Rangelands are composed of patchy, highly dynamic, herbaceous plant communities that are difficult to quantify at resolutions relevant to their characteristic spatial scales. Furthermore, differentiation of these plant communities using remotely sensed observations is complicated by their similar absorption profiles. Thus, understanding and quantifying the impacts of land management and climate dynamics on vegetation change, across extensive landscapes, is challenging. To overcome these limitations, we analyzed hyperspectral data produced by NEON (1m resolution and 426 spectral bands) at a 6500 ha experimental station (Central Plains Experimental Range) to quantify vegetation change over a 5 year timescale. The high resolution of the data posed challenges from a computational standpoint (e.g. >150 billion observations). We developed a set of scripts in Python to process the data and robustly implement a set of machine learning algorithms in a distributed computing environment on a HPC system.
Results/Conclusions
The spatial resolution (1m) of the data was able to resolve plant communities at a suitable scale and the information-rich spectral resolution (426 bands) was able to differentiate closely related plant communities. The resulting plant community map showed strong accuracy results from both formal quantitative measurements (F1 75% and Kappa 81%) and informal qualitative assessments. Over a five-year timescale, we found that plant community composition was impacted more strongly by weather than by rangeland management regime. We plan to further explore linkages among fine-scale plant community mapping, management strategies (e.g. stocking rates), and cattle foraging behavior and weight gains. Our work displays the potential to map plant community types across extensive areas of herbaceous vegetation and use resultant maps to inform rangeland ecology and management. Critical to the success of the research was developing computational methods that allowed us to implement efficient and flexible analyses on the large and complex data.