DEDAN KIMATHI UNIVERSITY OF TECHNOLOGY

PROJECT PROPOSAL FOR UNDERGRADUATE DEGREE CERTIFICATE.

TITLE: GIS AND REMOTE SENSING APPLICATION IN TEA HEALTH

DETERMINATION

CASE STUDY: KERICHO-TIMBILIL TEA ESTATE

BY: TERER KIPKIRUI WESLEY

EO32-1396/2011

INSTITUTE OF GEOMATICS, GIS AND REMOTE SENSING

(IGGReS)

BSC. GEOPATIAL INFORMATION SCIENCE

1

Table of Contents

Chapter One...................................................................................................................................................4

1.0. INTRODUCTION.............................................................................................................................4

1.1 Problem statement.............................................................................................................................6

1.2 Objective.............................................................................................................................................6

1.2.0 Main objectives................................................................................................................................6

1.2.1 Specific objective.............................................................................................................................6

1.3 Research Question.............................................................................................................................6

1.4 Justification........................................................................................................................................6

1.5 Study area...........................................................................................................................................7

Chapter Two...................................................................................................................................................9

2.0 LITERATURE REVIEW.................................................................................................................9

Chapter Three...............................................................................................................................................11

3.0 Methodology.....................................................................................................................................11

3.1 Data acquisition................................................................................................................................11

3.1.0 Remote Sensing Data....................................................................................................................11

3.1.1Topographic Map...........................................................................................................................12

3.1.2 Field Survey and Statistical Data................................................................................................12

3.1.3 Photographs of Tea Bush.............................................................................................................12

3.1.4 Statistical Data..............................................................................................................................12

3.2 Data Preparation..............................................................................................................................12

3.4 Georeferencing the Images................................................................................................................13

3.5 Classification....................................................................................................................................14

3.6 Accuracy Assessments.....................................................................................................................14

3.7 Spectral Signature Analysis............................................................................................................15

3.7.0 NDVI (Normalized Difference Vegetation Index) Analysis......................................................15

3.7.1 Tassled Cap Transformation.......................................................................................................16

3.7.2 Normalized Difference Moisture or Water Index (MDWI)......................................................17

Composite Bands and Visual Interpretations.....................................................................................17

3.8 Regression Analysis.........................................................................................................................17

4.0 QUALITATIVE IMAGE ANALYSIS AND RESULTS......................................................................18

4.1. Spectral Signature Analysis of Landsat-7 ETM+........................................................................19

4.2 Image Processing Analysis..............................................................................................................19

2

4.3 Unsupervised Classification unsupervised classification.............................................................19

4.4 Normalized Difference Vegetation Index (NDVI).........................................................................20

4.5 Normalized Difference Moisture or Water Index (NDMI or NDWI).........................................20

4.6 GIS Analysis.....................................................................................................................................20

5.0 Requirement...........................................................................................................................................22

5.1 Data sources......................................................................................................................................22

5.2 Expected result.................................................................................................................................22

5.3 Reference..........................................................................................................................................23

3

Chapter One

1.0. INTRODUCTIONTea (Camellia sinensis) is an important commercial crop in many subtropical and tropical areas

of the world. Tea, owing to its favorable effects on human health, currently enjoys a great

popularity among other beverages worldwide (Ruan and Härdter, 2001). Tea is a perennial plant,

which repeatedly pruned at different intervals (3-6 years). The shoots are plucked at regular

intervals (6-25 days) and removed a certain amount of various elements from the plant-soil

system (Verma, 1997). Certain major nutrients have to be supplemented through fertilizer

application. Tea yield increases sharply with increased levels of N and K to a certain point

(Barbora, 1996). Tea being a leaf crop, in the flush shoot the nitrogen content is the highest

followed by potassium (K), calcium (Ca), phosphorus (P), sulfur (S), magnesium (Mg) and zinc

(Zn). Nitrogen (N) is an important constituent of plants parts and plays a vital role in the

physiology of the tea plant. It is estimated that harvestable crop contains 3.5-5% N on dry matter

basis (Verma, 1997). Although applications of N can increase tea yields, it is recognized that the

quality of the manufactured product is suppressed by large N rates (Cloughly et al., 1983).The

study is conducted for identifying tea bush health and yield estimation.

