COMPARISON OF REMOTE SENSING TECHNOLOGIES FOR EARLY BLIGHT DETECTION ACROSS VARIETIES IN POTATO

Early blight, caused by Alternaria solani, is a worldwide problem in potato production and has the potential to create quality reduction and yield loss of up to 30% annually. While early blight has been a perennial disease concern in potato production of Wisconsin and other climatic regions of the globe, many factors are currently heightening disease risk, including: increasing concentration of potato production lands and reduced crop rotational capacities in commercial production, limited advancement of host resistance development, reduced availability of effective fungicides concurrent with industry and consumer desire to minimize pesticide reliance in food crop production, and increased spore production of Alternaria species resulting from global atmospheric carbon dioxide increases. Currently, no commercial potato cultivar is resistant to early blight, so management primarily relies on regular applications of one or more fungicides.The detection of disease in commercial potatoes is typically made by professional crop scouts or consultants, as well as growers, during weekly field inspections. Early blight is relatively easy to visually diagnose when necrotic symptoms develop, and is commonplace in most central and southern Wisconsin potato fields; however, its precise onset prior to visual symptom development and progress, as well as its direct resulting crop impact, by variety, is poorly understood and infrequently harnessed in prescriptively managing this costly disease. Enhanced, varietally specific, and more thorough field detection of earliest plant responses to early blight enables development of a tailored disease control approach to limit advancement of this polycyclic disease both within and between fields. Our primary objective is to test several emerging remote sensing technologies - including hyperspectral imaging and several sensor types on drones - for the early detection of early blight infected potatoes.Remote sensing has gained increasing traction in recent years as a tool for detection of diseases in crops. The literature is vast, leading to calls for global surveillance systems for detection of diseases in crops. The existing literature has documented that remote sensing is an effective tool for disease detection, but also points to the need for continued research to both identify the optimum set of measurement properties (e.g., sensor type and timing during the growth cycle) to develop effective, generalizable remote sensing algorithms. Some of the key limitations that remain include:1) comparisons among technologies and testing multi-sensor synergies;2) insufficient evaluation of differing responses among cultivars of a crop; and3) replication across years with differing weather conditions.In this research, we will address these limitations. We will test 4 approaches for early blight detection in potato. We will employ full-range hyperspectral imagery (400-2500 nm) from an airplane, as well as VNIR (visible-near infrared, 400-900 nm), 5-band multispectral from a drone, and thermal infrared (plant temperature) from a drone. Embedded in these four technologies are three other tests: aircraft vs. drone data, full-range vs VNIR hyperspectral imagery, and spectral vs. vegetation structure (i.e., height, derived via structure-from-motion) from multispectral imagery. Using field trials of diseased and protected/non-diseased potatoes from 6 varieties, we will also test whether different varieties show differing responses to infection in remotely sensed data. Finally, we will use the results of the analyses of remotely sensed data to better characterize the biological drivers behind the detectability of early blight across our experiments and across years.