Final report for GS22-258
Project Information
This project aims to evaluate how climate change affects the sustainability of perennial crop production in the Southern U.S. The value of deciduous fruit and nut production in the southern states is about two billion dollars every year, about 8.5% of the value of crop production in 2020. Given the long lifespan of deciduous fruit and nuts, it is essential to have long-term plans for adapting to climate change. Climate changes of the next couple of decades will be vital for crops already on the land now and for those cultivated in the next few years. With county-level crop production and climate data, I will estimate the responses of yield and acreage to climate variables by machine learning methods and then simulate the expected production loss or gain from climate change. The findings of this project will inform policymakers and practitioners of the potential benefits of adaptive decisions for deciduous crop production in the Southern states to climate change.
Objective 1: Estimate the responses of perennial crops, both yields and acreages, to climate change using historical crop and climate data
Objective 2: Simulate the impacts of changing climate on these perennial crops in the middle and end of the century using projected climate data.
Research
- This project plans to construct a county-level dataset on acreage, production, and revenue for the ten deciduous fruit and nuts in the Southern states in 1950-2020: apple, avocado, blueberry, peach, strawberry, pecan, orange, grapefruit, tangerines, and mandarins. Citrus fruit production in the Southern region is mainly from Florida. County-level data on citrus fruit (orange, grapefruit, and tangerine) in Florida from 1950 to 2020 can be found in the Citrus Summary from the United States Department of Agriculture National Agricultural Statistics Service (USDA-NASS). County-level data on noncitrus fruits and nuts will be obtained from the USDA-NASS Field Offices and state agricultural departments. The study period of noncitrus fruits and nuts might not be 1950-2020. It depends on the availability of county data.
- This project also plans to construct a county-level dataset of weather in the Southern region. I will use climate data from the two widely accepted data sources: Global Historical Climatology Network ‐Daily (GHCN‐D) data and Parameter-elevation Relationships on Independent Slopes Model (PRISM) Climate Data. The former is a collection of station-level weather data. The latter is grided data. Each type of data has its advantages and disadvantages. The choice depends on the further comparison of the two datasets in the Southern region.
- This project plans to consider different phenological stages when estimating the impacts of climate change. Some common characteristics of deciduous crops include dormancy, chilling requirement, development of bud, leaf, and shoot, flowering, fruit development, and fruit maturity (the latter six are from the BBCH-scale, which was developed to identify phenological stages of plants). For example, most tree crops require a certain length of freezing exposure (called chilling requirement) for flowering and fruiting. Projected climate data show that the chilling hours will shrink, threatening the production of these crops. Spring frost is another factor threatening production. An unusual warmth in March and temperatures below freezing in early April led to frost damage in the southeastern areas (Augspurger, 2013; Drepper et al., 2021; Ramirez & Kallarackal, 2015; Snyder, 2005).
- After constructing the two main datasets, this project plans to estimate the impacts of climate variables on yields and production with machine learning. The final dataset might be panel data whose maximum length is several decades. However, there are plenty of potential climate variables. To analyze data with high dimensions, this project plans to use the methods of machine learning, such as the LASSO model and regression tree modeling.
- After getting the fitted model of yields and production of those crops, this project plans to simulate the impacts of climate change on yields and production until the middle and end of this century. Projected climate under scenarios representative concentration pathways (RCPs) 4.5 and 8.5 of the southern states is obtained from the WCRP Coupled Model Intercomparison Project (CMIP).
- Under Objective 1, we conducted a thorough search for perennial crop production data available for the southern states. We successfully complied county-level data on citrus fruit production (oranges, grapefruits, and tangerines and mandarins) for Florida between 1948 and 2021 from the United States Department of Agriculture National Agricultural Statistics Service (USDA-NASS). Variables obtained include number of trees bearing age, production in boxes, and total acreage for part of the series (1992 to 2007). For non-citrus fruits, we obtained county-level data on pecans, peaches, grapes, and apples for Georgia from 2001 to 2022. Variables obtained include acres of production, yield per acre, and value of production. We also obtained citrus fruit production for Texas from the last six agricultural censuses.
- Under Objective 2, we constructed a county-level dataset of weather variables for all southern states. We obtained daily maximum temperature, minimum temperature, and precipitation from the weather dataset compiled by Wolfram Schlenker using data from the Parameter-elevation Relationships on Independent Slopes Model (PRISM) Climate Data. The dataset offers daily grid-level weather data spanning the years 1950 to 2019 across the United States. We aggregated the grid-level data to the county level, employing crop area weights provided by Schlenker. In addition, the daily maximum and minimum temperature observations provided in the dataset enable us to construct measures of chilling accumulation and growing degree days.
- Under Objective 3, we model phenological stages of different deciduous fruit and nut crops in the Southern States. Citrus is year-round development and harvest. We aggregate weather variables in four seasons and estimate the impacts of mean temperatures and total precipitation over four calendar seasons. For other crops, we aggregate daily mean temperature and precipitation according to four phenological phases: dormancy, bloom, fruit development, and fruit maturity.
- We have finished all data collection and developed preliminary estimation methods. We employ the framework introduced by Merel and Gammans (2021) to model adaptation and estimate both the long-run and short-run impacts of weather on crop yields. Under Objective 4, we also plan to apply the grid search method (GSM) outlined in Li and Ortiz-Bobea (2022) to construct appropriate aggregation windows for weather measures. This method entails fitting models with weather measures from every possible window within a calendar year, operating under the premise that if a weather measure is aggregated over a window that significantly influences yield, the model's fit should be superior in comparison to aggregations over alternative windows. Our preliminary findings indicate that weather impacts vary across phenological phases and the long-run impacts of adaptation are different from short-run impacts.
- We have not finished objective 5. We have developed the preliminary projection methods. We will continue with model estimation and plan to predict the impacts of climate change on yields and production for citrus in Florida and pecans, peaches, grapes, and apples for Georgia until the middle and end of this century.
Educational & Outreach Activities
Participation Summary:
Project findings have been incorporated into an undergraduate course at Virginia Tech (AAEC 3324: Environment and Sustainable Development Economics).
Working papers have been presented at professional conferences, including the annual meeting of Agricultural and Applied Economics Association in 2023.
Project Outcomes
Our project highlights the pivotal role of fruit and tree nuts in ensuring a robust food supply and providing essential nutrition for the human society. Empirical economic research focused on the impacts of climate change and adaptation on perennial crops remains limited. We contribute to the literature by modeling the intricate phenological characteristics of perennial crops and the climate impact and adaptation.
We developed a deeper understanding of the yield response and adaptation of perennial crops to climate change.