Use of Remote Sensing (RS) and Geographical Information System (GIS) are discussed and

analyzed to come up with solutions for effective decision making.

Tea is a hardy, multi-stemmed, slow-growing evergreen shrub which, if allowed, can grow up to

twelve feet in height. It takes one to five years to mature. During cultivation it is pruned for easy

picking (two leaves and the bud are removed by hand) as a low spreading bush to a maximum

crop of young shoots (Hadfield, 1974). Tea cultivation in Kericho District. Currently the estate

have started reducing the old species which are 60 to 100-year-old tea bush. Also the small scale

farmers in Kericho have improved in species.

The most currently species is the purple teaand they are grown in many gardens in central part of

Kericho county, Kenya. Plucking season begins with the first flush of new 1 growth in March

and April after the dry seasons of December to February. Following a short period of dormancy,

4

the plants put forth a second flushthat is picked from May into June. The summer months bring

heavy rains from July until September, yielding a monsoon flush. The autumnal flushis picked in

October and November. The cold winter months of December to February are a period of

dormancy. Top grade first and secondflush tea leaves bring some of the highest prices found at

the tea auctions. The tea estate and the small scale farms have one major problem that is facing

tea, the low quality of tea which then bring impact on economic and marketing strategies.

The gardens are all located at elevations up to over 2000 meters above mean sea level. Due to the

unique and complex combination of agro-climatic conditions prevailing in the region tea grown

here has a distinctive and naturally-occurring quality and flavor which has won recognition all

over the world for well over a century. The effort of the tea manufacturer is generally aimed at

maintaining the natural aroma as much as possible. The final quality of tea depends primarily on

the nature and chemical composition of the plucked leaf which is again dependent on the type of

bush, the growing conditions and the characteristics of plucked leaf, like coarseness and fineness.

Only careful and proper processing will bring out the full potential of the green leaf.

The factors affecting tea quality apart from those involved in processing can be distinguished in

three groups: genetic, environmental, and cultural.

(i) Tea quality is primarily determined by the genetic properties of the tea plant in

particular area.

(ii) Both soil and climate influence the quality of tea. Climatic conditions including

temperature, humidity, exposure to sunlight, and rainfall are also important in

determining quality.

(iii) Field operations like pruning, fertilizing, shading, plucking round, and plucking

standards also play an important role in determining the quality of tea.

5

1.1Problem statementTea Management Board currently use field statistical result and laboratory research in

determining the progress and states of tea health. This method take a lot of time because it

requires field visit study and laboratory result has to be analyzed for decision making to be made.

This thesis is an attempt to explore tea bush health by studying and analyzing spectral responses

in remote sensing imagery to differences in tea plant quality. This technology will not only serve

the area of study but it will be anywhere.

1.2 Objective

1.2.0 Main objectivesTo apply GIS and Remote Sensing in monitoring and managing tea health.

1.2.1 Specific objective To monitor tea plant health and production and the quality of tea by use of remote

sensing imagery.

To analyze conditions and come up with solutions that could be shared with the tea

garden management for effective decision making in the near future and help maximum

profit for the tea industry.

1.3 Research Question

(i) How helpful are the spectral signatures in the assessment of the features on the

ground in the context of tea canopy characterization?

(ii) Is there any relationship between vegetation index NDVI and tea leaf yield and their

quality?

(iii) How can the results obtained be helpful in overcoming the problems of conventional

tea producing methods?

1.4 JustificationThe tea yield is determined mainly by the number of shoots plucked at each harvest and the

quality of tea depends on health of the bush and limited use of pesticides and fertilizers. Remote

sensing could help to identify areas within a field which are healthy or experiencing difficulties,

6

so that preventive measures can be taken. For instance, the correct type and amount of fertilizer,

pesticide and need for irrigation if necessary can be decided. Using this approach, the

management committees could improve not only the quality and productivity but also help

reduce farming costs and minimize environmental impacts. If the data are georeferenced, and if

the planter has a GPS (Global Positioning System) unit, precise location of the healthy and

stressed plants can be found by matching the coordinates of a given location to that on the image.

