Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

48
Orchard LAI Estimation and Land-use Correlation using Geospatial Technology By By Dr. Sudhanshu Sekhar Panda Dr. Sudhanshu Sekhar Panda Associate Professor, GIS/Environmental Science Joshua Nolan & Lee Irminger Joshua Nolan & Lee Irminger Undergraduate Student Institute of Environmental Spatial Analysis

description

Orchard LAI Estimation and Land-use Correlation using Geospatial Technology. By Dr. Sudhanshu Sekhar Panda Associate Professor, GIS/Environmental Science Joshua Nolan & Lee Irminger Undergraduate Student Institute of Environmental Spatial Analysis. Background. - PowerPoint PPT Presentation

Transcript of Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Page 1: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Orchard LAI Estimation and Land-use Correlation using Geospatial

TechnologyByBy

Dr. Sudhanshu Sekhar PandaDr. Sudhanshu Sekhar PandaAssociate Professor, GIS/Environmental Science

Joshua Nolan & Lee IrmingerJoshua Nolan & Lee Irminger

Undergraduate Student

Institute of Environmental Spatial Analysis

Page 2: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

BackgroundBackground

Major quantity of soil and plant available water is lost by evapotranspiration (ET).

Most Irrigation scheduling for crop is conducted based on ET. The loss of water due to ET varies with different crops

including horticultural plants. It is difficult to estimate ET for large orchards. Hydrologic parameters like stomatal conductance, soil

moisture, leaf area index (LAI), plant canopy temperature, and wind velocity are functions of plant ET.

These hydrologic attributes together can be modeled to estimate ET.

Page 3: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

BackgroundBackground

Remote sensing has the proven ability for measuring some of Remote sensing has the proven ability for measuring some of those parameters in quick and cost-effective mannerthose parameters in quick and cost-effective manner

Hydrologic parameters like stomatal conductance, soil moisture, LAI, and plant canopy temperature can be estimated on a spatial basis using remotely sensed imagery.

Horticultural plants of different species and growth stages Horticultural plants of different species and growth stages demonstrate different values for these hydrologic parameters demonstrate different values for these hydrologic parameters that ultimately help estimate the ET.that ultimately help estimate the ET.

Geospatial Technology has the ability to decipher those plants of different species and growth stages in an orchard.

Page 4: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

ObjectivesObjectives

The objectives of this study is to 1. Estimate LAI of a blueberry orchard (Z-Blu

Farm) and surrounding windbreaker forest cover using high resolution orthoimagery.

2. Conduct the land-use classification or orchard speciation using Object-based Image Analysis (OBIA) technique.

3. Correlate the LAI to different species present in the orchard.

Page 5: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology
Page 6: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Study AreaStudy Area

Z-Blu Farm (130 acres blueberry Z-Blu Farm (130 acres blueberry plantation) in Nahunta, GAplantation) in Nahunta, GA

Very Well managed orchardVery Well managed orchard Two varieties of blueberry – High Two varieties of blueberry – High

Bush and Rabbit Eye and 3 stages of Bush and Rabbit Eye and 3 stages of blueberry plants due to expansion of blueberry plants due to expansion of the orchard since 2004the orchard since 2004

Other land-uses present are forest Other land-uses present are forest (windbreaker), bare soil, and grass(windbreaker), bare soil, and grass

Page 7: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Data & SoftwareData & Software

6” (15 cm)6” (15 cm) resolution Color Infrared (CIR) imagery ( resolution Color Infrared (CIR) imagery (RR, , GG, , and and NIRNIR bands) through bands) through Photoscience Geospatial Inc.Photoscience Geospatial Inc.

