
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Granco, Gabriel, Haoji He, Brandon Lentz, Jully Voong, Alan Reeve, and Exal Vega. 2023. "Mid- and End-of-the-Century Estimation of Agricultural Suitability of Californias Specialty Crops" Land 12, no. 10: 1907. https://doi.org/10.3390/land12101907
  • Type: Conference Papers and Presentations Status: Accepted Year Published: 2023 Citation: Cornejo, K., G. Granco. The Evapotranspiration of Almonds Crops during Drought Periods. Poster presentation at the CARS Conference, Cal Poly Pomona, Pomona, CA.
  • Type: Conference Papers and Presentations Status: Accepted Year Published: 2023 Citation: Misa, B., Kawasaki, J., G. Granco. The impacts climate change and how it affects California's agricultural production of grapes Poster presentation at the RSCA Conference, Cal Poly Pomona, Pomona, CA
  • Type: Conference Papers and Presentations Status: Accepted Year Published: 2023 Citation: Russell, D., G. Granco. Pistachios on the move: Examining the impact of climate change on cash crops in California. Poster presentation at the RSCA Conference, Cal Poly Pomona, Pomona, CA.
  • Type: Conference Papers and Presentations Status: Accepted Year Published: 2023 Citation: Cisneros, E., G. Granco. Impacts of Climate Change on California Almonds. Poster presentation at the RSCA Conference, Cal Poly Pomona, Pomona, CA
  • Type: Conference Papers and Presentations Status: Accepted Year Published: 2023 Citation: Lentz, B., G. Granco. Modeling Climate Change and its Potential Effects on California Lettuce Oral presentation at the RSCA Conference, Cal Poly Pomona, Pomona, CA
  • Type: Conference Papers and Presentations Status: Accepted Year Published: 2023 Citation: Lentz, B., G. Granco. Modeling Current and Future Climatic Conditions for California Lettuce Oral presentation at the Annual Meeting of the American Association of Geographers (AAG), Denver, CO.
  • Type: Conference Papers and Presentations Status: Accepted Year Published: 2023 Citation: Rosales, S., G. Granco. California Drought: Desirability to grow specialty crops based on response to drought conditions and economic profitability. Poster presentation at the RSCA Conference, Cal Poly Pomona, Pomona, CA


Progress 01/01/22 to 12/31/22

Outputs
Target Audience:During the first year of this project we reached a target audience of: 1. Undergraduateand graduatestudents at a minority serving institution. Internships - Fall 2022. We were able to hire undergraduate students in late Fall. Formal classroom instruction: i) a module on the applications of GIS for climate change and suitability analysis was deployed in GEO4450 - Enviromental Modeling with GIS- Spring 2022, 26 students; ii) three undergraduate students in Geography conducted indenpendent studies on agricultural suitablility - Spring 2022, 3 students; iii) a module on the use of python and other programing languages for GIS as applied to agricultural suitability analysis was deployed in GEO 3220 - GIS Programing and Applications - Fall 2022, 13 students. 2. Broader community in California. We participated in three news report in the Los Angeles region, with one broadcast targeting Spanish-speaking population. Changes/Problems:A major problem that impacted the expenditure of funding was related to adelay in organizing the hiring of students research assistents. Our university andthe CPP Foundation have been facing a shortage of staff which slowed down the process of settingup the account, issuing the hiring call, allowing PIs access to the system to examine applications, and Foundationprocessing of students applications and onboarding. We have worked out the issues, and we should be ready for the next report period. We have two changes to the project: 1)The CPP farm is no longer planting strawberries so we don't have access to a strawberry field. To accommodate this new situation, we plan to continue to investigate the agricultural suitability of grapes, citrus, and strawberry under Objective 1 - Estimate current and future agricultural suitability using satellite data, and we will add another crop to Objective 1, we will investigate tomatoes as well. Under Objective 2 - Estimate current and future plant-health suitability using UAV data, and Objective 3 - Geospatial data fusion for agricultural suitability, we plan to deviate from our initial set of specialty crops, grapes, citrus, and strawberry,by replacing strawberry with tomatoes.It is important to note that tomatoes are a valuable crop for California and tha the methods used to investigate tomatoes are expected to generalize to strawberries and other crops. 2)We had a change in personal asDr. Chaichi retired and cannot participate in the research project. We have been working with Dr. Priti Saxena, an Assistant Professor in the Department of Plant Science at Cal Poly Pomona. 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? Estimate current and future agricultural suitability using satellite data Download new climate datasets at 30 seconds spatial resolution Prepare data for analysis Manuscript on agricultural suitability is under review with LAND journal Prepare manuscript with improve suitability analysis Estimate current and future plant-health suitability using UAV data Collect UAV data for entire growing session (late spring to late fall) Process imagery for NDVI and other remote sensing products Geospatial data fusion for agricultural suitability Test algorithm for data fusion of satellite and UAV imagery Perform suitability analysis with data fused layers

Impacts
What was accomplished under these goals? Estimate current and future agricultural suitability using satellite data Gathered data for suitability analysis. Performed analysis at 2.5 minutes spatial resolution Preparation of manuscript on agricultural suitability Preparation of manuscript on viticultureadaptation to climate change Estimate current and future plant-health suitability using UAV data Collected UAV imagery during late summer and early fall Created protocol for data collection and data sharing for group Initiate data processing for NDVI Geospatial data fusion for agricultural suitability Identified current data fusion algorithms for satellite imagery Optimizing algorithm performance on different satellites

Publications

  • Type: Journal Articles Status: Accepted Year Published: 2023 Citation: Sun, Q., G. Granco, L. Groves, J. Voong, S. Van Zyl. Viticultural Manipulation and New Technologies to Address Environmental Challenges Caused by Climate Change. Climate, Volume 11, no. 4: 83. https://doi.org/10.3390/cli11040083
  • Type: Conference Papers and Presentations Status: Accepted Year Published: 2022 Citation: 2022. J. Pizzorno, D. Sirisack, G. Granco. Climate Changes Effects on Californias Specialty Crops in the Mid to Late Century at the CARS Conference, Cal Poly Pomona, Pomona, CA. 2022. G. Granco, H. He, B. Lentz, A. Reeves, J. Voong, E. Vega, Y. Zhang, S. Reyes. Agricultural suitability modeled using MaxEnt for climate change at the Annual Meeting of the American Association of Geographers (AAG), virtual. 2022. A. Reeves, G. Granco. Mapping Future Suitability of Walnut Crops at the RSCA Conference, Cal Poly Pomona, Pomona, CA. 2022. H. He, G. Granco. Investigating Potentially Suitable Farmland of Citrus in California using SDM at the RSCA Conference, Cal Poly Pomona, Pomona, CA.