Detecting tea bush locations and monitoring tea health requires moderate to high-resolution

imagery and multispectral imaging capabilities.

1.5 Study area

The study area is Kericho County, the highlands of Kenya, an estate called Kapkorech. In this

highland area tea is grown between 4500 feet (1500m) and 6750 feet (2250m) above the sea

level. On tropical, red loam soil and decomposed volcanic deposits. The soils are well drained

and have a ph. on the range of (4.5-6.5). Straddling the equator,Kericho County as one of the tea

growing zone, have an ample supply of sunlight and even distribution of rainfall throughout the

year. Providing the optimal conditions for growing tea. The rainfall in this area ranges between

1200mm and 2500mm annually (the rainfall pattern is unimodal) while the temperature ranges

between 12°c and 28°c.KerichoCounty is characterized by small holder farmers and tea is the

dominant cash crop.

However Kericho county is also known to host the multinationalcompanies that are involved

directly in tea production e.g.Uniliver,James Finlay. Williamson tea, Kapchebet and the new

risingcompanies,Kabianga tea factory which are owned privately and other new ones are coming

up, for example purple tea factory in which the construction is going on currently around the

Brooke Centre on your way to Kericho town. There is also the government owned factories;

KTDA and they all compete for the tea leaves from the small holder farmers. This makes it

suitable to undertake the study.

Map of Kerichocounty in Kenya map context

7

8

Chapter Two

2.0 LITERATURE REVIEWThe study was toassess the impacts of tea leaf quality and challenges faced by management of

tea farmers in determining the teabush and their produce. The need for the research arose from

the realization that tea quality for both small and large scale tea leaf produce is currently faced

with various challenges.

There is need for agricultural policy makers to get empirical insight in the levels of tea farmers’

efficiency as well as different determinants factors in order to be able to more effectively address

the performance of tea leaves by the farmers. This study attempts to investigate the challenges

faced by tea farmers in their plots even after applying fertilizers.

Omosa (2003) argued that KTDA had failed in executing its functions and farmers were losing

through low prices. The returns to the tea delivered by farmers were low and made the farmers to

be trapped in a vicious cycle of poverty. This made the farmers to look for alternative market

outlets. She found out that the movement of farmers to other alternative market had a direct

impact on the quality of tea and the Kenyan tea prices fetched on the world market. This was

because the alternative markets did not follow the two leaves and a bud standard of plucking that

KTDA follows strictly as required by the Kenya tea board.

Hazarika (2011) conducted a study in India on the changing scenario of market for tea in India.

She realized that in the tea industry producers are not the actual markets and not want to be.

Most of the owners of tea farms were satisfied with the auction system. This was because tea

growers did not give much attention on marketing aspect as they always have a ready market for

the tea leaves. The study also showed that the quality of tea had a different effect on prices in the

auction. She recommended that the tea producers take a proper initiative in the marketing field

they can achieve a higher profit margin which ensure a higher revenue.

Tea farming covers a large proportion on the tea production on the total tea production in the tea

producing countries. A study carried out by CPDA, 2008 showed that by2006 the small holder

farmer was contributing 60% of the Kenya’s total tea production. However the small holder

farmers are faced by several challenges, and end up making the small holderfarmers losers in the

industry. In a case study conducted in Malawi, relates the challenges faced by small holder

9

farmers in supplying their produce to poor governance and management of the lucrative sector

and also poor resource allocation. The authors argued that the government allocated much

resources to subsidize production of other crops like cotton, making a tea farmer to incur a lot of

cost during production. The farmers had to pay taxes for their produce and transportation cost to

deliver their produce. Small holder farmers in the tea growing zone areas have devoted all their

land to mainly tea production with an expectation of making an income from it.

