Utilizing Satellite Fusion Methods to Assess Vegetation Phenology in a Semi-Arid Ecosystem

Gallagher, Megan

doi:10.18122/td/1425/boisestate

Cited by 3 publications

(9 citation statements)

References 70 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…STARFM-a weighted function-based methods that weight the spatial and temporal aspects, thus capable of reconstructing multi-time data that have gaps and then producing images with good spatial and temporal resolution, developed by Gao et al (2006) has been widely used and has shown success in producing synthetic data like Landsat which has good spatial and temporal resolution for identifying phenological events. Similar studies utilizing STARFM have demonstrated promising results in generating images with adequate spatial and temporal resolution, particularly for detecting phenology (Gallagher, 2018;Onojeghuo et al, 2018;Son et al, 2016;Vincent, 2021). In addition, downscaling MODIS and Landsat-8 data can produce a high-quality time-series data since MODIS and Landsat have similar orbital characteristics with only 30-minutes time difference when crossing equator (Hwang et al, 2011).…”

Section: Introductionmentioning

confidence: 92%

Untitled

2024

View full text Add to dashboard Cite

Papua, known as one of the wettest regions in the world with annual precipitation ranging from 2400 to 4500 mm, faces a high risk of flooding, especially during La Niña, as observed in 2018 and 2022. Conversely, the region also experiences forest fires, mainly in the southern areas of Papua, during periods of extreme dry conditions brought about by El Niño events, as seen in 2002, 2004, and 2015. Given the increasing frequency of extreme climate events in the context of climate change, understanding the impact of global climate phenomena such as El Niño Southern Oscillation (ENSO) on Papua is crucial. This research aims to analyze the influence of ENSO and the Indian Ocean Dipole (IOD) on forest fires and flood risk in Papua, Indonesia. The analysis of forest fires utilizes MODIS hotspot and ERA5 precipitation data, employing quantitative modeling techniques such as Lasso and Elastic Net Regression. It integrates both ENSO and IOD indices into the precipitation indicators. In contrast, flood analysis is carried out through distribution and joint pattern analysis. The Elastic Net Regression yields promising results in modeling, with more than 96% of the tested models successfully predicting the total annual hotspots for each year, achieving an R-squared value of 90%. This suggests that the method and algorithm used can serve as a robust model for early hotspot prediction in the analyzed area. The warm phases of ENSO and IOD consistently exhibit a positive correlation with the dry season. However, the cold phases of these phenomena do not significantly impact heavy and extreme precipitation indicators in the studied region. The flood analysis reveals that La Niña has only a slight effect on all three precipitation patterns in the analyzed area, primarily by increasing the risk of extreme precipitation indicators. Conversely, negative IOD demonstrates inconsistency across all three precipitation patterns in Papua.

show abstract

Section: Introductionmentioning

confidence: 92%

Untitled

2024

View full text Add to dashboard Cite

show abstract

“…They generally rely on platforms such as MODIS which trade high revisit frequency with spatial and spectral resolution [Iiames, 2010]. Other methods relate the temporal resolution of MODIS with the spatial and spectral qualities of Landsat [Gallagher, 2018]. Yet others 'reconstruct' time-series data by filling gaps between observations or by fitting mathematical functions to the observations [Zhou et al, 2016].…”

Section: Phenology and Time-seriesmentioning

confidence: 99%

“…Known spectral endmembers ('pure' pixels, either from known locations in the images, or a reference signal) can then inform unmixing of the pixel into its likely constituents. This process has been used with remarkable success, even in shrub-steppe environments [Poley, 2017] It may not be possible to pick a singular date that can minimize confusion between all classes since phenologies of semi-arid vegetation can be disparate [Gallagher, 2018].…”

Section: Phenology and Time-seriesmentioning

confidence: 99%

See 1 more Smart Citation

Remote Sensing Time-Series Analysis, Machine Learning, and K-Means Clustering Improves Dryland Vegetation and Biological Soil Crust Classification

