Moving Beyond Anecdotes: Making Data Driven Decisions to Adapt to Climate Change

Through this grant, we learned that access to relevant, site-specific information is an important tool to analyze the causes and effects of various situations on our farm. The aerial imagery collected over the course of the past season illustrated the spread of disease, the breadth and resilience of weeds, and the extent and impact of flooding. Weather station data provided insight into macro-trends and micro-trends. For instance, these objective tools better informed us of when we need to start looking at various field activities (rather than when we felt we should) and gave clues and insights into why crops performed the way they did (rather than guesses). Some specific examples of what we learned over the course of the 2019 season can be found in the Research Results and Discussion section.

These are new tools for our farm and there have been a number of successes and challenges. We have learned a lot about how best to use them and how they can improve what we do and how we can better tell the story of our work. In addition to the research we are continuing to do we have also been able to share our farm with our customers and community in a whole new way. One example was the flooding we experienced last September. After all the rain we received an outpouring of concern. Because we had access to weather data and aerial images illustrating the breadth of the flooding we were better able to convey what happened and how it was impacting our crops with both data and images supporting each other. 

The impact of extreme weather events is varied and complex, and as many farmers already know, it's useful to look at impacts through a variety of lenses and data sets. Extreme weather events cause stress on crops which manifests in different ways per crop severely impacting overall crop health, viability, flavor and postharvest quality. It can be challenging to track all of these factors across a wide variety of crops. Therefore, if looking to correlate weather and the success/health of a crop farmers should pay particular attention to one crop group (e.g., broccoli) or one succession (broccoli in the heat of the summer, or that grew through a wet spell) per season to allow time to deep dive into the various impacts of weather. Over time this will allow farmers to see commonalities in the correlation between weather and crop performance.

Additionally, the predictive signs and trends depend on these deep dives into the various issues per weather event which also differ in their makeup (i.e., no two weather events are exactly the same). A key takeaway is to think about the information into two key categories: 

1) Action-Oriented Data - information that allows us to make better decisions during the growing season, and

2) Planning Data - information that we can look back on to determine factors in the success or lack of success of crops when reviewing the past season. 

Action-Oriented Data

Defined as information that allows us to make better decisions during the growing season:

• Weather Station Data:
  • In addition to visual cues and “hands-on” data points, the soil moisture and rainfall data allow us to make more accurate decisions about when to complete field activities (planting, tillage, cultivation, irrigation, etc.) and when to hold off and put the soil’s health first. 
• Drone Imagery:
  • As opposed to walking and re-walking fields in multiple directions and some amount of guesswork, aerial videos provide time savings as they provide useful larger-scale data that complements and builds on what we can collect from the ground level. Some examples include:  
    • Instead of sending tractors or farm crews into a field, farmers can use aerial images to better assess the extent of weed pressure and better determine the best tool(s) to use and/or whether it's able to be controlled and/or might jeopardize the success of a crop.
    • Before spending money and time on additional re-seeding efforts, farmers can use aerial images to assess the germination of cover crops and consider whether location-specific reseeding efforts could be more efficient.
  • Furthermore, aerial photos are useful in seeing the health, disease pressure and variety differences in crops because you can see plantings as a whole in one view. 

Planning Data

Defined as historical information that may be referred to determine factors of crop success or failure when reviewing the past season. 

• The ongoing automatic temperature and humidity data logs (as opposed to keeping written logs) are helpful touch points (in addition to those mentioned above) when striving to identify factors leading to plant disease. In addition to the tools purchased, we would recommend farmers consider adding a leaf wetness meter to add an additional, automated data collection tool.
• The observations we’ve made (and continue to make) through the drone imagery supported by the weather data are providing valuable information as we develop and improve crop and field rotations. This information is allowing us to see the nuances of how different fields hold water and when they are dry enough to be able to grow consistent crops. We are using this information to classify different areas of our fields into groups that are reliably available at different times of the year. These groups are in turn being used to decide what crops get planted when and where. This is an ongoing process that continues to evolve as we observe the impacts and repercussions of various weather events.

Most of the challenges we experienced were related to using new technologies. We had a learning curve with the drone finding the best height, speed, angle and when to use photographs versus videos to capture what we wanted. As we look forward to the second year of the grant we will be developing a more standardized process to make these variables more consistent. The weather data is collected automatically, however, we are still learning the intricacies of configuring the logger to collect the right amount of information for our needs. It also took some experimentation to put the various data points together and use that information to draw conclusions. 

This project demonstrated the need for patience and additional time for experimentation when transitioning to new measurement tools. Just as farmers need additional time, practice and patience to integrate new implements into the farm’s workflow, we have found this is also the case for these tools. We look forward to continuing to learn and become more predictive as we learn more about the tools and data.

Beyond integrating the regular use of these measurement tools into our workflow, we will continue to adjust how and when we review, analyze and then integrate the data into our crop and field planning process. We would recommend farmers consider each measurement tool and who is responsible for consistent measurement and data collection, and wherever possible automate/standardize the data collection process so there are fewer decision points. 

Our action-oriented data saved us hours in moving around our 40 acre farm. As opposed to checking on field conditions too soon, we used the data collected to better predict the ideal timing of field activities (i.e., capturing data on the rates at which soil was drying). We better prioritized work based on the data for soil conditions and we feel better making the hard decision when to terminate a failing crop based on data collected in both imagery and through the weather station (e.g. how quick/slow soil is drying, weed and/or disease pressure, humidity levels, etc.) While there is no replacement for boots on the ground, these data points aid in making more objective decisions using a more complete picture of what is and has happened.