A study by Mwaura and Muku(2012) further showed that a large proportion of farmers are living

below the poverty line. They argued that the small holder farmers faced a lot of challenges and

there were no returns to their production. Kalunda (2013) conducted a study in Nyeri County on

financial inclusion of small scale tea farmers. She realized that the level of poverty among the tea

small holder farmers denied them access to credit. Mainly because of the ever fluctuating tea

prices and furthermore the low prices, making the financial institution indifferent to lending

them. The study also showed that a largepercentage of farmers had difficulty in repaying the

loan. She recommended that a financial literacy be made available to the farmers.

The reviewed literature shows that a proper marketing and farm management in the subsector by

the small holder farmers presents a huge opportunity in increasing the country’s GDP. With

better market prices, and fair channels of tea leaves delivery system by small holder farmers

offers an opportunity in increasing farm incomes to small holder farmers and facilitate the

millennium goal of poverty reduction. The current study considers tea hawking as a major

challenge facing small holder farmers. The previous studies have not shown in detail the

economic effects of tea hawking. Tea hawking affects prices, revenue and quality of tea.

Thisstudy will try to enumerate the reasons for tea hawking, awareness of farmers of the law

implications of tea hawking and also suggest the possible ways of resolving the problem.

10

Chapter Three

3.0 MethodologyTo initially detect and examine the healthy and stressed tea patches, different bands as well as

true color and false color compositions are compared and discussed here.

3.1 Data acquisitionRemote sensing processes in that it takes into account, source of data, type of data, temporal resolution,

spectral resolution and even type of sensor for imagery. In fact it is the step that ensures that all

requirements and indeed the character of data is desired.

Landsat imagery will be downloaded from USGS’s Earth Explorer. The data used for this study

are: Landsat-7 ETM+ remote sensing data, topographical map data, field survey data, and

recorded data of the Tea Board of Kenya from TRFK Kericho.

3.1.0 Remote Sensing Data

Landsat-7 TM+ data will be obtained from USGS archives for the study area. Images for the year

2013 and 2014 will be used for this study. Both images will be downloaded during April-

November, which is the time when the tea shoots are matured and ready for plucking which is

done manually by hand, but currently they are using mechanical to pick tea.

These satellites have a major component of NASA’s Earth observation program with primary sensors

evolving over thirty years: MSS (Multi-Spectral Scanner), TM (Thematic Mapper), and ETM+

(Enhanced Thematic Mapper plus). Landsat supplies high resolution visible and infrared imagery, with

thermal imagery and a panchromatic image also available from the ETM+ sensor. The collection of

Landsat available through GLCF is designed to complement overall project goals of distributing a

global, multi-temporal, multi-spectral and multi-resolution range of imagery appropriate for land cover

analysis. Analyzing vegetation cover change using Landsat Satellite data is considered very much

suitable since it has NIR (Near Infrared Band) and the Red Band.

11

3.1.1Topographic Map

A topographical map of Kericho for the area will be used to georeferenced to delineate

approximate garden boundaries. The map will be obtained from Survey of Kenya. It will be

scanned, and geo-referenced by use of ArcGIS software. This map will aid in identification of a

true location of infrastructures and identification of potential land use.

3.1.2 Field Survey and Statistical DataA field survey was conducted during the month of November in 2007, coinciding with the season

when the images were actually captured. Tea bush are seasonal hence the timing is important for

analysis and conclusions. Sample locations within the Timbilil tea garden in Kericho will be

characterized and photographed in terms of tea health.

Ground truth data will be collected using a handheld GPS for establishing the geographic

coordinates of the gardens. The readings were then checked and converted as necessary to use

for image analysis. While these 2007 observations do not coincide with the 2013 and 2014

Landsat data, they are nevertheless useful for establishing garden locations and characterizing

the nature of varying tea health.

3.1.3 Photographs of Tea BushGround-level photographs is taken using a 8 mega pixel camera, the timing being April to late

November when actually the bush are matured and ready for plucking shoots. Representative

photographs of healthy, moderately healthy, and unhealthy tea.

3.1.4 Statistical DataData from the Timbilil Tea garden manager's office as well as the Tea Board of Kenya and

KRFK for validation and comparison of the image analysis results.

