Wildlife, Crops, $1B Risk

Maximizing value through a combination of scientific analysis, software, and financial consulting.

A customer requested remote sensing expertise and recommendations on a project dealing with an airport and the types of agricultural crops planted around it. The interaction between wildlife and crops was the focus, specifically pertaining to reducing wildlife collisions. Annually wildlife collisions with aircraft cost nearly $1B in damages. Obstacles and risks are typically addressed through wildlife countermeasures, which require continued investments. Although deterrence countermeasures can be effective over local areas, by themselves they are not always effective over the entirety of the area of interest. Also, standard procedures for mitigation may not acknowledge the seasonal variation of risk probabilities in some areas that is related to migratory patterns and timing of crop harvest.

The target goal was to identify scenarios (and locations) with elevated risk of wildlife collision, and search for patterns among those scenarios in particular related to crop type and season. The desired outcome was to produce best-use land management practices for each specific location based on it’s local landscape and wildlife, which would inform guidance on reducing wildlife collisions through careful management of the kinds of crops planted. These recommendations required the best available analyses from detailed satellite imagery, particularly because of the important role agriculture plays in commerce and life in the area.

While the expertise and resources necessary to provide the client with a set of targeted recommendations exists, no simple “off the shelf” product exists that can immediately provide answers. This gap in knowledge exists because 1) many of the existing data sources on planted crops in the field are not updated on an annual basis and furthermore are generally used to examine the abundances of different crop types over an area rather than the exact crops planted in a specific field, and 2) while much is known about the foraging preferences of animals when studied in controlled situations, their actual foraging behavior among the complex set of circumstances in the landscapes around airports may differ, particularly depending on the lack of availability of prefered resources. These circumstances contribute to uncertainty and variability in the effectiveness of best-practice countermeasures for wildlife collisions and suggest that scenario specific guidance may be of some benefit in mitigating wildlife collisions.

The cost-effective solution provided leveraged the abundance of satellite remote sensing imagery against the complexity of the problem, and developed location specific guidelines that quantified risk and accounted for differences in crops, wildlife species, air travel, and climate.


Assessing Risk, Road Blocks, and Opportunities

No comprehensive database exists that tracks crops planted or harvested in specific fields.

Researchers instead rely on modeled information (USDA’s Crop Data Layer) that provides multi-year snapshot maps of crops. However, CDL’s were designed with the intent of tracking land cover changes over time not referencing, which crops were in a specific field at a given point in time.

While wildlife collisions are not uncommon, the risk probabilities remain very low due in large part to substantial investments in deterring wildlife and engineering planes against damage. Nonetheless, in areas with intensive agriculture these mitigation efforts come at a financial cost, which incentivises research into alternative methods of wildlife control. Furthermore, wildlife quickly recognize that many of these deterrence (scarecrow) systems do not pose an actual threat, and so they eventually ignore them and return to their original behavior.

A complementary strategy for any wildlife mitigation system is to manage the landscape around airports, such that they do not invite wildlife into the area to begin with. However, the linkages between crops, season, location, and wildlife are not so completely understood that exacting rules on crops and rotations can be applied to each airport in order to drive down collisions with complete confidence. There are several reasons for this. Control experiments (such as no crops or no vegetation) are not feasible. Failing this, comprehensive crop monitoring (by an observer) of all crops in all fields around airports does not scale-up economically. Lastly, crop classification using satellite remote sensing, although economical and efficient, is still a field of intense research especially with regard to reducing attribution error. Rather than placing initial efforts toward identifying crop types, we used an alternative approach to identify the remote sensing “signatures” of crops that were most associated with collisions, which included where such patterns were occurring throughout the landscape and through time. Our approach conserved expenditures, and focused expensive field observation or surveying on the areas of greatest concern as identified by their remote sensing signature.


Delivering Clear & Measurable Value

Our ultimate goal in consulting was to rapidly provide wildlife collision solutions that:

Contributed to improved safety over time

Were location specific and could be used to inform future site selection decisions

Were cost-effective and did not require large continual financial inputs

Were consistent with best management practices for wildlife species as well as meeting public consensus about ethical wildlife management

In approaching the problems we were hired to resolve we considered solutions that did not rely too heavily on technology alone. Rather than focusing on crop type identification from imagery and correlations to collisions we used high tech resources to provide leverage for human expertise. This is in part because although hyperspectral imagery can be used to discriminate crop types it probably will not scale-up economically in most situations. Wildlife collision risk is is actually determined by a variety of of factors including those not captured by remote sensing, such as feeding habitats, disease, weather, and population cycles. The solution we arrived at with the customer took advantage of the knowledge base of wildlife specialists who have spent years in the field dealing with animals by providing specialists with specific and targeted information about the locations of greatest concern as determined by patterns extracted from satellite imagery. We called this a field-based approach boosted by remote sensing.

The targeted approach is efficient because it filters-out areas that are not of interest based not on an exact identification of crop type but a simpler identification of the vegetation temporal signature, which may include multiple plant types but is consistently associated with collision risk. The goal basically became not to identify all the crop sources leading to collisions, but to produce continually updating maps that filter all the crop fields that are not of interest in order to prioritize the places that require attention or control measures.