Published on

International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.

Published in: Technology
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide


  1. 1. Amol D. Vibhute, Dr. Bharti W. Gawali / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.081-09181 | P a g eAnalysis and Modeling of Agricultural Land use using RemoteSensing and Geographic Information System: a ReviewAmol D. Vibhute*, Dr. Bharti W. Gawali**1Research Scholar, 2Associate Professor*, ** (Department of Computer Science & Information Technology,Dr. Babasaheb Ambedkar Marathwada University, Aurangabad, (M.S), INDIA-431004)ABSTRACTGIS, remote sensing and Globalpositioning System are the most widely usefultools for land use planning and decision supportsystem. Remotely sensed imagery is beneficial foragricultural production. It gives the accurateinformation of agricultural activities such asdifferent crop identification and classification,crop condition monitoring, crop growth, croparea and yield estimation, mapping of soilcharacteristics and precision farming.Information from remotely sensed imagery,geographic information system and globalpositioning system allows farmers to carry onlyaffected areas of a field. Problems within the fieldmay be identified before they create a bigproblem in the agricultural production usingremotely sensed images. This paper attempts toreview different techniques for variousapplications of GIS and Remote sensing for landuse/land cover change detection, cropidentification and classification, crop conditionmonitoring, crop growth, crop area and yieldestimation, mapping of soil characteristics andprecision farming. Thus implementating GIS andRS for better production of the crops as well asland use/land cover change detection can beachieved.Keywords – Classification, Crop acreage and yieldestimation, Precision farming, RS and GIS, Soilmapping1. INTRODUCTIONDue to rapid growth of population,urbanization, industrization, the agricultural land isdecreasing day by day. In twenty first century, landuse and land cover plays an important role in globalenvironmental change. The use of informationtechnology to support decision making in detectingland use and cover is essential and recent. For local,regional and macro level planning and managementLand cover/land use play an important role [1] byusing RS & GIS it is possible to make planning anddecision process more realistic, effective [2]. Landcover is nothing but the physical material on theEarth surface which is covered by different parts ofthe nature or also the man made parts on the Earthsurface i.e. soil, rocks, water bodies, vegetation,built-up areas, trees, etc. Land use means those areaswhich are used by humans for their need. Peoplesagreed to produce, maintain the change on landcover through the adjustments, movements andinputs [3].Both developed and undeveloped countriesare dependent on agriculture for the economic aswell as food security. Agriculture is the primarysource of many countries to maintain the food foreveryone and it plays a dominant role in almostevery country to develop the economical condition.Food production in a cost-effective manner is thegoal of every farmer as well as large scale farmerand agricultural agencies, so there is need to beinformed a farmer in an efficient way to get theknowledge and inform them about food production.Remote sensing, GIS and GPS these technologieswill useful for farmer to monitor their crops beforethey got damage, yield estimation, soil condition aswell as precision farming. Commodity brokers arealso very interested in how well farms areproducing, as yield (both quantity and quality)estimates for all products control price andworldwide trading [4], [5]. So there is need to securethe food in an effective way. Because good crophealth is a part of every person’s diet. In everycountry food production through sufficient quantityand quality is essential for the well-being of thepeople. To growth of crop in the field of agricultural,as living organisms, it requires water and nutrientsand is sensitive to extreme weather phenomena,diseases and pests. So to overcome these problemsand identify as well as monitor crops, remotesensing data is very much useful. GeographicalInformation System along with Remote Sensing andother types of data will helpful in decision supportsystem about crops and agricultural strategies [6].Extracting land use/land cover (LU/LC) informationis crucial exercise for agricultural land which is mostuseful for decision support system, planning anddevelopment in agriculture [7]. In the study of Landuse/ Land cover pattern analysis different sourcesare important for effective planning andmanagement like soil survey manuals, topographicmaps, aerial photographs, vegetation surveys, floodmaps, hydrology maps, and property surveys etc.,[8]. Due to the population and urbanization land useand land cover rapidly change, to acquire the exact