3.2 Data PreparationLandsat-7 ETM+ data for the years 2013 and 2014 will be studied. The TIFF and FAST format

scenes will be imported using IMAGINE for further analysis and the study area was clipped out

in IDRISI environment. The latitude and longitude coordinates of the study area were checked

12

with topomap and as image restoration and rectification, are intended to correct for sensor- and

platform-specific radiometric and geometric distortions of data.

3.4 Georeferencing the ImagesThe images of each band are projected from reference system UTM-EGS 1984 and units in

"meter" reference systems with units "degree" using IDRISI image processing software. The

resulting data files for 2013 and 2014 hold the following characteristics

Radiometric corrections may be necessary due to variations in scene illumination and viewing

geometry, atmospheric conditions, and sensor noise and response. Each of these will vary

depending on the specific sensor and platform used to acquire the data and the conditions during

data acquisition. Also, it may be desirable to convert and/or calibrate the data to known

(absolute) radiation or reflectance units to facilitate comparison between data.

Variations in illumination and viewing geometry between images (for optical sensors) can be corrected

by modeling the geometric relationship and distance between the areas of the Earth's surface imaged the

sun, and the sensor. This is often required so as to be able to more readily compare images collected by

different sensors at different dates or times, or to mosaic multiple images from a single sensor while

maintaining uniform illumination conditions from scene to scene.

Geometric registration process involves identifying the image coordinates (i.e. row, column) of several

clearly discernible points, called ground control points(or GCPs), in the distorted image and matching

them to their true positions in ground coordinates (latitude, longitude). Once several well-distributed

GCP pairs have been identified, the coordinate information is processed by the computer to determine

the proper transformation equations to apply to the original (row and column) image coordinates to map

them into their new ground coordinates. In order to actually geometrically correct the original distorted

image, a procedure called resampling is used to determine the digital values to place in the new pixel

locations of the corrected output image. The resampling process calculates the new pixel values from the

original digital pixel values in the uncorrected image. There are three common methods for resampling:

nearest neighbour, bilinear interpolation, and cubic convolution. Nearest neighbour resampling

uses the digital value from the pixel in the original image which is nearest to the new pixel location in

the corrected image. This is the simplest method and does not alter the original values, but may result in

some pixel values being duplicated while others are lost. This method also tends to result in a disjointed

or blocky image appearance. Bilinear interpolation resampling takes a weighted average of four pixels

13

in the original image nearest to the new pixel location. The averaging process alters the original pixel

values and creates entirely new digital values in the output image. Cubic convolution resampling goes

even further to calculate a distance weighted average of a block of sixteen pixels from the original image

which surround the new output pixel location. As with bilinear interpolation, this method results in

completely new pixel values. However, these two methods both produce images which have a much

sharper appearance and avoid the blocky appearance of the nearest neighbor method.

Mosaicking will be undertaken blend together several arbitrarily shaped images to form one large

radiometrically balanced image so that the boundaries between the original images are not seen

Pre-processing also aims at removing the noise from imagery, where noise is any unwanted disturbance

in image data that is due to limitations in sensing, signal digitization or the data recording process. This

noise has to be removed so as to restore image to as close to the original scene appearance as possible.

Noise could termed as dropping line or stripping.

Radiometric correction will be undertaken to improve the spectral resolution of the by reducing haze

effects. Visual interpretation will also aid in image pre-processing especially by identifying areas

obscured by clouds.

3.5 ClassificationThe intent of the classification process is to categorize all pixels in a digital image into one of several

land cover classes. Its objective of image classification is to identify and portray, as a unique gray level

(or color), the features occurring in an image in terms of the object or type of land cover these features

actually represent on the ground. Both supervised and unsupervised classification will be undertaken.

Unsupervised classification is normally undertaken before field work while for supervised classification

training areas are established based on ground truth information taken during field work or using high

resolution imagery from Google Earth software. Image classification helps in identifying land use/ land

cover types are identified in order to detect forest cover change.

