A Collaborative Phenology Modeling System to Enhance Crop Management on Vegetable Farms

A Collaborative Phenology Modeling System to Enhance Crop Management on Vegetable Farms

Summary

In the spring of 2013 we selected the initial crops and varieties and weed species to be modeled. We also identified some priority features of the website and named the website and the project CROPTIME. During 2013 and 2014 we drafted two fact sheets and detailed growth stage descriptions for all 16 vegetables being modeled. We launched the outreach and educational “front page” of the CROPTIME website: http://smallfarms.oregonstate.edu/croptime. The modeling interface is housed at OSU’s Integrated Plant Protection Center site: http://uspest.org/dd/multmodel. The models will also be housed at this site. During 2013 and 2014 we collected extensive field data to support development of vegetable and weed degree-day (DD) models. We now have enough data to start developing the first models and have started to verify threshold temperatures and DD accumulation for various key growth stages.

Objectives/Performance Targets

1. Participatory research and system development.
   1. Identify crops and varieties to model.
   2. Identify weeds and N management parameters to model.
   3. Gather input from growers on CROPTIME website functions.
   4. Develop Extension expertise in DD modeling.
   5. Conduct workshops to teach people how to use CROPTIME and develop models.
   6. Develop online instructions to teach CROPTIME users about DD models and how to use the website. Develop the site to be user-friendly.
2. Online degree day planning tool.
   1. Develop website with features prioritized by growers and Extension faculty.
   2. Develop a restricted photo-period parameter for day-length sensitive crops and weeds.
   3. Develop growth stage descriptions and record keeping templates for crops.
3. Phenology model development.
   1. Identify lower and upper development thresholds for the crops and weeds modeled.
   2. Determine DDs for at least 50 vegetable variety models, six weed models and a DD sensitive nitrogen management tool.
   3. Investigate tomato and pepper development with and without black plastic mulch.

Accomplishments/Milestones

1. Participatory research and system development.
   1. Identify crops and varieties to model. Seventy-eight varieties from 16 different crops were prioritized for modeling. Crops are snap beans, broccoli, Brussels sprout, cabbage, carrot, cauliflower, cucumber, kale, head lettuce, parsnip, sweet pepper, pumpkin, summer squash, sweet corn, tomato, and winter squash. The full list of varieties is at: Crop and Variety List.
   2. Identify weeds and N management parameters to model. Participants selected pigweed, nightshade, common lambsquarters, crabgrass, chickweed, and smartweed to model. Sullivan is working on nitrogen management information that can be incorporated into an Extension publication that will complement the Croptime tool.
   3. Gather input from growers on CROPTIME website functions. During the March 2013 seminar and follow-up discussions we surveyed growers about key functions for the website. Growers and vegetable processors would like to enter multiple start dates for different crops; seed company representatives would also like to be able to enter multiple varieties. They want CROPTIME to be able to model transplanted and direct-seeded crops, as well as day-length sensitive crops and weeds. They would also like to predict crop development on black plastic mulch (especially peppers and tomatoes). They would also like to have options other than 30-year average for long-term forecasts and to be able to display average first and last frost dates. They were interested in being able to “dial up” or “dial down” DDs to maturity from known standard varieties (i.e. based on relative days to maturity reported in seed catalogs). Some of these features are being incorporated into the website; others would require additional funding.  
   4. Develop Extension expertise in DD modeling. Throughout the project, Coop is mentoring Andrews and Noordijk in DD model development. The website is being developed to facilitate “open-source” data collection for further model development after the grant period.
   5. Conduct workshops to teach people how to use CROPTIME and develop models. Participants rated the quality of our 2013 seminar at 4.8/5. Insight from this workshop will help us enhance future public workshops planned for the outreach phase of this project (starting in 2015).
   6. Develop online instructions to teach CROPTIME users about DD models and how to use the website. Develop the site to be user friendly. Andrews, Coop, and Myers drafted two online fact sheets: “An introduction to vegetable degree-day models” and “How to develop a vegetable degree-day model using CROPTIME.” They are at http://smallfarms.oregonstate.edu/croptime-fact-sheets. With funding from Clackamas County (see impacts and contributions), we partnered with OSU Extension and Experiment Station Communications (EESC) to survey participants about preferred learning methods. Participants rated online fact sheets, ask an expert, informational “flash cards” and brochures, and pop-up instruction bubbles built into the CROPTIME website as their favored methods for learning. Other popular methods include a multi-media user’s manual (text and video), one-on-one training, face-to-face workshops, and short instructional videos.
2. Online degree day planning tool.
   1. Develop website with features prioritized by growers and Extension faculty. Coop developed the first draft CROPTIME calculator in early 2013 which we are using in the project. It is under continual revision: http://uspest.org/dd/multmodel. Coop developed a google maps interface to locate weather stations (see link above) and he has also added the ability to run up to four planting dates for each model run (select CROPTIME models under model category on above website). In 2014 we worked with a Microsoft usability engineer to conduct usability tests which should help us make the CROPTIME interface relatively user-friendly and minimize the need for user manuals. We plan to continue intermittent usability tests in 2015 and future years as the CROPTIME interface is developed.
   2. Develop a restricted photo period parameter for day-length sensitive crops and weeds. In 2013 Coop found a formula that can calculate day-length at any farm or weather station (using latitude). In 2014 he incorporated this formula into the interface (http://uspest.org/dd/multmodel). In 2015 we plan to identify ways to incorporate information about plant response to day-length into CROPTIME.
   3. Develop growth stage descriptions and record keeping templates for crops. In 2013, we drafted the first growth stage descriptions and record keeping templates for 17 crops (those mentioned above plus Brussels sprout): http://smallfarms.oregonstate.edu/sites/default/files/growth_stage_descriptions_online_2014-08-25.pdf. These descriptions are continually being updated and refined.
3. Phenology model development.
   1. Identify lower and upper development thresholds for the crops and weeds modeled. Our goal with this grant is to develop DD models for 50 vegetable varieties and six weed species. In 2013 and 2015, we collected field data for about 80 vegetable varieties. We now have more than 500 “site-years” (growth stage observations for one variety at one location) of data for these varieties. With this two-year data set we now have enough data to start the modeling work. 2015 data will be needed to meet our goal of 50 vegetable models. Data was also collected for all six weed species we are modeling. We hope to complete four weed models by the end of this project, due to technical complications with weed data. Coop is training Andrews, Noordijk, and Heinrich to use lowest co-efficient of variance to determine lower and upper development thresholds. We anticipate that 6-10 site years will be required to identify threshold temperatures.
   2. Determine DDs for at least 50 vegetable variety models, six weed models and a DD sensitive nitrogen management tool. Data collected in 2013-15 will be used to calculate average DDs to critical growth stages (i.e. maturity, beginning of period of rapid N uptake). When temperature thresholds have been identified, we hope to determine DDs to specific growth stages with less than five site years of data for each variety.
   3. Investigate tomato and pepper development with and without black plastic mulch. In response to grower interest, we conducted a side experiment at the North Willamette Research & Extension Center to observe development of three sweet pepper varieties and five tomato varieties both with and without black plastic mulch in 2013 and 2014. Plots were replicated three times and we recorded crop development, ambient air temperature, and soil temperature at 2”. This data needs to be analyzed and interpreted. We hope that it will inform development of a “correction factor” for black plastic mulch.
   4. Investigate growth rates of transplanted vs. direct seeded crops. In response to grower interest we are also trying to address differences in growth rates between transplanted and direct seeded crops. We are discussing whether to use separate models for direct seeded and transplanted crops, or incorporate both practices into one model.