  2. 2. Amol D. Vibhute, Dr. Bharti W. Gawali / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.081-09182 | P a g eland pattern remote sensing & GIS technique is mostessential [9], [10].Remote sensing is the technology to acquire theexact information from the earth surface patternswithout making any physical contact to it. In today’sdate, this technology uses the satellites to detect andclassify objects on the earth surface and inthe atmosphere and oceans by using thesignals of electromagnetic radiation emittedfrom aircraft or satellites [11].Geographic Information System (GIS) is acomputer system designed for capturing, storing,integrating, analyzing and displaying data from ageographic angle. The measurement of natural andhuman made phenomena and processes from aspatial angle. These measurements emphasize threetypes of properties commonly associated with thesetypes of systems: elements, attributes, andrelationships. The storage of measurements in digitalform in a computer database. These measurementsare often linked to features on a digital map. Thefeatures can be of three types: points, lines, or areas(polygons). The analysis of collected measurementsto produce more data and to discover newrelationships by numerically manipulating andmodeling different pieces of data. The depiction ofthe measured or analyzed data in some type ofdisplay - maps, graphs, lists, or summary statistics[12].2. WORK DONE SO FARThere is lot of techniques used to identifyland use/land cover change detection, and landuse/land cover monitoring from satellite imagery.Fuzzy C-mean supervised classification algorithmhas been used in land use/land cover changedetection through Landsat TM5 satellite images withoverall 90% and 89% accuracy [13]. To classify theland use/land cover from remotely sensedmultispectral imagery and to detect the change onLULC from multitemporal imagery researchers hasbeen used supervised maximum LikelihoodClassification [1], [7], [14], [15], [16], [17], [18],[19], [20], [21], minimum distance [16], [21],[22],[23], parallepiped [16], [21], fishersclassification [16], unsupervised classification forISODATA [14],[23],[24], Hybrid classification [14],Box classifier[25] mostly.Many researchers identified and classifiedcrops by using multispectral and multitemporal dataand supervised and unsupervised classificationtechniques.Fuzzy classifier, SR (Simple Ratio), NDVI(Normalized Difference Vegetation index), TNDVI(Transformed Normalized Difference VegetationIndex), SAVI (Soil-Adjusted Vegetation Index) andTVI (Triangular Vegetation Index) are used toidentify cotton crop using temporal multi-spectralremote sensing imagery. The overall fuzzy highestclassification accuracy 93.12% was achieved fromSAVI temporal data base of data set 2 [26].Maximum Likelihood and Artificial Neural Networkalgorithms are used with SPOT HRV and Land satTM data for classifying agricultural crops combiningmulti-temporal, multi-spectral and multi-sourceremotely-sensed data to identifying generalagricultural crop classes over an area in East Anglia(UK) [27]. Sugarcane, cassava and maize crops areidentified through Radarsat-2 multipolarization datain Thailand, because high cloud cover availability innature and these sensors cannot give the wholeinformation about the objects. Therefore, the radarsensor is the alternative option to obtain satellite dataespecially RADARSAT-2 which provides fullpolarizations (HH, VV, HV, VH) [28]. UnsupervisedISODATA and NDVI (normalized differencevegetation index), these two methods was used toclassify crop. Method first based on unsupervisedclassification of Landsat visible and Near-infraredsatellite images obtained at multiple dates in thegrowing season, followed by traditional, manualclass identification. And second method based onunsupervised classification as well as NDVI mapsperformed MODIS images. Tree analysis wasapplied for grouping similar classes into clustersusing NDVI mask data and for crop type of eachcluster was identified from ground truth. They got3% more overall classification accuracy usingunsupervised classification and NDVI masks whichis the second method than first method [29].Multispectral images obtained from the SPOT 5satellite acquired two different date images has beenused to identify crop area using pixel based i.e.unsupervised and maximum supervisedclassification and object based classificationtechniques [30]. Hyperspectral remote sensing hasalso helped enhance more detailed analysis of cropclassification [31]. Maximum Likelihood classifierwas performed on hyperspectral data of the Hymapsensor for different crop classification. This methodwas used to extract a subset of individual bands fromhyperspectral data and the classification results arecompared to those achieved by using principalcomponents transformation as a well establishedfeature extraction technique to optimize theperformance of the maximum likelihood classifier[32]. Rice Crop Identification and Classificationusing Hyper-Spectral Image Processing System, wasreviewed through the comparison of multispectraland hyper spectral image processing techniques [33].Support vector machine method is better than neuralnetworks classifiers as multilayer perceptrons(MLP), Radial Basis Functions (RBF) and Co-Active Neural Fuzzy Inference Systems (CANFIS)for crop classification [34]. A number of studieshave been carried out on different crop acreageestimation using remote sensing and GIStechnology.
