Evaluating the Utility of Remote Sensing Time Series Analysis for the Identification of Grassland Conversions in Alberta, Canada by Jacob Mardian A Thesis presented to The University of Guelph In partial fulfillment of requirements for the degree of Master of Science in Geography Guelph, Ontario, Canada © Jacob Aaron Mardian, September, 2020 ABSTRACT EVALUATING THE UTILITY OF REMOTE SENSING TIME SERIES ANALYSIS FOR THE IDENTIFICATION OF GRASSLAND CONVERSIONS IN ALBERTA, CANADA Jacob Aaron Mardian Advisor: Dr. Aaron Berg University of Guelph, 2020 Committee Member: Dr. Bahram Daneshfar Grasslands are an important source of ecosystem services and play a critical role in climate regulation. However, this biome is threatened by agricultural expansion and intensification. Remote sensing offers a unique opportunity to monitor these grassland to cropland conversions through the collection of data at various spatial and temporal scales. This research examines the applicability of remote sensing time series analysis as a tool for identifying grassland to cropland conversions in Alberta, Canada. The Breaks for Additive Seasonal and Trend (BFAST) Seasonal method was the most effective model tested, identifying the correct year of change for 76% of rangeland to cropland conversions and 66% of pasture to cropland conversions. However, the results were strongly dependent on model parameterizations, the datasets used, and the subsequent crop planted. Overall, this research provides a new data-driven approach for identifying grassland to cropland conversions and can be used to improve monitoring of Canada’s grassland resources. iii ACKNOWLEDGEMENTS I would like to thank Agriculture and Agri-Food Canada, NSERC, the University of Guelph and the Canada First Research Excellence Fund: Food from Thought Initiative for providing funding for the research completed in my thesis. I would like to express my gratitude to my advisor, Dr. Aaron Berg, and committee member, Dr. Bahram Daneshfar, for their invaluable guidance, support and expertise over the last two years. I also appreciate the staff, faculty and students within the Department of Geography, Environment and Geomatics at the University of Guelph who have provided assistance and encouragement throughout this process. Finally, thank you to my friends, family and partner Emma for your continual support. iv TABLE OF CONTENTS ABSTRACT ........................................................................................................................ i ACKNOWLEDGEMENTS ............................................................................................ iii LIST OF ACRONYMS .................................................................................................... vi LIST OF FIGURES ........................................................................................................ vii LIST OF TABLES ......................................................................................................... viii Chapter 1.0: Introduction ................................................................................................. 1 1.1 Background .............................................................................................................. 1 1.2 Research Aim and Objectives ................................................................................. 3 1.3 Thesis Outline .......................................................................................................... 3 Chapter 2.0: Literature Review ....................................................................................... 5 2.1 Introduction ............................................................................................................. 5 2.2 Overview of Grasslands .......................................................................................... 5 2.2.1 Significance of Grasslands ................................................................................. 6 2.3 Grassland MonitorinG ............................................................................................. 7 2.4 Remote Sensing Platforms for Grassland MonitorinG ......................................... 8 2.4.1 Multispectral Sensors ......................................................................................... 8 2.4.2 Radar Sensors ................................................................................................... 10 2.4.3 Multisource Data and Image Fusion ................................................................. 12 2.5 Vegetation Phenology and Spectral Band Indices .............................................. 14 2.6 Time Series SmoothinG .......................................................................................... 16 2.7 Change Detection ................................................................................................... 18 2.7.1 Threshold-Based Change Detection ................................................................. 18 2.7.2 Harmonic Analysis ........................................................................................... 19 2.7.3 Structural Break Methods ................................................................................. 20 2.7.3.1 BFAST ........................................................................................................ 21 2.7.3.2 BFAST Monitor ......................................................................................... 22 2.7.3.3 BEAST ........................................................................................................ 23 Chapter 3.0: Structural Breaks in Earth Observation Time Series Data for Grassland Change Detection in Alberta, Canada ........................................................................... 24 Abstract ........................................................................................................................ 24 3.1 Introduction ........................................................................................................... 25 3.2 Methodology ........................................................................................................... 30 v 3.2.1 Study Areas ...................................................................................................... 30 3.2.1.1 Grassland Conversion Polygons ............................................................... 32 3.2.2 Image Acquisition and Pre-Processing ............................................................. 34 3.2.2.1 MODIS ....................................................................................................... 34 3.2.2.2 Landsat ...................................................................................................... 34 3.2.3 Time Series Smoothing .................................................................................... 35 3.2.4 Change Detection ............................................................................................. 38 3.2.4.1 BFAST ........................................................................................................ 38 3.2.4.2 BEAST ........................................................................................................ 40 3.2.4.3 Parameter Tuning ...................................................................................... 41 3.2.4.4 Influence of Data Source and Lambda ...................................................... 42 3.2.5 Accuracy Assessment ....................................................................................... 43 3.3 Results and Discussion .......................................................................................... 43 3.3.1 Method Comparison ......................................................................................... 44 3.3.2 BFAST Parameterization .................................................................................. 45 3.3.2.1 BFAST Seasonal ........................................................................................ 45 3.3.2.2 Detrending ................................................................................................. 47 3.3.2.3 Empirical Fluctuation Process .................................................................. 48 3.3.2.4 Harmonic Order ........................................................................................ 49 3.3.3 Data Source Comparison and Lambda ............................................................. 51 3.3.4 Change Detection Timing ................................................................................. 53 3.3.5 Effect of Crop Type on Change Detection ....................................................... 55 3.3.6 Limitations ........................................................................................................ 56 3.4 Conclusions ............................................................................................................ 56 Chapter 4.0: Conclusions ................................................................................................ 58 References ........................................................................................................................ 60 vi LIST OF ACRONYMS AAFC – Agriculture and Agri-Food Canada ACI – Annual Crop Inventory BEAST – Bayesian Estimator of Abrupt change, Seasonal change and Trend BFAST – Breaks For Additive Seasonal and Trend EFP – Empirical Fluctuation Process EVI – Enhanced Vegetation Index LCLUC – Land Cover and Land Use Change MODIS – Moderate Resolution Imaging Spectrometer