Abstract

Leaf area index (LAI) is a versatile indicator of crop growth that is used to estimate evapotranspiration (ET), monitor nitrogen status, and estimate crop yield. Traditional methods for measuring LAI can be improved using high resolution remote sensing. The aim of this study was to compare approaches for estimating LAI from UAV-derived visible vegetation indices. Coincident ground-based and remotely sensed data were obtained from two irrigated wheat fields and were sampled at a total of 5 events in 2019 and 2020. Ground-based LAI was measured with a ceptometer and remotely sensed images were collected using a consumer-grade UAV. Mosaiced orthophotos were resampled from native (0.06m) spatial resolution to increasingly coarser spatial resolutions up to 3 m by either a direct or ladder resampling method. Visible band color information (RGB) was extracted from the orthophotos at the points that LAI was collected within field and 12 different visible vegetation indices (VVIs) were calculated. Linear regression was performed to evaluate the relationships between wheat LAI and each calculated VVI for all spatial resolutions and resampling methods. Three VVIs, visible atmospherically resistant index (VARI), normalized green-red difference index (NGRDI), and modified green-red vegetation index (MGRVI), estimated LAI equally well (R2= 0.66, 0.66,0.66; RMSE=0.74,0.73,0.73; MAE=0.57,0.56,0.56) when resampled to 3 m spatial resolution with the ladder resampling method. These results demonstrate the potential to remotely estimate LAI using only RGB cameras and consumer grade drones. An additional aim of this study was to evaluate use of a remotely sensed visible vegetation index to characterize the spatial variability of LAI within irrigated wheat fields. Variation of LAI was measured with a ceptometer on random nested grids at two sites with pre-determined management zones in 2019 and 2020. Coincident digital imagery was collected using a consumer-grade unmanned aerial vehicle (UAV). A visible atmospherically resistant index (VARI) LAI estimation model was applied to red, green, blue (RGB) UAV imagery using a ladder resampling approach from 0.06 m to 3 m spatial resolution. There was significant within-field spatial and temporal variation of mean LAI. For example, in May at the Grace, ID location measured LAI ranged from 0.21 to 2.58 and in June from 1.68 to 4.15. The relationship of measured and estimated LAI among management zones was strong (R2=0.84), validating the remote sensing approach to characterize LAI differences among management zones. There were statistically significant differences in estimated LAI among zones for all sampling dates (P=0.05). We assumed a minimum difference of 15% between zone LAI and the field mean for justifying variable rate irrigation among zones, a threshold that corresponds with approximately a 10% difference in evapotranspiration rate. Three of the five sampling dates had LAI differences that exceeded the threshold for at least one zone, with all three having mean LAI of less than 2.5. The VARI model for estimating LAI remotely is more effective at identifying LAI differences among management zones at lower LAI. Application of this approach has potential for applications such as estimating evapotranspiration of irrigated fields and delineation of zones for variable rate irrigation.

Degree

College and Department

Life Sciences; Plant and Wildlife Sciences

Rights

https://lib.byu.edu/about/copyright/