Crop Scouting with Drones Identifying Crop Variability with UAVs

Transcription

1 DroneDeploy Crop Scouting with Drones Identifying Crop Variability with UAVs A Guide to Evaluating Plant Health and Detecting Crop Stress with Drone Data

2 Table of Contents 01 Introduction Crop Scouting with Drones Identifying Crop Variance Using Plant Health Algorithms on the Farm Eyes in the Sky Cameras for Crop Scouting VARI vs. NDVI Understanding the Different Plant Health Algorithms Conclusion Measuring Plant Health is Just One Piece of the Puzzle...13 Crop Scouting with Drones 2

3 01 Introduction: Crop Scouting with Drones No longer do you have to rely on costly, time-intensive methods of collecting aerial crop data such as satellite imagery or manned aircraft. Drones enable you to assess plant health and spot crop variability in minutes. This puts you in the driver's seat and allows you to lower costs and maximize crop yields along the way. Using drone mapping for crop scouting, you can measure plant health, identify crop stress, and rapidly eliminate threats to your field. But where do you start? Plant health algorithms help highlight variability and make it easier to spot crop stress, but there s a lot of confusion out there about NDVI and VARI. And let s face it it s a complicated topic. For many, it can be hard to determine where to start or how you should start collecting actionable insights with your drone. In this guide, we ll address some of the most common questions, like "what are plant health algorithms", what kind of camera should I buy, and what is the difference between NDVI, false-ndvi and VARI. Let s get started. Crop Scouting with Drones 3

4 02 Identifying Crop Variance: Using Plant Health Algorithms on the Farm We re able to measure plant health using drone imagery because healthy plants reflect light differently than unhealthy plants. Plants that are healthier tend to reflect more green light than red light, which is why they look green. Plants also reflect near-infrared light that is invisible to the naked eye, but can be detected with near-infrared sensors. Fig. Maps with plant health algorithms are assigned colors based on numerical values, alloing you to identify crop variance. Plant health algorithms such as NDVI and VARI (more on this later) compare the proportions of light captured across different bands (red, green, blue and sometimes near-infrared) to compute numerical values for each pixel or area of a given drone map. Maps with plant health algorithms are then assigned colors based on those numerical values, which makes it easy to identify the variance between healthy and unhealthy areas. Using plant health algorithms, you can identify key areas of crop stress and diagnose problems such as pests, disease, irrigation or fertilizer issues, and more. Once you have a better understanding of what's happening in your field, you can work with your agronomist to diagnose and treat issues fast, before they spread. Fig. left (healthy) and right (unhealthy) with NDVI reflectance algorithm. Image courtesy of NASA. Crop Scouting with Drones 4

5 03 Eyes in the Sky: Using Aerial Cameras for Crop Scouting What Kind of Camera Do I Need to Evaluate Plant Health? There are several types of cameras that can help you evaluate plant health, but ultimately you should choose one to meet your individual requirements. We recommend starting simple to see what value can be extracted from your existing drone hardware before investing in more expensive systems. You can always upgrade and invest more in your drone hardware later. Let s take a look at the most common types of cameras used for drone mapping in agriculture. RGB Cameras RGB or standard cameras capture Red, Green, and Blue light. This is the camera type that comes stock with most drone models. These are multipurpose photo and video cameras that can be used to make highdefinition 2D orthomosaic maps as well as 3D models. Additionally, they can be paired with plant health algorithms such as the Visible Atmospherically Resistant Index (VARI) to assess plant health and crop stress in precision agriculture use cases. Fig. RGB cameras come off-the-shelf with any popular drone model, like these models from DJI Crop Scouting with Drones 5

6 Benefits of RGB Sensors We recommend that growers and service providers who are just getting started with drone mapping stick with RGB sensors for a couple of reasons. For one, RGB imagery is more accessible and can be captured with the average camera making it an affordable option that doesn t require the need to invest in additional sensor technology. Additionally, RGB cameras capture images similar to what is seen with the human eye, making the data more familiar and easier to interpret even without the use of plant health algorithms. RGB cameras are also multipurpose. They are able to take pictures for making detailed orthomosaic drone maps and capture high quality videos. This makes them a great choice for anyone getting their start in aerial mapping. Fig. A DJI Inspire flies over crops fields to gather plant health data. Drawbacks of RGB Sensors Remember that RGB cameras are all-purpose and not specifically designed for agriculture. Additionally, they can only be used to determine relative plant health within the same map. If you fly half of your field one day and the other half the next, you won t get a useful reading of crop stress. This is because the amount of sunlight and cloud cover needs to remain constant throughout the duration of the drone flight to attribute the relative data analysis. You may need to compare maps over time to understand what's happening in your fields from month to month, or year to year. RGB cameras will not be the best solution for this type of analysis. Crop Scouting with Drones 6