  3. 3. Amol D. Vibhute, Dr. Bharti W. Gawali / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.081-09183 | P a g eSupervised classification i.e. maximum likelihoodclassification algorithm was used to estimate thewheat crop acreage of Indore district, M.P. usingsingle date, and cloud free Resourcesat-1 (IRS P6)LISS-III data. This work has been carried out usingthe ground truth through GPS(Global PositioningSystem) and georeferenced by image-to-imageregistration existing master images then theimportant steps were used, like, extraction of districtmask of the image, generation of training signatures,test of statistical class separability and maximumlikelihood supervised classification [5]. MaximumLikelihood Classification and unsupervisedclassification techniques were used for Crop AreaEstimation on IRS-P6 LISS-III(Resourcesat-1) dataof different seasons kharif, Rabbi And Zaid in yearin Bundi Tahsil of Rajsthan. The visualinterpretation also performed for this particular studyand classified different land use and land coverobjects for extracting exactly crop area fromagricultural land Finally conclude that 688.92 sq.kmarea was occupied in agriculture while 73.12sq.km.Area is fallow land [7]. NDVI algorithm andunsupervised classification technique were used toidentify the sugarcane area as well as its conditionassessment by using toposheet and IRS LISS IIremote sensing data were used. NDVI image is verymuch useful in sugarcane crop area identificationbecause of its canopy cover, crop biomass and vigor.For this particular study two bands were used i.e.NIR and RED of LISS-II data then georeferencedthrough ground control points, NDVI image hasbeen generated and sugarcane area were extractedfrom NDVI image. For crop area estimation NDVIvalue of Band 4(NIR) and Band 3(RED) of IRSLISS-II data have been used and for classificationthe NDVI image cluster using ground truth havebeen used. And for assessing the sugarcane cropcondition, sugarcane area map has been used as a‘mask’ over the NDVI value map to generaterespective NDVI mask for sugarcane. NDVI value isvery much sensitive to crop canopy, biomass andvigour i.e. health condition of the crop. So, NDVImask map was then classified and given the classnames as ‘very good’, ‘good’, ‘moderate’ and ‘poor’for sugarcane health condition according to verifiedcrop condition by field observation and farmersknowledge. In referring healthy crops, reflectance inthe blue and red parts of the spectrum is low sincechlorophyll absorbs this energy. In contrast,reflectance in the green and near-infrared spectralregions is high. Stressed or damage crop experiencea decrease in chlorophyll content and changes in theinternal leaf structure. So examining the ratio ofreflected infrared to red wavelengths is an excellentmeasure of vegetation health. Healthy plants have ahigh NDVI value because of their high reflectance ofinfrared light, and relatively low reflectance of redlight. [35].MODIS (TERRA) time series (multitemporal) dataof three years using NDVI values and supervised,unsupervised as well as Hybrid classificationtechniques has been used for estimation of soya binacreage and crop yield estimation in districts of M.P.They performed NDVI on all time series images andcompared the scene average NDVI values forindividual district over the years, to get the predictedyield of soyabin. Crop yield was calculated bycomparison of the NDVI values of soyabin for theseyears [36]. Optical and microwave remote sensingdata has been used for corn monitoring and Cropyield estimation and production efficiency model(PEM) has been used to estimate crop yield onQuickBird imagery and getting higher accuracy ofpredicted results than Landsat TM imagery [37].Precision farming is a best technique to establish theagricultural management and variable rateapplication (VRA) is also plays a significant role inprecision farming [38].Due to the lower absorption of lightcomparing other objects on earth surface agriculturalcrops have higher spectral reflectance of signature.Vegetation frequently decreases absorption andincrease