3.6 Accuracy AssessmentsIn thematic mapping from remotely sensed data, the term accuracy is used typically to express the

degree of ‘correctness’ of a map or classification. Accuracy assessment evaluation will include an error

matrix which is a report of the overall proportion of correctly classified pixels. Finally, Kappa Statistics

will be calculated for the different areas that were classified. Kappa indices will also be employed

where the kappa coefficient result values are between 0 and 1. The latter shows complete agreement,

14

and is often multiplied by 100 to give a percentage measure of classification accuracy. Kappa

values are grouped into three groups: a value of kappa coefficient greater than 0.80(80%) will

represent strong agreement, a value of kappa coefficient between 0.40 and 0.80 (40 to 80%)

represents moderate agreement, and a value of kappa coefficient below 0.40(40%) represents poor

agreement.

3.7 Spectral Signature AnalysisThe spectral reflectance of tea plants can indicate variations of signature with respect to the:

I. phenology (each shoot is divided into three phases corresponding to the resting,

quiescent and bud-burst phases),

II. plant type, and

III. tea health

Healthy vegetation contains large quantities of chlorophyll, which gives most vegetation its

distinctive green color. In referring to healthy crops, reflectance in the blue and red parts of the

spectrum is low since chlorophyll absorbs this energy; in contrast, reflectance in the near-

infrared (NIR) spectral regions is high. Stressed or damaged crops experience a decrease in

chlorophyll content and changes to the internal leaf structure. The reduction in chlorophyll

content results in a decrease in reflectance in the green region and internal leaf damage results in

a decrease in near-infrared reflectance. These reductions in green and infrared reflectance

provide early detection of plant stress. The detection of chlorophyll content and greenness in

plant is the basis behind some vegetation indices, including the NDVI (Rajapakse et al. 2001),

the Wetness component of the Tasseled Cap Transformation, and the Normalized Difference

Moisture or Water Index (NDMI or MDWI). The use of each of these indices is examined and

shown here.

3.7.0 NDVI (Normalized Difference Vegetation Index) AnalysisVegetation indices are arithmetic transformations aimed at simplifying data from multiple

reflectance bands to a single value correlation to physical vegetation parameters. The most

commonly used of theses indices is the

15

Normalized Difference Vegetation Index (NDVI) used by researchers for extracting vegetation

abundance from remotely sensed data. It divides the difference between reflectance values in the

visible red and near-infrared wavelength to give an estimate of green vegetation abundance.

NDVI is (Jensen. 2007):

NIR-R

NDVI =

NIR + R

In this study, the NDVI transformation is done for the Landsat-7 ETM+ images for different

dates. The NDVI image is analyzed along with RGB and false color composite image.

3.7.1 Tassled Cap TransformationThe TASSCAP in IDRISI/ArcGIS undertakes a "Kauth-Thomas" tasseled cap 4- dimensional

transformation on 6 bands of TM data (excluding the seventh thermal band) to extract three new

index bands, which are designated

Brightness, Greenness and Wetness. The Brightness band commonly characterizes soil

brightness. The Greenness band highlights green vegetation cover or biomass above ground. The

Wetness band provides subtle information concerning the moisture status of the wetlands

environment. For a Landsat image, these three bands are calculated from the original six bands

for each pixel by means of the following formulas (Jensen. 2007).

Brightness = (TM1*0.3037) + (TM2*0.2793) + (TM3*0.4343) + (TM4*0.5585) +

(TM5*0.5082) + (TM7*0.1863)

Greenness = (TMl*-0.2848) + (TM2*-0.2435) + (TM3*-0.5436) + (TM4*0.7243) +

(TM5*0.0840) + (TM7*-0.1800)

Wetness = (TM1*0.1509) + (TM2*0.1793) + (TM3*0.3299) + (TM4*0.3406) +

(TM5*-0.7112) + (TM7*-0.4572)

These three bands are also considered in this study for the analysis of tea bush. During the bud-

burst phase, the garden is greener compared to the surrounding areas; unique values can be found

in the derived indices using the Tassled Cap method.

16

3.7.2 Normalized Difference Moisture or Water Index (MDWI)The NDWI, based on the near and middle-infrared bands, highly correlates with vegetation water

content. Water stress of tea plant can be tracked from NDWI transformed image results and can

be validated with Tassled Cap resulted moisture index for studying tea bush health (Jensen.

