Projects Efforts

Influence of cover crops on cash crops growth and development: To assess the influence of no cover crop vs. cover crops on cash crops plant health, unmanned aerial vehicle (UAV) images were collected using multiple sensors (Micasense Rededge, LIDAR, HSI, and DJI RGB), at multiple times (18 times on cover crops; 24 times on cash crops) and multiple altitudes (200' and 400') during the crop growth life cycle covering both of cover crop and cash crops. To verify the results of UAV images, a total of 10 (corn) to 15-week (cotton) destructive samplings were collected across treatments from planting to maturity. The supervised and unsupervised classification algorithms are currently standardizing to identify potential VIs that best differentiate the plant stands, plant health and predicting yield in response to cover cropping treatments under rainfed environments (Bheemanahalli et al., 2020; Bheemanahalli et al., 2021).

Early detection of root-knot nematode infestation in cotton using hyperspectral data: Temporal hyperspectral signatures were collected to study the effect of root-knot nematode (RKN) (Meloidogyne incognita) on leaf reflectance under controlled environmental conditions. The supervised learning algorithms to classify the RKN infested cotton from the control group using hyperspectral signatures. We found that the 350-1000nm range bands are good enough to produce a reasonable classification, encouraging because commercially available drone mountable hyperspectral sensors produce imagery only in this range. The effect caused by the RKN on the root system of cotton can be non-invasively diagnosed using hyperspectral data at the early growth stage (Samiappan et al., 2021).

Impacts of temperature on cover crops vegetative growth and development: Extreme temperatures during the early seedling stage significantly affected the growth and development of cover crop biomass accumulation. We identified cardinal temperatures and functional algorithms for growth and developmental traits of four cool-season cover crops (Cereal rye [Secale cereale], crimson clover [Trifolium incarnatum], triticale [Triticum x Secale], and Winter wheat [Triticum aestivum] and one warm-season cover crop [Mighty Mustard Pacific Gold, Brassica juncea] using controlled-environment temperature conditions (Munyon et al., 2021). Temperature optimum (Topt) for shoot and root traits varied from 23.9 to 26.5°C and from 22 to 25.7°C, respectively. On average, the Topt for root traits was significantly lower than shoot traits in four out of five species. Our results show that extreme temperatures (low and high) negatively affect the growth and development of cover crops. For colder (sub-optimal) climatic conditions, cereal rye would usually be the best species to grow. At warmer climatic regions (Topt and above), crimson clover and mighty mustard pacific gold may yield higher biomass and be the best selections. However, in all treatments, mighty mustard pacific gold was top among the five species in root and shoot growth. With the results presented here, a producer could choose potential species to grow based on local climatic conditions (Munyon et al., 2021).

Advanced machine learning algorithms for soil moisture estimation from hyperspectral imagery data: Soil moisture has a major impact on vegetative growth. Assessing soil moisture is crucial in developing an effective management of farmland irrigation to improve crop yield. This research explored advanced machine learning algorithms to estimate soil moisture from hyperspectral imagery data. Hyperspectral measurements of the reflectance of illumination varies when different amounts of water are presented in the soil. Hyperspectral sensors mounted on UAS allow us to collect high-spatial resolution imagery data over research fields. We have developed and evaluated data-driven machine learning algorithms to learn representative features (i.e., dimensionality reduction) and make predictions (i.e., soil moisture estimation) from high-dimensional hyperspectral images. The success of this research will enable us to estimate the soil moisture for a large area in an automatic and robust manner.

Applying radio frequency (RF)/ microwave remote sensing from UAS to map soil moisture: This project aimed to answer two fundamental questions. Can low-frequency Signals of Opportunity (SoOp) be used to reliably map soil moisture at surface and root-zone level in irrigated and rainfed farms at high spatiotemporal resolution? If so, can low-cost, and ubiquitous platforms (i.e., smartphones and drones) be leveraged to use the SoOp approach in a way that is immediately available for use in agriculture-based societies? To answer these questions, researchers applied radio frequency (RF)/ microwave remote sensing from unmanned aerial systems (UAS) to map soil moisture at Mississippi State University’s North Farm. A comprehensive UAS-based RF testbed was developed using reflectometry from smartphone and GPS-receivers. UAS-based RF uniters were flown on a regular basis (twice a day, Monday – Friday, over a yearlong period). The testbed was accompanied with proximal sensing via unmanned ground vehicles that acquired in-situ soil moisture and vegetation geophysical parameters to provide appropriate datasets for training and testing physics aware, machine learning-based models. The primary goal is to enable non-intrusive high-resolution soil moisture estimates at multiple depths of soil via UAS-based microwave instruments.