Crop monitoring workflow from UAV data collection to plant health indication using ENVI analytics from Harris Geospatial Solutions

Agricultural operations work on thin margins, and conventional field inspection methods are time consuming and inefficient.

Missing a small area of pest or insect infestation can result in a big loss when it comes to harvest, and over-fertilizing can be just as costly as using too little fertilizer over the long term. Analyzing airborne imagery from low cost unmanned aerial vehicles (UAV) can hereby offer an efficient and cost-effective way to aid in the assessment of field health and the estimation of crop yields, particularly when it comes to critical information for high value specialty crops such as strawberries. The following case study hereby presents a remote sensing processing workflow from UAV data collection to automated image analytics for agriculture monitoring on the example of a strawberry field using ENVI Modules from Harris Geospatial Solutions.

The test site is located near Plant City / Florida / U.S.A. This place is known as the winter strawberry capital of the world and hosts the annual Florida strawberry festival. Data collection was employed by Highland Agriculture, a US provider of state-of-the-art precision agriculture tools, using a MicaSense RedEdge multispectral camera on board of a UAV. The five spectral bands of this camera are specially targeted to agricultural applications (blue, green, red, red edge, near IR), with high spatial (8 cm GSD at 120 m AGL), radiometric (12-bit) and temporal (1 capture / sec) resolutions.

Preprocessing of the collected UAV imagery included band-to-band alignment and stacking of the individual multispectral layers into one single MultiPage Tiff file, as well as creation of an additional GPS formatted file with the sensor locations and orientations during acquisition. Using the advanced photogrammetric algorithms of ENVI OneButton the UAV images were georeferenced, orthorectified and finally mosaicked. For the orthorectification, a Digital Elevation Model was derived from the overlapping UAV images. Postprocessing consisted of atmospherically correcting the UAV orthomosaic to relative surface reflectance by removing instrument effects, the solar irradiance curve and atmospheric effects from scattering and gas absorption.


Agricultural analysis with the ENVI Precision Agriculture Module was divided in three sections. First, the number, position, and size of the individual strawberry plants were extracted from the final UAV orthomosaic. For this purpose, the Sum Green Index (SGI, after Lobell et al. 2003) was calculated. In the resulting index image, plants could be clearly distinguished from the background soil. Secondly, based on this image and an estimate of the minimum and maximum crop size, strawberry plants were automatically counted using a crop counter tool. Finally, given the results from plant counting and the post-processed multispectral UAV orthomosaic, a specific crop health tool was used to investigate the relative and absolute health distribution of the individual plants based on the Normalized Difference Vegetation Index (NDVI).

This precision agriculture workflow is suitable for diverse crops and provides farmers with a fast and accurate solution for tailored crop management to guide activity. The extracted metrics of the individual plants in the crop area help to predict yield and allow farmers to efficiently identify crop stress.

This workflow can be made turnkey for operational use and deployed to enterprise environments. For example, its analytics are implemented in Highland Hub, a web-based farm management system of Highland Precision Ag providing services and analytics for crop monitoring.

To find out more about Harris Geospatial Solution ’s precision agriculture tools, attend the company`s oral presentation at the Commercial UAV Expo and Conference Europe on 20-22 June in Brussels or visit the Harris stand (#427)