2007):

NIRTM4 - MidIRTM5

NDWI =

NIRTM4 + MidlRTMs

Composite Bands and Visual Interpretations

True color and false color compositions were prepared for an initial visual interpretation of

images. True color imagery portrays a region as the human eye would see it. False color

(specifically, color infrared) imagery assigns the color red to the near infrared band, rather than

the red band. This results in the portrayal of vegetation in shades of red, rather than green. In this

study the Landsat-7 ETM+ remote Image Processing and GIS Analysis.

Unsupervised image classification was conducted for extracting natural clusters (in IDRISI

environment) of similar pixel values in the image. A user defined post-classification assignment

of particular clusters to a tea class was used to select regions in the NDVI images for further

analysis. The final interpretation is the result of identification of features, field verification, map

preparation and data validation.

3.8 Regression AnalysisLinear regression is used to explain tea health analysis. The result is represented by scatter plots

of NDVI vs. NDML and various combinations of Tasseled Cap bands vs. NDVI and NDMI vs.

Tasseled Cap bands attribute values follow a linear pattern, then there is a high linearcorrelation,

17

while the data do not follow a linear pattern, there is no linearcorrelation. If the data somewhat

follows a linear path, then we say that there is a moderate linear correlation. Regression is used

here to help establish which indices may provide similar or redundant information, and which

provide different, uncorrelated information.

4.0 EXPECTED QUALITATIVE IMAGE ANALYSIS

The methodology used in this study are divided into a qualitative phase, which seeks to establish

regions of tea cultivation in terms of tea health categories, and a quantitative phase, which seeks

quantitative estimates of areas for each category, estimates tea yield, and compares the results

with the statistical data. The qualitative phase is described in this section and results discussed in

section five of Tea bush health mapping. The quantitative phase is described in section six and

section seven.

To initially detect and examine the healthy and stressed tea patches.

The healthy, moderately healthy and unhealthy tea patches are visually interpreted. False color

compositions (FCC) were generated using Landsat-7 TM+ bands 4, 3, 2, and true color

compositions were generated using the visible bands 3, 2, 1, respectively. The healthy and

stressed tea patches can be distinctly separated from the other land use classes in FCC

composition image. Following is an interpretation of the FCC image:

Tea Bush: In the FCC image the expected results is that tea patches will have their leave appear

in reddish color. The healthy tea bush appears in bright red color due to the higher reflectance

while the moderately healthy tea bush appears in reddish brown. A reference map will be

prepared using the three interpreted tea health categories. Linear features and the regions of tea

indicated by the topomap are included to provide context. Note that these regions do not entirely

coincide; the interpreted image, as well as field examination, indicates the presence of tea

cultivation that extend beyond the boundaries indicated in the topomap.

18

4.1.Spectral Signature Analysis of Landsat-7 TM+Spectral signature analysis was performed for representing the reflectance variability between

land cover classes and for evaluating spectral responses within land cover classes. The

reflectance curves for all bands were generated for studying the pattern of spectral responses of

each landuse/landcover classes.

In Landsat imagery vegetation can be easily distinguished by its unique spectral signature.

The reflectance value of Landsat imagery ranging from 0-255 are represented by Digital Number

DN). The reflectance DN value of healthy, moderately healthy, and unhealthy or stressed, as well

as severability of tea's from other vegetation peaks in the near infrared (NIR) band are discussed

here.

The study of reflectance value for the different bands suggests that the NIR band could be the

most useful for analyzing vegetation health. Any useful index should rely on this band. The low

peak appearing in the red bad (band 3) and middle infrared (band 7) is mainly due to the

absorption of chlorophyll. The river bed shows its peak in the red, TIR band, blue, green, MIR

and NIR respectively. This portrays the representative spectral characteristics for selected

classes, derived through careful interpretation. Figure 18 portrays this information in graphical

form.

4.2Image Processing AnalysisIn the process of image analysis, unsupervised classification using the

ISODATA algorithm will be performed to determine data clustering, which could be used to

derive spectral classes that exist in the image. The NDVI transformation is performed to derive a

tea bush health index that could be used to classify healthy, moderately healthy and stressed tea

bush areas. NDMI and Tasseled Cap image transformations were also attempted to determine the

health of the bush along with NDVI index.

