THAI NGUYEN UNIVERSITY UNIVERSITY OF AGRICULTURAL AND FORESTRY `“ CC ——= /T Di 1⁄4 a TRAN HOANG SON DIFFERENTIATING CRESTED WHEATGRASS (AGROPYRON CRISTATUM) IN SASKATCHEWAN LANDING PROVINCIAL PARK, CANADA WITH REMOTE SENSING BACHELOR THESIS Study Mode: Full-time Major : Environmental Science and Management Faculty : International Programs Office Batch : K47 Thai Nguyen, May 2020 DOCUMENTATION PAGE WITH ABSTRACT Thai Nguyen Ủniversity of Agriculture and Forestry Degree Program Bachelor of Environmental Science and Management Student name Tran Hoang Son Student ID DTN1554290027 Thesis Title Differentiating Crested Wheatgrass (Agropyron Cristatum) in Saskatchewan Landing Provincial Park, Canada with Remote Sensing Supervisor (s) Assoc. Tran Van Dien Supervisor’s Signature Abstract: Invasive species are a topic of interest to researchers, soil resource managers and environmentalists. Exotic grass species threaten the biodiversity and carrying capacity of an ecosystem, and cause negative impacts on the livelihoods of native plants. To date, there has been no effective method for detecting exotic species in mixed prairie ecosystems.
This project focuses on the use of remote sensing data (biophysical, hyperspectral) to distinguish Crested Wheatgrass (CW) (Agropyron cristatum) in Saskatchewan Landing Provincial Park (SLPP), Saskatchewan, Canada. The objectives are to: 1) Investigate the spectral characteristics of Crested Wheatgrass from ground hyperspectral data; 2) Find the best vegetation indices to differentiate Crested Wheatgrass; and 3) Investigate the behavior of Crested Wheatgrass in different Grazing regimes. Hyperspectral data were used to indicate the spectral characteristics of Crested Wheatgrass. Besides, reflectance at red and near infrared and green wavelengths were used to calculate 5 Vegetation Indices: NDVI (Normalized Difference Vegetation Index), RDVI (Renormalized Difference Vegetation Index), SAVI (Soil Adjusted Vegetation Index), MSAVI (Modified Soil Adjusted Vegetation Index), PSRI (Plant Senescence Reflectance Index).
The results of Tukey Post Hoc Test showed that NDVI is the best Vegetation Index to distinguish Crested Wheatgrass but using Vegetation Indices is not the optimal method to evaluate behavior of Crested Wheatgrass in different grazing regimes in this study. Invasive species, Native Plants, Crested Wheatgrass (Agropyron Keywords: cristatum), Saskatchewan Landing Provincial Park, Hyperspectral Data, Wavelength, Reflectance, Vegetation Indices, NDVI, RDVI, SAVI, MSAVI, PSRI. Number of pages: | 55 Date of IMay 15, 2020 Submission: 1 ACKNOWLEDGEMENT First and foremost, I wish to express my endless thanks and gratefulness to my supervisor Dr. Her kind support and continuous advices went through the process of completion of my thesis at University of Saskatchewan.
I thank Saskatchewan Landing Provincial Park for their cooperation and support with this project in providing an ideal location for field data collection. Moreover, I want to express my gratefulness to Ms. Thuy Doan for her enthusiastic help in providing valuable documents. Last but not least, I would like to give my special thanks to my parents for their endless love, care and having the mental assistance and motivation throughout my whole life.
Thai Nguyen, May 2020 TRAN HOANG SON ili TABLE OF CONTENT LIST OF FIGURES. vi LIST OF ABBREVIATIONS o.oo cece esesseecseseeeeseeesssesesssesseessessssssesenesseeseeeaes Vili CHAPTER I. cecscseeecseseeecscseeesscsesesessesessesseenseesea 1 In co .-- c6 1S S1 11 1 91T HH TH TH TH TH ngàn Hàn TH Hi 5 PP Cà.---- «kg TH Tàn HH HH rà 7 2. Crested Wheaftgrass CharaCf€TISEICS.
- + HT TH HT ri, 8 CHAPTER IIL. - 6 E11 S11 TT HT TH HT TH Tàn Hàn nhà 10 3. Là HH HH HH HH TH TH TT HT TH HT TH TT TH TH TH tre 10 3. Dafa COlÏ€CfIOH.