Field LAI data with LAI-2200 Canopy Analyzer (Field LAI data with LAI-2200 Canopy Analyzer (LI-COR LI-COR Biosciences, Lincoln, NEBiosciences, Lincoln, NE ) )

IDRISI Taiga (IDRISI Taiga (Clark Labs, Clark University, Worcester, MAClark Labs, Clark University, Worcester, MA) ) softwaresoftware

eCognition (eCognition (Trimble Geospatial, Westminster, COTrimble Geospatial, Westminster, CO) ) Developer softwareDeveloper software

QT Modeler (QT Modeler (Applied Imagery, Silver Spring, MDApplied Imagery, Silver Spring, MD) software) software ArcGIS 10 (ArcGIS 10 (ESRI, Redlands, CAESRI, Redlands, CA) software) software

Page 8: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Orthoimagery (15 cm resolution)Orthoimagery (15 cm resolution)

Page 9: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

LAI-2200 Plant Canopy AnalyzerLAI-2200 Plant Canopy Analyzer

Page 10: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

LAI-2200 Calibration & Data LAI-2200 Calibration & Data Collection Collection

Page 11: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

LAI Raster Development LAI Raster Development

LAI raster was developed for the Z-Blu LAI raster was developed for the Z-Blu orchard using the relationship developed orchard using the relationship developed by Schultz and Engman (2000) by Schultz and Engman (2000)

LAI   =  - ln (SAVI + .371)/.48 LAI   =  - ln (SAVI + .371)/.48 Which calculates LAI from the soil Which calculates LAI from the soil

adjusted vegetation index (adjusted vegetation index (SAVISAVI) from ) from NAIP orthoimageryNAIP orthoimagery

Page 12: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

SAVI Raster Development SAVI Raster Development

SAVISAVI raster is created by the algorithm developed raster is created by the algorithm developed Huete (1988)Huete (1988)

where r and ir are spectral reflectance from the R-and NIR-band images, respectively, and the L is a constant that represents the vegetation density.

Huete (1988) defined the optimal adjustment factor of L = 0.25 to be considered for higher vegetation density in the field, L = 0.5 for intermediate vegetation density, and L = 1 for the low vegetation density.

Page 13: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

OBIA Based Image SegmentationOBIA Based Image Segmentation

Individual bands (R-, G-, and NIR-band) were Individual bands (R-, G-, and NIR-band) were extracted from the CIR imageextracted from the CIR image

LiDAR data was used to produce ground elevation LiDAR data was used to produce ground elevation raster (DEM) and the plant height raster (nDSM) raster (DEM) and the plant height raster (nDSM) using the using the QT Modeler QT Modeler softwaresoftware

NDVI raster was developed using the algorithmNDVI raster was developed using the algorithm

Another raster was developed following the WATER algorithm developed by Nolan (2011)

where r , g and ir are spectral reflectance from the R-, G- and NIR-band images, respectively

NDVI = (ir - r) / (ir + r)

WATER = (g - ir) /(g + ir)

Page 14: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Rule Sets Developed for Image Rule Sets Developed for Image SegmentationSegmentation

All these rasters (All these rasters (previous slideprevious slide) were ) were imported to eCognition softwareimported to eCognition software

Rule sets were developed for distinguishing Rule sets were developed for distinguishing individual classes including blueberry individual classes including blueberry species from the study area imagespecies from the study area image

Multiresolution segmentation (Multiresolution segmentation (MRSMRS) or ) or other segmentation techniques were used other segmentation techniques were used in many steps to have classes distinguishedin many steps to have classes distinguished

Page 15: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Rule Sets (Rule Sets (Step by StepStep by Step))

Page 16: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Rule Sets (Rule Sets (Step by StepStep by Step))

Page 17: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Rule Sets (Rule Sets (Step by StepStep by Step))

Page 18: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Rule Sets (Rule Sets (Step by StepStep by Step))

Page 19: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Rule Set Example from eCognitionRule Set Example from eCognition

Page 20: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Rule Set Example from eCognition Rule Set Example from eCognition ((Contd…Contd…))

Page 21: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Blueberry LAI CorrelationBlueberry LAI Correlation

Segmentation of the orchard image Segmentation of the orchard image provided different blueberry species along provided different blueberry species along with bare soil and grass.with bare soil and grass.