4.3Unsupervised Classification unsupervised classification Unsupervised Classification unsupervised classification could also be used as an alternative

method of derivingthree tea health classes. The ISODATA algorithm will performed for both

datesrepeatedly with different numbers of clusters. The classifiedimages that will be generated

will be compared with the NDVI and field studyresults.

19

4.4Normalized Difference Vegetation Index (NDVI)The NDVI vegetation index characterizes vegetation through a combination of the visible red

and near infrared bands of any multispectral data. Tea bush with healthy, moderately healthy and

unproductive or stressed tea patches are recognized as higher to lower NDVI values respectively.

Typical NDVI values for different categories of tea bush health

4.5Normalized Difference Moisture or Water Index (NDMI or NDWI)

NDWI image processing analysis will be done to find the moisture index in the healthy,

moderately healthy and stressed tea bush. The water index is (0.337) for the healthy tea canopy

and very low water index for stressed tea bush (0.079).

Negative NDWI index is found on concrete/manmade features. Typical NDWI index for

different categories of tea bush health.

4.6GIS Analysis NDVI-based Classification

The NDVI image will compared with landuse/landcover map as well as with RGB band

composite image and FCC image for extracting the feature.

NVDI results is expected to be the most reliable and practical basis for refining tea class into its

three health subclasses (Dutta et al, 2005). Toward this end, a user defined remapping of values

was performed for the transformed NDVI images.

The range of positive and negative NDVI values is studied thoroughly through comparison with

ISODATA results and field observations, and a standard set of upper and lower threshold values

is establish for classification.

20

21

Subset

Image

Statistical Data of the area

DATA ACQUISITION

Topographic Map

Landsat-7

TM+ Raster

5.0Requirement

1. ArcGIS software

2. ERDAS IMAGINE software

3. Microsoft 2013

4. Quantum GIS software

5. QGis software

5.1Data sourcesData type Format Source

Topographical map JPEG Survey of Kenya

Soil Data Raster TRFK (Kericho)

Landsat imagery Raster Earth explorer

Digital images JPEG TRFK (Kericho)

Yield statistical data raster TRFK (Kericho)

5.2Expected result1. After exploratory analysis and comparison of results, depending on the outcome of the

analysis decision may be made thatthe NDVI transformation of Landsat-7 TM+ image

data formed an adequatebasis of analysis the mapping of tea bush health. The user

defined classificationrepresents about the health of the tea bush canopy.

2. The resulting map represents tea health, and is meant to identify locations whichneed

immediate action.

22

5.3Reference

Wijeratne, M. A. (2004). Vulnerability of Sri Lanka tea production to global

climate change. Water, Air, & Soil Pollution.

Jensen, J. R. (1996). Introductory Digital Image Processing: A remote sensing

Perspective. 2nd edition.

Dutta, R. (2005). Assessment of Tea Bush Health and Yield Using Geospatial

Techniques.

Mwaura, F., &Muku, O. (2007). tea farming enterprise contribution to small holders' well-being

in kenya. (pp. 307-313). Nairobi: Tea Research Foundation of Kenya.

Omosa, m. (2003). the interplay between commodity markets and rural livelihoods.

Markham, B. L. and Barker, J. L., 1986, Landsat MSS and TM post-calibration dynamic ranges,

exoatmosphericreflectances and at-satellite temperatures, EOSAT Landsat Technical Notes.

Crist, E. P. and Cicone, R. C., 1984, A physically-based transformation of Thematic Mapper data --

the TM Tasseled Cap, IEEE Trans. on Geosciences and Remote Sensing, GE-22: 256-263.

Kauth, R. J., and G. S. Thomas, 1976. The Tasseled Capa graphic description of the spectral-

temporal development of agricultural crops as seen by Landsat. Proceedings of the Symposium

on Machine Processing ofRemotely Sensed Data, Purdue University, West Lafayette, Indiana,

pp. 4B41-4B51.

23