Location: Water Management and Conservation Research
Project Number: 2020-13660-009-006-S
Project Type: Non-Assistance Cooperative Agreement
Start Date: Aug 1, 2022
End Date: Jul 31, 2025
Objective:
The overall goal is to develop remote sensing methods and data processing pipelines to estimate crop cover, height, water stress, leaf area index, and/or and biomass from imaging systems onboard modern satellites and small unoccupied aerial systems (sUAS). Timely collection and processing of the remote sensing data is required, because the plant growth estimates must be immediately incorporated with irrigation scheduling models and agroecosystem models used for in-season crop management decisions as part of ARS-managed field experiments. Additional efforts will focus on identifying how in-season plant growth estimates from remote sensing can be used for predicting crop yield. Specific objectives of the project are to: 1) Develop remote sensing techniques to estimate various plant growth metrics for integration with irrigation scheduling decision models, and 2) Evaluate how the RS-based plant growth estimates can also be used for crop yield prediction.
Approach:
A fleet of sUAS carrying multispectral (MS) and digital color (red, green, blue; RGB) cameras will be flown by the University of Arizona team over cotton field trials (managed by ARS) to monitor crop growth and yield. Methods for estimation of plant growth variables, such as crop height and fractional canopy cover, will be developed with a goal to incorporate the data as inputs to process-based models used for irrigation scheduling. Once the remote sensing methodologies are deemed reliable and robust, ARS field trials will be designed to include treatments with irrigation management based on the scheduling models driven by remote sensing data. Agronomic outcomes, including yield, applied irrigation, and water use efficiency, will be evaluated to demonstrate whether remote sensing can lead to improvements. Naturally, a follow-on investigation will evaluate how the in-season plant growth estimates from remote sensing can be used to predict crop yield.
The UA team will use two drones carrying MS sensors: an Inspire 2 sUAS carrying a MicaSense Altum sensor with six bands (3 visible, 1 red edge, 1 near infrared, and 1 thermal) and a Phantom 4 sUAS carrying a built-in MS sensor with similar bands to the Altum sensor, excluding the thermal band. Furthermore, the UA team will use two sUAS carrying RGB cameras: a Phantom 4 Pro V2 and a Mavic 3. All sUAS are equipped with batteries and accessories required to complete long missions (e.g., 3D missions for plant height evaluation using RGB data and low altitude missions for high resolution MS data). The UA team also possesses perpetual licenses for necessary image processing software such as Pix4Dmapper and Pix4Dfields.
Satellite data products are now becoming more reliable for use in crop management decisions. For example, Sentinel-2 data is now available globally with 5-day revisit frequency, providing 13 channels in the visible, near infrared, and shortwave infrared wavebands at 10, 20, or 60 m spatial resolution depending on the waveband. Access to these data is also being made easier through use of the Google Earth Engine platform. Another example is Planet Scope, a group of 130 satellites capable of imaging the land surface of Earth every day at 3 m spatial resolution, which now provides optical satellite data products free for researchers. As sUAS platforms will likely be impractical at the scale of commercial farms, the project will seek to transfer the methodologies for plant growth estimation from sUAS scale to satellite scale. If practical and possible, treatments that incorporate satellite-based plant growth estimates into irrigation scheduling models will be incorporated into field trials conducted by ARS.
Overall, the remote sensing methods developed in this project will be used to enhance the precision irrigation management field studies that ARS is currently conducting. The approach seeks to identify remote sensing methodologies that can provide data for irrigation scheduling decision models and improve cotton yield and/or water use efficiency.