7 Near Infrared (NIR) Cameras There are three main types of NIR cameras: Modified RGB cameras for Near Infrared, factory-made Near Infrared, and Multispectral cameras. These tend to be more expensive in general compared to RGB cameras, but there are a wide variety of options available on the market. Let s take a closer look. RGB Cameras Modified for Near Infrared Modified cameras are fitted with a filter to capture some combination of Near Infrared, Red, Green and Blue light depending on the model. Fig. An example of an RGB camera modified for near infrared available from MaxMax. Fig. An example of a drone map made with an RGB camera modified for near infrared. Near Infrared Cameras* Factory-made near infrared cameras come manufactured to capture a combination of Near Infrared, Red, Green and Blue light. Fig. An example of a near infrared camera available from Sentera. Fig. An example of a drone map made with a near infrared camera. Crop Scouting with Drones 7

8 Multispectral Cameras* Multispectral cameras typically capture some combination of visible and near infrared light using narrow filters to increase the sensitivity and specificity of the measurements. Some multispectral sensors, such as those from SLANTRANGE, can be configured with a custom set of filters to meet the requirements of specific applications or advanced research. Fig. An example of a multispectral camera available from SLANTRANGE. Fig. An example of a drone map made with a multispectral camera. Benefits of Modified RGB, Near Infrared, and Multispectral Cameras One major benefit to some models of modified RGB and multispectral cameras are their ability to be calibrated with incident light sensors, making it possible to compare absolute plant health data over time. SLANTRANGE multispectral cameras automatically calibrate in flight, others have this capability with additional setup. Calibration makes images and map data comparable over time, despite changes in sunshine and cloud cover from day to day, or week to week. These cameras often provide earlier detection of crop stress, provide more quantifiable scientific NDVI analysis, and enable more advanced crop analysis when you're ready to move beyond basic NDVI metrics. This capability is particularly important for agronomists working with farmers throughout the season and service providers selling drone mapping for hire year round. Drawbacks of Modified RGB and Multispectral Cameras RGB cameras modified with NIR filters and multispectral cameras deliver high performance and accurate (absolute) NDVI imagery, but they require substantially higher investment than standard RGB cameras. Quality NIR-capable cameras can cost anywhere from $1,200 on the lower end, to $7,000 on the high end. Additionally, while many aftermarket camera conversions for RGB cameras are available, image quality is inconsistent across manufacturers, meaning some may lead to poor map quality (see our documentation for a full explanation). We recommend getting quality control samples and a warranty from your hardware vendor prior to any purchase. * DroneDeploy now supports factory-made Near Infrared and Multispectral image processing as of 7/20/2017. Crop Scouting with Drones 8

9 04 VARI vs. NDVI: Understanding the Different Plant Health Algorithms What Type of Plant Health Algorithm Should I Use? We recommend you use the algorithm that is intended for the type of imagery you collect. You can start with the RGB imagery and move up from there with increased investment in other systems. Here's a breakdown of the most common plant health algorithms being used in the industry today. VARI Plant Health Algorithm The Visible Atmospherically Resistant Index (VARI) was designed and tested to work with RGB sensors and is a good measure of plant health using the visible light spectrum. VARI is not intended as a substitute for a NIR camera, but it can provide useful insight when working with non-ndvi imagery. RGB images with the VARI algorithm applied make it possible to detect areas of crop stress in a field. Fig. An example of the VARI algorithm applied to a drone map of a crop field captured using an RGB camera. Why Use the VARI Plant Health Algorithm? Many growers are only seeking to understand the relative plant health of their fields. RGB imagery with the VARI algorithm provides useful information related to variance at a single point in time. VARI paired with ground-truthing can be a valuable tool to help you identify stress and generate variable rate prescriptions. DroneDeploy customers often report that RGB imagery is more than enough to identify variability and gain actionable insights about their crops. Crop Scouting with Drones 9