Enterkine¹

View full text Add to dashboard Cite

Dryland and semi-arid vegetation communities, although appearing to the casual observer as relatively simplistic and homogeneous, are in fact the opposite. Upon further inspection, semi-arid vegetation is highly complex and heterogeneous at almost any scale. The same holds true for biological soil crust. Growing concern about global changes in climate, nutrient cycles, and land use have required increasing scrutiny of our understanding of these communities and all of their constituents, as we seek to improve forecasting models and inform land management decisions. This thesis aims to provide insight to the paradigm of how we create and interpret vegetation classifications in a semi-arid ecosystem. In the first chapter, I examine the potential of new remote sensing imaging platforms in combination with machine learning algorithms and cloud computing as they apply to time-series analyses for vegetation classification. The results of this indicate that sinusoidal approximations ("Harmonic Models") of vegetation indices are able to predict vegetation cover with nearly the same accuracy as monthly composites, and that a combination of both perform no better than either. Additionally, I examine how assigning classes to training data (e.g. species-level, plant functional type) influence the classification accuracy, interpretability, and potential uses. Stricter class membership requirements at increasingly aggregated scales (e.g. PFT) lead to greater accuracies. Finding the implication of this conclusion unsatisfactoryat the extreme, everything was either cheatgrass or bare, the shrub class succumbing almost entirely to vii errors of omission-I investigated approaching other methods to assign classes to field data that captured more of the realities of semi-arid vegetation within our study area. To this extent, k-means clustering was used to determine what community classes were present in the field data. The outcome of this approach is a class where each potential constituent cover has a known distribution. Overall accuracies were found to be lower for this approach. However, the classification outcomes quantify overlapping distributions of cover types (e.g. 'sagebrush' or 'shrub') between classes. These accuracies are assessed using 'fuzzy' confusion matrices. This enables more information to be preserved through the remote sensing classification process, and reserves more interpretation for the map user than a typical 'hard' classification. Importantly the distributions of cover types are likely most representative of field conditions and thus more useful to land managers making holistic decisions about restoration or fuel management. For the second chapter, I delved deeper into the potentials of new remote sensing and computing platforms to predict biological soil crust cover. The growing field of research on biological soil crust points to potentially significant implications for nutrient and water cycling, in addition to positive effects on native vascular vegetation. However, spatial data are lacking due to remot...

show abstract

“…Machine-learning techniques within GEE can be leveraged to improve the accuracy of species classification by finding trends in remotely sensed data (Gallagher, 2018). Random forest (RF) is an ensemble machine learning algorithm that can be used to take training data and create either a classification or regression model (Belgiu & Drăgu, 2016;Pavlov, 2019).…”

Section: Google Earth Engine For Classificationmentioning

confidence: 99%

A Sense of Scale

Vermillion¹

View full text Add to dashboard Cite

The native vegetation communities in the sagebrush steppe, a semi-arid ecosystem type, are under threat from exotic annual grasses. Exotic annual grasses increase fire severity and frequency, decrease biodiversity, and reduce soil carbon storage amongst other ecosystem services. The invasion of exotic annual grasses is causing detrimental impacts to land use by eliminating forage for livestock and creating a huge economic cost from fire control and post-fire restoration. To combat invasion, land managers need to know what exotic annual grasses are present, where they are invading, and estimates of their biomass. Mapping exotic annual grasses is challenging because many areas in the sagebrush steppe are difficult to access; yet field measurements are the main method to identify and quantify their existence. In this study, we address this challenge by exploring the use of both landscape-scale and plot-scale observations with remote sensing. First, we use satellite imagery to map where exotic annual grasses are invading and identify the native species which are being encroached upon. Second, we investigate the use of fine-scale imagery for non-destructive measurements of biomass of exotic annual grasses. Understanding the location of exotic annual grasses is important for restoration efforts, e.g. large swath (~100m) herbicide spraying. Restoration efforts are expensive and often ineffective in areas already dominated by exotic annual grasses. Early detection of exotic annual grasses in sagebrush and native grasses communities will increase the chances of effective ecosystem restoration. We used Sentinel-2 satellite imagery in Google Earth Engine, a cloud computing platform, to train a random forest (RF) machine learning algorithm to map vegetation in ~150,000 acres in the sagebrush steppe in southeast Idaho. The result is a classification map of vegetation (overall accuracy of 72%) and a map of percent cover of annual grass (R2 = 0.58). The combination of these two maps will allow land managers to target areas of restoration and make informed decisions about where to allow grazing. In addition to knowing what exotic annual grasses exist and their percent cover, detailed information about their biomass is important for understanding fuel loads and forage quality. Structure from Motion (SfM) is a photogrammetry technique that uses digital images to develop 3-dimensional point clouds that can be transformed into volumetric measurements of biomass. The SfM technique has the potential to quantify biomass estimates across multiple plots while minimizing field work. We developed allometric equations relating SfM-derived volume (m3) to biomass (g/m2) for a study area in southeast Oregon. The resulting equation showed a positive relationship (R2 = 0.51) between the log transformed SfM-derived volume and log transformed biomass when litter was removed. This relationship shows promise in being upscaled to larger surveys using aerial platforms. This method can reduce the need for destructively harvesting biomass, and thus allow field work to cover a greater spatial extent. Ultimately, increasing spatial coverage for biomass will improve accuracy in quantifying fuel loads and carbon storage, providing insights to how these exotic plants are altering ecosystem services.

show abstract

Utilizing Satellite Fusion Methods to Assess Vegetation Phenology in a Semi-Arid Ecosystem

Cited by 3 publications

References 70 publications

Untitled

Untitled

Remote Sensing Time-Series Analysis, Machine Learning, and K-Means Clustering Improves Dryland Vegetation and Biological Soil Crust Classification

A Sense of Scale

Contact Info

Product

Resources

About