Our farm’s viability depends on economic, social and environmental factors. As climate extremes continue to affect our outdoor growing environment, we seek to embrace technical solutions and tools to help us learn, adapt and protect the quality of our crops from the extremes. This takes the form of investing in hoophouse structures to protect key crops, technology to better track planting, harvest and yield data, and access to objective data that allows us to make better decisions and become smarter growers. While there is similar weather information available through various sources such as the National Weather Service, we have found that our proximity to Lake Michigan means that both precipitation and temperatures can vary widely between their nearest reporting site and our farm. We found highly detailed, place-based data is important to our learning how to better work with the land and soil conditions and not against them. There is also value in having the information easily accessible.

We realize the future of our farm’s viability also depends on technology tools to help us manage smarter as well. For instance, we program alarms and notifications to our weather station to prevent potential crop loss in both our hoophouses and greenhouse due to both hot and cold temperature extremes. We also have alarms for wind so we are sure to close structures to prevent damage. We will use data on growing degree days and leaf moisture reading to watch for pest and disease outbreaks. Our viability will continue to rely on learning about new pests, diseases and being proactive in the prevention of diseases.  

We believe our viability also depends on the viability of other farmers, and thus we will continue to openly share information. We plan to register our weather station with Cornell’s NEWA network to share information and utilize their models to forecast pest and disease pressure.

We see a large potential for both these technologies to work together and independently to benefit sustainable farming. As we move forward and build a better understanding of their strengths and weaknesses we will begin to share this information with other farms and advise them on how these technologies could fit into their farms to address challenges they might be having.

Summary Lessons Learned and Recommendations
In summary, these are lessons we share with farmers and growers:

• We found having two sets of data sources is useful, but at times duplicative. We would recommend using the two sets of data for different uses. 
• Synergies were found in data correlation when looking for larger trends or to answer certain hypotheses about disease pressure or fertility needs.
• These tools are also useful for additional reasons: proactive temperatures control within hoophouses and greenhouses, watching weather trends, sharing the importance of moisture levels within soils with others, using data to better predict the timing for farm tasks and the potential for disease pressure.
• When extreme external factors come into play (e.g., global pandemic), we found consistently gathering data was challenging for farmer-dependent tools, whereas we saw the benefit of automated data logging and capture tools that didn't require a farmer's time.
• We found the scalability and flexibility of the technology tools an essential aspect for the tools to grow with our knowledge.  
• We’ve found that our work is more predictive and less reactive due to additional information. 
• Aerial imagery captures new pieces of information that you didn’t realize you could find useful (e.g., finding trends within plantings).
• Consider the time available to maintain your technology (e.g., to maintain drone license, a yearly certification is required).
• We would recommend shared or collaborative purchasing of equipment and share data between farmers.
• The cost for maintaining both pieces of technology is relatively low. For continued commercial use, the aerial drone license fee of $150, along with passing a safety test, is required every two years. We also foresee costs in replacement batteries (approx $185). However, over the course of two years, we have not seen any noticeable decreases in battery life. Thus far, the only ongoing cost associated with the weather station was the addition of sensors. Most sensors are in the range of $150-200. We’ve decided to invest in additional sensors to track temperatures in our greenhouse and hoophouses, and we invested in a leaf moisture sensor to begin to better predict disease outbreaks. In our experience, the most useful sensors have been, temperature, humidity, rainfall, and soil moisture.
• Time Requirements:
  • Learning
    • FAA License - I spent about 10-15 hours over a couple of weeks to feel prepared for the test. After completing the test I felt like I may have been a little over-prepared. This is only required for commercial operations.
    • Drone Operation - The drone we use has a “training mode” that limits the height and distance from the operator. I spent 1-2 hours in this mode getting a better sense of the controls before taking it out on a full flight. After that, it was mainly a process of learning the finesse of the control to make smoother turns and adjustments. There are quite a few modes and settings you can use to make flying safer (height limits, distance from operator). 
  • Ongoing
    • I aim to fly drone every other week throughout the growing season plus additional flights depending on conditions. Each flight was about 30 minutes plus 15-20 minutes for set-up and putting everything away. I am generally on my own while flying, however, if you are trying to cover a more extensive range or an area with lots of obstructions you would want a helper to maintain line of sight with the drone.
    • Reviewing the information generally took an additional hour depending on the amount of detail I was looking at. Whereas the timing of the flight is dependent on conditions and farm work the review of the information is very flexible.  
    • Field observations fit into the regular field walks we do to stay on top of our farming operations and didn’t add much time. Many times these observations were more productive because we caught things in the drone flights that called for additional investigations.
• Our recommendation to farmers would be to standardize the data collected to ensure comparing and contrasting opportunities in the future. We found the sweet spot for data collection on our farm was to fly at a height of 80-100’ above ground to receive the best combination of detail and scope. We also began taking three different views: 1) a bird's eye video flying at ~100’, 2) a bird’s eye photo, and 3) a photograph looking straight down at a height to capture most of the field/planting. Three different views gave us the most useful information. 
• Given the drone’s abilities, we tested various aspects of data we could collect. While we initially focused mainly on using the drone to create videos, we gradually shifted to using the drone to capture more still images from above. While videos are useful we would recommend farmers consider still images to capture clearer images and to observe detailed aspects of images through the ability to more easily focus and zoom.
• A filter for the drone’s camera is required to get usable video and images without and wobble. We found ND8 filter to be the most versatile and cost-effective. If purchased as an aftermarket add on the ND8 filters are not too expensive and can be found as a set with multiple different filters which can be useful for different types of light conditions.