Impacts and Contributions/Outcomes

In our view, an important factor for long-term adoption of this tool will be adoption of DDs as a planning tool by seed companies. If seed companies list DDs in their catalog and direct their customers to the CROPTIME website, we expect more rapid adoption than through Extension-only outreach. To this end we are developing relationships with Bejo Seed, Vitalis Seed (organic branch of Enza-Zaden), High Mowing Seed, Osborne Seed, Johnny’s Seed, Wild Garden Seed, Adaptive Seed and Territorial Seed. Some of these seed companies conduct variety trials in Oregon and are allowing us to collect degree-day data at their trials. This approach is increasing the efficiency of our data collection.

At the March 2013 seminar we evaluated how much collaborating growers learned. They rated their knowledge before and after the workshop on a scale of 1-7. On average, growers rated their knowledge at 3.0 (low end of medium) before the workshop and 5.4 (high) after the workshop. Their knowledge of all topics increased. Topics included the list below, average knowledge ratings are shown as (before/after):

• Understand the purpose of DD models (4.4/6.0)
• Identify good online weather stations (2.3/5.3)
• Ability to run vegetable DD models (2.3/4.3)
• Use crop models to make decisions on your farm (3.1/5.3)
• Understand how weed DD models could benefit your farm (3.1/5.8)
• Understand how DDs can help with nitrogen management (2.6/4.9)

All respondents plan to use CROPTIME to manage their crops. They all also intend to develop DD models for their own varieties. Farmers expect to tell 40 other farmers about CROPTIME. Collaborators from seed company and vegetable processors work with an estimated 550 farms managing >60,000 acres. They expect to share CROPTIME information with most of these growers. They all plan to develop DD models for their own varieties. One seed company expects to use DDs in their catalog. Another company may use them, and one doesn’t yet plan to.

We secured additional local funding in 2013 and 2014 (Clackamas County) to fund usability testing of the CROPTIME website and work with OSU-EESC to identify and develop the most effective learning materials.

We plan to estimate additional project impacts and outcomes towards the end of the project.

Collaborators: