Using unmanned aerial vehicles and multispectral sensors to model forage yield for grasses of semiarid landscapes

Authors: Hernandez, Alexander; Jensen, Kevin; Larson, Steven; LARSEN, ROYCE - University Of California; Rigby, Craig; Johnson, Brittany; Spickermann, Claire; SINTON, S - Avenales Ranch

Submitted to: Grasses
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/29/2024
Publication Date: 5/17/2024
Citation: Hernandez, A.J., Jensen, K.B., Larson, S.R., Larsen, R., Rigby, C.W., Johnson, B.L., Spickermann, C.S., Sinton, S. 2024. Using unmanned aerial vehicles and multispectral sensors to model forage yield for grasses of semiarid landscapes. Grasses. 3(2):84-109. https://doi.org/10.3390/grasses3020007.
DOI: https://doi.org/10.3390/grasses3020007

Interpretive Summary: Problem: It is difficult to obtain reliable and easy to replicate estimates of forage yield for grasses of the inter-mountain west of the USA due to complexities of the canopy, variations in the physical environment and limitation on acquisition of field estimates to be used as samples. Forage yield estimates are extremely important to be able to quantify many of the ecosystem services that forage grasses provide in vast rangelands of the United States of America. Accomplishment: A transparent protocol to quantify forage yield based on field forage wet weights was developed using sensors onboard unmanned aerial vehicles, and the code is provided for anybody to replicate on other areas. The significance of using a lone predictor: a volumetric 3D space is demonstrated. This is quite important because the volumetric space can be generated using red, green and blue RGB sensors which are quite affordable, thereby expanding the potential for applications to many rangeland users. Contribution to solving the problem: Users from many rangeland areas and with similar grasses to the ones used in this study can utilize the mapping protocol to apply it on their own area of interest, and their specific grasses. The only element that users will need to provide is their forage wet weights survey measurements. With the code that is provided, users can replicate and obtain results for areas of any size, thereby augmenting the opportunities for a wide audience to contribute to the understanding of forage yields and relevant ecosystem services on semiarid rangelands.

Technical Abstract: Forage yield estimates are relevant information to manage and quantify ecosystem services in grasslands. We fitted and validated prediction models of forage yield for several prominent grasses used in restoration projects in semiarid areas. We used field forage harvests from three different sites in Northern Utah and Southern California, USA in conjunction with multispectral, high-resolution UAV imagery. Different model structures were tested with quite simple models using a unique predictor: the forage volumetric 3D space, and more complex models where RGB, Red Edge and Near Infrared spectral bands and associated vegetation indices were used as pre-dictors. We found that for most dense canopy grasses using a simple linear model structure could explain most (R2 0.7) of the variability of the response variable. This was not the case for sparse canopy grasses where a full multispectral dataset and a non-parametric model approach (ran-dom forest) were required to obtain a maximum R2 of 0.53. We developed transparent protocols to model forage yield where in most circumstances acceptable results can be obtained with af-fordable RGB sensors and UAV platforms. This is important as users can get rapid estimates with inexpensive sensors for most of the grasses included in this study.