Án HT HT HH TH HT TT TT HT HH TH it 11 3. 6 (121211211 1 1 3121 H1 HT HH TT HT TH 11 3. Ground hyperspectraÌ (Ì4fa.--‹- - ¿+ + 1S 11T HT HH TH HH nhiệt 15 3. Pre — PLOCESSING nh ốc ố ố ốằ.
Dafa DTOC€SSITE. TT HT HH TT TH TT HT TH TT HT TH Hiệp 16 3. Hyperspectral data analysis oo. cece sees ceseeseseseescsesessseessesessssesesessesseeegs 16 iv 3.
Vegetation Indices caÏCuÏafIOTI. ¿6S 11t 1S 1T TH HH HH ri, 17 3. án TH HT HH TH HT HH TH TH HT HH Hi 19 CHAPTER IV: RESULTS AND DISCUSSION. Spectral characteristics of Crested WheatðTaSS.
The best vegetation indices to differentiate Crested Wheaftgrass. The behavior of Crested Wheatgrass in different Grazing regimes. Limitation and futher SfUYy .- «+ St 11B 1n TT HH nành 32 5. 41 FY235))00I00:TTĐee.
44 LIST OF FIGURES Figure 1. Remote sensing process (researchgaf€. Location of Saskatchewan Landing Provincial Park. Saskatchewan Landing Provincial Park Grassland.
Sampling design 1n each SI{©.--- ‹- + 11k HT HH TH ri, 12 Eigure 5. Designing the sampÏing aT€a. - ¿+ + 1E E111 1 111 11T TT TH Hàn Hàn ghế 14 Eigure 7. - -- 6 + kh TT HH TH HT TH HT HH TH HT HH ngà 14 Figure 8.
Sage in SLPP oo.-- -- 6 + 11111 H111 11T TT HH TH TT HH ngà 14 Figure 10. Crested Wheatgrass and grazing activity in SLPP. Spectral characteristic of CW and 4 non-grass types in SLPP. Five Vegetation Indices of CW & 4 non-grass types showed on graph.
Vegetation Indices of 10 sites study indicated on graph. ---- 27 vi LIST OF TABLES Table 1. List of Hyperspectral Vegetation Indices have been used. Average reflectances at wavelength 800, 670 and 500 ofŠ grass types.
Vegetation Indices of CW & 4 non-gTaSS fyp€S. cà Sex, 21 Table 4. Comparisons of NDVI, RDVI, SAVI and MSAVI (CW and 4 non-grass types) .- kh TH TT TT TT HT HH TH ngư 24 Table 5. Boxplots for 4 Vegetation Indices of CW and 4 non-gTass types.
Vegetation Indices for 10 s1f€S SfUỦy. Grazing & Dominant specIes Information of 10 sites study in SLPP. Boxplots for 5 Vegetation Indices of IŨ sites study. ---‹--<<<5+ 30 vii LIST OF ABBREVIATIONS SLPP: Saskatchewan Landing Provincial Park CW: Crested Wheatgrass Vis: Vegetation Indices NDVI: Normalized Difference Vegetation Index RDVI: Renormalized Difference Vegetation Index SAVI: Soil Adjusted Vegetation Index MSAVI: Modified Soil Adjusted Vegetation Index PSRI: Plant Senescence Reflectance Index vi CHAPTER I.
Rationale Non-native species are a threat to agricultural and native prairie communities around the world. The existence of invasive species affected negatively on the native species population, crowding out the existence of native plants. Invasive species are any species that do not originate from that ecosystem and are capable of self-propagation, have introduced or likely caused harm to the environment and other factors (Pejchar and Mooney, 2009). Non-native species are also affected by climate change.
Predicted environmental changes, such as changes in rainfall and temperature, nutrient content and soil disturbance, may increase habitat sensitivity to non-native plants (Hufnagel and Garamvélgyi, 2014). When non-native plants invade the environment, they are able to overcome native plants through direct or indirect competition (Robert et al. A study by Driscoll et a/. (2014) shows that environmental weeds are non- native plants that establish in natural areas, often harming to the native plants, damage to ecosystem function and cost billions of dollars to manage each year.
Crested Wheatgrass threatens native ecosystems, which are important reservoirs of biodiversity (Mooney and Drake, 1989; D’ Antonio and Vitousek, 1992). This species impacts both fauna and flora (Sutter and Brigham, 1998; Heidinga and Wilson, 2002). They also affect nutrition and energy sources (Christian and Wilson, 1999). Non-native species are a threat to agricultural and native prairie communities around the world.