reflectance and transmittance in the visiblewavelength due to the dietetic scarcity in thevegetation. Hence, agricultural crops have higherreflectance of the spectral signature. In thisSituation, hyperspectral remote sensing plays animportant role as tool for monitoring and estimatingagricultural land use and cover because it canprovide earlier information than conventional landuse mapping methods. So, using hyperspectralremote sensing imagery development should beincreased in the field of agricultural land use andland cover for future planning and management [39].For managing the natural resources on theearth surface mapping of vegetation and land coveris essential. For planning, management anddevelopment the information on the land use andland cover in the form of maps and data is veryimportant [40].3. AGRICULTURAL APPLICATIONS3.1. Crop InventoryThe importance of spectral reflectance ofdata to identify and classify different crops is verymuch useful in crop area estimation, crop conditionmonitoring, and crop yield estimation. Remotesensing data have unique spectral signature whichuseful in crop identification and crop classification.When two crops have same spectral signature occurin a given date, then multi date data is required toidentify them [41]. Remotely sensed images are usedas mapping tools to classify crops, monitor the crophealth and capability and monitor farming practices.These are the some main agricultural applications inremote sensing: 1) Crop type classification and Crop
  4. 4. Amol D. Vibhute, Dr. Bharti W. Gawali / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.081-09184 | P a g eidentification, 2) crop condition (monitoring)assessment, 3) crop area and yield estimation, 4)mapping of soil characteristics, 5) Precision farmingpractices, etc.3.1.1. Crop type classification and CropidentificationThere are number of reasons to identify andclassify the crops. National/multinationalagricultural agencies and other agriculture relatedagencies are responsible to produce different croptype maps to prepare an inventory of what wasgrown in certain areas and when [4]. Properlyidentification of crops is very much beneficial forcrop acreage and crop production estimation [42].Remote sensing technique plays a vital role inidentifying and classifies different crops in acreage.It shows the vegetation health and growth based onthe spectral reflectance of the vegetation and otherpatterns. For crop identification and classificationthey use the multispectral and multitemporal dataand mainly supervised and unsupervisedclassification techniques. In Supervised techniquemachine require to give training signature whichclass is in the pixel, then identify examples of theInformation classes (i.e., crop type) of interest in theimage. These are called "training signatures". Butunsupervised classification is a method whichexamines a large number of unknown pixels anddivides into a number of classed based on naturalgroupings present in the image values. Unsupervisedclassification does not require analyst-specifiedtraining data [7], [43].3.1.2. Crop condition (monitoring) assessmentCrop condition monitoring is one of themain important advantages using remote sensing. Topreserve the countries food it is necessary to monitorthe crops timely and accurately [42]. Crop conditionmonitoring is very much useful in crop acreage andcrop yield estimation. Availability of water andnutrients, pest attack, diseases outbreak and weatherconditions these are the some factors that could beaffect to crop condition [41]. Crop condition isprimarily focused on individual physical parametersas well as different indices of the crop [31].It is difficult to ensure good agriculturalproductivity due to crop health assessment and earlydetection of crop infestations. Some diseases shouldbe detected early enough to provide an opportunityfor the farmer to mitigate due to for example,moisture deficiencies, insects, fungal and weedinfestations. To early detection of diseases remotesensing imagery provides frequently images atminimum within 2 days. The growth of the