Mannequin predicts presence of animal feeding operations with 87 % accuracy
31 Could 2025
4 minute learn
Understanding the place cattle are raised is essential for managing their environmental impacts and growing technological options, however gaps in knowledge usually make it difficult to get the complete image.
Becca Muenich, organic and agricultural engineering researcher, got down to fill the hole with a brand new approach for mapping animal feeding operations.
With out correct management methods, the waste generated by these operations can pose vital ecological hurt, Muenich mentioned, similar to floor water contamination with extra phosphorus and nitrogen. Animal feeding operations are outlined as amenities that feed animals for no less than 45 days per yr in a confined space that doesn’t develop grass or forage. For Muenich, a water high quality engineer who focuses on how water strikes by means of landscapes and the way it can pollute areas by choosing up and shifting poisonous supplies, this problem piqued her curiosity.
“We will’t actually handle one thing if we don’t know the place the issue is,” mentioned Muenich, an affiliate professor with the Faculty of Engineering on the College of Arkansas and researcher for the Arkansas Agricultural Experiment Station, the analysis arm of the College of Arkansas System Division of Agriculture.
“We don’t have a very good nationwide — even at many state ranges — understanding of the place livestock are within the panorama, which actually hinders our capability to do a few of the research that I used to be eager about,” she mentioned.
Muenich mentioned there was an increase in these feeding operations in response to growing inhabitants dimension and international demand for livestock merchandise.
Contemplating key predictors of feeding operation presence similar to floor temperature, phosphorus ranges and surrounding vegetation, Muenich’s crew constructed a machine studying mannequin that may predict the situation of feeding operation places with out utilizing aerial photos. Machine studying fashions are a sort of pc program that may use algorithms to make predictions primarily based on knowledge patterns.
The mannequin was developed utilizing knowledge encompassing 18 U.S. states. The info was damaged up into particular person parcels primarily based on possession. Testing in opposition to a dataset of identified animal feeding operations, the mannequin predicted their location with 87 % accuracy.
The examine, “Machine learning-based identification of animal feeding operations in america on a parcel-scale,” was revealed in Science of the Complete Atmosphere in January.
Filling within the gaps
Earlier makes an attempt at figuring out animal feeding operations have usually relied on aerial photos, Muenich mentioned, however livestock amenities usually look completely different between states and by animal, so she and her crew aimed to make use of additional methods.
She defined the lack of knowledge surrounding livestock places usually comes from variations in how states interpret the Clear Water Act, which requires farms categorized as “concentrated animal feeding operations” to get permits by means of the Nationwide Pollutant Discharge Elimination System. These amenities are a sort of animal feeding operation with greater than 1,000 animal items.
Regardless of the nationwide regulation, states administer this allowing otherwise, resulting in variations in accessible knowledge.
For instance, Muenich constructed a watershed mannequin in an space with of Michigan and Ohio that included a number of feeding operations. Information was available by means of the pollutant elimination system for Michigan as a result of state’s allowing necessities. The identical knowledge, nevertheless, wasn’t accessible for a similar operations in Ohio, which set Muenich down this path of investigation.
Advancing in direction of a greater accounting of livestock may also help with growing methods that may enhance environmental outcomes of livestock administration whereas creating financial alternatives for farmers by means of the scaling up of applied sciences geared toward combating animal waste, Muenich mentioned. Scaling these applied sciences in economically possible methods requires information of the place livestock are most prevalent and spatially related, she defined.
Co-authors of the examine included Arghajeet Saha, previously a postdoctoral researcher on the College of Arkansas and presently an assistant scientist with the Kansas Geological Survey; Barira Rashid, Ph.D. scholar on the College of Arkansas; Ting Liu, a analysis affiliate with the College of Arkansas organic and agricultural engineering division; and Lorrayne Miralha, an assistant professor with The Ohio State College’s division of meals, agricultural and organic engineering.
The analysis was supported by the Science and Applied sciences for Phosphorus Sustainability Middle below Nationwide Science Basis award quantity CBET-2019435. The Information with Objective program from Regrid, a supply for nationwide land parcel knowledge, supplied knowledge used within the analysis.