The existence of invasive species affected negatively on the native species population, crowding out the existence of native plants. Invasive species are any species that do not originate from that ecosystem and are capable of self-propagation, have introduced or likely caused harm to the environment and other factors (Pejchar and Mooney, 2009). Non-native species are also affected by climate change. Predicted environmental changes, such as changes in rainfall and temperature, nutrient content and soil disturbance, may increase habitat sensitivity to non-native plants (Hufnagel and Garamvélgyi, 2014).
When non-native plants invade the environment, they are able to overcome native plants through direct or indirect competition (Robert et al. A study by Driscoll et a/. (2014) shows that environmental weeds are non- native plants that establish in natural areas, often harming to the native plants, damage to ecosystem function and cost billions of dollars to manage each year. Crested Wheatgrass threatens native ecosystems, which are important reservoirs of biodiversity (Mooney and Drake, 1989; D’ Antonio and Vitousek, 1992).
This species impacts both fauna and flora (Sutter and Brigham, 1998; Heidinga and Wilson, 2002). They also affect nutrition and energy sources (Christian and Wilson, 1999). Objectives The purpose of this study is to see whether remote sensing can be an effective method for detecting Crested Wheatgrass in native mixed-grass prairies. Until now, an effective method has not been developed to detect non- native plants in mixed grassland ecosystems with medium-resolution imagery.
At an affordable cost, it will be suitable for resource managers to work with. The objectives are: a. Investigate the spectral characteristics of crested wheatgrass from ground hyperspectral data; b. Find the best vegetation indices to differentiate Crested Wheatgrass; c.
Investigate the behavior of Crested Wheatgrass in different Grazing regimes. Remote sensing technique 2. Remote Sensing Remote sensing is the collection of information about an object or phenomenon that does not physically contact the object and is therefore the opposite of local observation. In modern use, this term often refers to the use of aerial sensor technologies to detect and classify objects on Earth (both on the surface and in the atmosphere and oceans) by means of signal transmission (for example, electromagnetic radiation).
It can be divided into active remote sensing (when the first signal is emitted from aircraft or satellites) or passively (for example, sunlight) when information is only recorded (Curran, 1985). Sensor B Sensor A Sun ~~ YE AVA Iđ)- đc „8O —— Processing station Target Figure 1. Remote sensing process (researchgate.net) An advanced image synthesis technique recently developed to provide a means to detect the structure and functional properties of invasive plants at different canopy levels is the integration of passive energy. and actively collected at the same time by image spectrometer and scanning-waveform light detection and ranging system (LIDAR) (Huang and Gregory, 2009).
A study by Ustin et al. show map based on high spectral (224 10 nm bands) and spatial (/ spl sim / 4 m) resolution spectra of several invasive species, including eggplant, jubata, fennel and giant reed from a range of habitats at Camp Pendleton and Vandenberg Air Force Base in California using AVIRIS data (2002). Through the advent and dissemination of imaging systems born in air and space, researchers have been able to measure the physical and chemical properties of surface geology of the earth and planets, monitoring changes in plant biomass of continents, monitoring global migration of water, measuring surface heat flux, visualizing deformation of the earth's surface due to human processes and nature and many other applications (Beasley and Barnhart, 2017). In the study of northern mixed-grass prairie in Canada, Zhou and Guo used SPOT-5 imagery to detect Crested Wheatgrass invasion.
The results of their study showed that a single-date SPOT-5 imagery with a resolution of 10 m would be useful in distinguishing CW from native species in mixed grasslands (2007). Using remote sensing can be a cost effective strategy to detect invasive weeds. Historically, invasive species have often been identified by natural resource managers and volunteers who manually scouted (Shaw, 2005). Hyperspectral data Hyperspectral data refers to the analysis and measurement of the reflection, transmission or absorption of electromagnetic radiation with very high spectral resolution (Lukas et al.
According to Zebin et al., the large dimensions and mass and hundreds of contiguous spectral channels are characteristic of hyperspectral remote sensing images. These images obtained from the Earth's surface contain a variety of information about space, radiation and spectrum, which helped a lot for researchers in analyzing, processing and monitoring information on the Earth's surface (2016). Hyperspectral remote sensing can form a ground surface reflection images at several hundred wavelengths simultaneously, with wavelengths ranging from 0.