10 NDVI Plant Health Algorithm The Normalized Difference Vegetation Index (NDVI) is a commonly used remote sensing algorithm intended to measure a plant s overall health. NDVI has been the standard for understanding plant health in the agriculture industry for many years. In the past, Near Infrared imagery has typically been captured by satellites or manned aircraft, but now drones are changing the game. Previously, it may have taken weeks or more to receive NDVI imagery from satellites. Today, you can purchase drones and NIR-capable cameras to collect NDVI imagery on demand, making it both more affordable and accessible than ever. Fig. DroneDeploy calculates the NDVI value for each pixel of your map, giving The NDVI algorithm relies on the comparison of near infrared light to measure healthy plant life across a wide range of conditions. When combined with ground-truthing, this data can give you powerful insights into your area of interest allowing you to better understand where crop stress is occuring and how to address it. Why Use the NDVI Algorithm? Growers, agronomists, and drone service providers (DSPs) who are seeking to understand comparative plant health data over time may wish to use NDVI imagery captured using RGB cameras that are modified with an NIR filter, Near Infrared or use Multispectral cameras. These cameras are specifically designed for agricultural needs. This means that any NDVI images produced are better suited to analyze and detect the smallest amount of crop stress by reading the near-infrared light and picking up on any reflection from plants. Sometimes it can be possible to identify crop issues a few days earlier than with standard RGB sensors. Fig. An example of an NDVI map calculated with the standard formula Crop Scouting with Drones 10

11 Cost Comparison of VARI and NDVI with Common Drone Platforms We know growers and service providers are concerned about the financial risk of purchasing drone platforms. Rest assured that you don t need to drain your wallet to see variability in the field. If you re looking to maximize ROI per acre, we recommend getting started with the DJI Phantom 4 Pro system and an RGB camera using the VARI plant health algorithm. It s a low level of financial risk that can deliver powerful results. See a comparison of drone platforms using VARI and NDVI algorithms below for yourself. Fig. Comparison of VARI and NDVI plant health algorithms across common drone platforms from SenseFly and DJI. As we mentioned before, there are clear reasons why you may want to consider both the NDVI and VARI algorithms to detect crop variability. That said, it's important to realize that you can capture a significant amount of data and make smarter crop management decisions without breaking the bank. Additionally, both Sentera and SLANTRANGE offer sensors that are compatible with some of the more affordable DJI drone platforms pictured above. Addressing False NDVI In your research, you may have come across the term false-ndvi. This refers to the NDVI algorithm applied to RGB imagery. While some growers may find false-ndvi useful, we don t recommend applying NDVI to RGB imagery because the algorithm was not designed for that purpose. The use of "false-ndvi" results in the reporting of inaccurate data. As mentioned previously, the VARI formula is a much more useful way to calculate plant health using RGB imagery. If you plan on using RGB imagery to detect crop variability, we strongly urge you to stick with the VARI algorithm. Crop Scouting with Drones 11

12 05 Conclusion: Measuring Plant Health is Just One Piece of the Puzzle Now that you have a greater understanding of aerial imagery and the various ways to measure plant health with drones, you can begin to leverage this information on the farm. But before you get started, there are a few things you should keep in mind when using these techniques. Here is some parting wisdom from seasoned Ag pros we'd like to share with you. Ground-Truth All Your Findings It is extremely important to put your boots on the ground every NDVI map should still be ground-truthed. Norm Lamothe, Deveron UAS Corp Read the Case Study Remember that no matter which algorithm you choose to measure plant health, you should follow up with inspection. Measuring plant health allows you to see variability, but it won t tell you why that variability exists. You should always get out in the field to ground-truth any crop variance you discover. Take Historical Crop Data into Account The power right now is in understanding your own field. Truly understanding it. Soil type, history all of it. Chad Colby, Colby AgTech Watch the Webinar You ll also want to analyze historical crop data. This, combined with plant health data and ground-truthing, will allow you to see the full picture of what is happening in the field. Comparing data from the previous growing seasons including soil type and health, irrigation methods, and history of pests or disease will help you make smarter crop management decisions. Learn More About NDVI and VARI Plant Health Algorithms If you're seeking more information about drones in agriculture, consider watching our three-part drone clinic series. We dig into the many ways you can use DroneDeploy's platform to minimize costs and maximize yield on the farm throughout the growing season. Agriculture Drone Clinic Level 1: Putting Your Drone to Work in the Field Agriculture Drone Clinic Level 2: Crop Scouting and Field Analysis Agriculture Drone Clinic Level 3: Advanced Crop Scouting & Field Management Happy Flying, The DroneDeploy Team Crop Scouting with Drones 12

13 Areas Mapped with DroneDeploy 7 Continents 160 Countries 10 Million Acres About DroneDeploy DroneDeploy is the leading cloud software platform for commercial drones, and is making the power of aerial data accessible and productive for everyone. Trusted by leading brands globally, DroneDeploy is transforming the way businesses leverage drones and aerial data across industries, including agriculture, construction, mining, inspection and surveying. Simple by design, DroneDeploy enables professional-grade imagery and analysis, 3D modeling and more from any drone on any device. DroneDeploy is located in the heart of San Francisco. To learn more visit us online and join the conversation on Twitter. DroneDeploy Crop Scouting with Drones 13

14