Visual correlation was conducted on Visual correlation was conducted on classified image versus LAI rasterclassified image versus LAI raster

The field collected LAI values were The field collected LAI values were correlated with the LAI values obtained correlated with the LAI values obtained from the image analysis to observe from the image analysis to observe relationship.relationship.

Page 22: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology
Page 23: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

SAVI RasterSAVI Raster

Page 24: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

NDVI RasterNDVI Raster

Page 25: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

LAI RasterLAI Raster

Page 26: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

OBIA Segmentation Raster SeriesOBIA Segmentation Raster Series((Tall & Short Features separationTall & Short Features separation))

Page 27: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Multiresolution Segmentation for Differentiating Multiresolution Segmentation for Differentiating Various Classes of Tall ObjectsVarious Classes of Tall Objects

Page 28: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Multiresolution Segmentation for Differentiating Multiresolution Segmentation for Differentiating Various Classes of Short ObjectsVarious Classes of Short Objects

Page 29: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Trees & Buildings SeparationTrees & Buildings Separation

Page 30: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Evergreen & Deciduous Trees Evergreen & Deciduous Trees SeparationSeparation

Page 31: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Water & Shadow SeparationWater & Shadow Separation

Page 32: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Bare Soil ClassificationBare Soil Classification

Page 33: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

High Bush Blueberry Variety Distinction High Bush Blueberry Variety Distinction (NDVI between 0.2 and 0.4)

Page 34: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Rabbit Eye Blueberry Variety Distinction Rabbit Eye Blueberry Variety Distinction (compactness greater than 3 )

Page 35: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Chessboard Segmentation to Chessboard Segmentation to Obtain Smaller ObjectObtain Smaller Object

Page 36: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

MRS Region Grow Segmentation to MRS Region Grow Segmentation to Obtain Larger Objects in Rabbit EyeObtain Larger Objects in Rabbit Eye

Page 37: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Grass ClassificationGrass Classification

Page 38: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Final Segmented ImageFinal Segmented Image

Page 39: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

LAI Raster & Field LAI LAI Raster & Field LAI Collection LocationsCollection Locations

Page 40: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

Field LAI Collected Values Field LAI Collected Values

Page 41: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

LAI Raster & Field LAI LAI Raster & Field LAI CorrelationCorrelation

Page 42: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

ConclusionsConclusions

LAI raster was developed using the SAVI raster of the study area 0.15 cm resolution CIR.

The LAI values very well visually correlated with two blueberry varieties (High Bush and Rabbit Eye) in the field.

The LAI values from the LAI raster and the values collected from field did not correlate well. It was attributed to the faulty LICOR instrumentWe will collect LAI data in the field later with a better

instrument

Page 43: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

ConclusionsConclusions

The OBIA based image segmentation process was very exhaustive but worth every extra rule sets we develop

SAVI raster very clearly distinguished different vegetation features in the image including both blueberry varieties (High Bush and Rabbit Eye) We would use that in the OBIA segmentation process and it

would save a great deal of effort Ground truth will be conducted later for accuracy assessment

but initial assessment shows the result more than 80% accurate Ground truth points were collected in 2009 for a previous study

but 2010 orthoimagery is used in this study Different stages (year of plantation) of blueberry plants in the

orchard will be separated with the nDSM raster in a later stage

Page 44: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

AcknowledgementAcknowledgement

This work was supported in part by the Georgia Space Grant Program, managed by the Georgia Institute of Technology on behalf of the National Aeronautics and Space Administration, and by State and Federal Funds allocated to Georgia Agricultural Experiment Stations

Hatch project GEO1654..

Page 45: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology
Page 46: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

SAVI Raster

Page 47: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

LAI raster for entire land-use (Ranges from -7.59 to -11.377)

Page 48: Orchard LAI Estimation and Land-use Correlation using Geospatial Technology

LAI raster showing the LAI values (very homogenous) of only Kudzu (Ranges from -11 to -11.315)