crops inthe field is not even, it can vary in the field from oneplace to another place of the farm. The differences ingrowth are result of various factors such as soilnutrition or other forms of stress. The remotesensing technology can help the farmer to identifysuch places in the farm where the growth isdecreased or slow; this will allow farmer to provideadequate amount of fertilizers, pesticides andherbicides to the crop in such areas. It will not onlyincrease the productivity of the farm land but it willalso reduce the farm input cost and theenvironmental impact will be minimum. The remotesensing image provides the required spatial overviewof the farm land. It allows a farmer to observe imageof his fields and make timely decisions aboutmanaging the crops. Remote sensing can be helpfulin identifying crops affected by conditions that aretoo dry or wet or that can be affected by insect, weedor fungal infestations or weather related damage.Images obtained throughout the growing seasonswill help to detect to problems but also to monitorthe success ratio of the treatment [4].To monitor the crop condition we need a cloud freedata, but crop signatures are not unique or thevariance in the signature of a single crop is too large,so it’s difficult to monitor the crop. In this condition,radar remote sensing plays a significant role tovegetation biomass and structure as well as thesesensors are very much useful in crop monitoring[37]. What is the crop condition means, there is noclear definition for crop condition and researchers inthe field of this area don’t have any idea about thisdefinition. If we don’t have any exact informationabout crop condition, then how could we monitor thecrop condition? Therefore it is urgent requirement toknow the crop condition means [44]. The use ofremote sensing methods in crop monitoring has beenreported by many researchers. These are the somemain useful methods for crop condition monitoringas following: Direct monitoring methodVegetation indices which are very muchuseful in crop condition monitoring, because ifvegetation indices higher, then the crop condition isbetter. This direct monitoring method is depends onvegetation indices which are easy to use andpromised but if large area or complex area will bebig then it should be failed [44]. Following table 1shows the main important indices useful in cropcondition monitoring.
  5. 5. Amol D. Vibhute, Dr. Bharti W. Gawali / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.081-09185 | P a g eTABLE 1. Useful Vegetation Indices for Crop Condition MonitoringVegetation Index Formula UsableBandsRVI(Ratio Vegetation Index)REDNIRRVI NIR, RED[26]NDVI(NormalizedDifference VegetationIndex) REDNIRREDNIRNDVINIR, RED[26]NDWI(NormalizedDifference Water Index)SWIRNIRSWIRNIRNDWINIR,SWIRSAVI(Soil AdjustedVegetation Index)LREDNIRREDNIRLSAVI))(1( NIR, RED[26]TVI (Triangular VegetationIndex))(200))(120(5.0 GREENREDGREENNIRTVI  NIR, RED,GREEN[26]TNDVI(TransformedNormalized DifferenceVegetation Index) 5.0REDNIRREDNIRTNDVI 1/2NIR, RED[26] Image classification methodTo monitor the crop growth statusgenerally having two methods which are veryimportant i.e. supervised and unsupervisedclassification methods using multitemporal satelliteimagery [44]. Same-period comparing methodWhile we are comparing previous yearremote sensing data (NDVI, LAI for instance) withnew one then we come to know the crop growingstate. Some vegetation indices such as differences,ratios are mostly useful in this method [44]. Crop growth profile monitoring methodThe crop growth profiling method is thedifferent between the year and year for the cropgrowth profile can it can be seen by the cropgrowing in the prolong time duration of thegrowing season. The crop growth profiles areproduced by getting the statistic of NDVI at thelevel of province. The time series data of NDVIduring the crop season is used for crop growthprofile. Different crops have differentcharacteristics in the crop growth profile usingNDVI; even the same crop grown in differentenvironment is having the different crop growthprofile. The NDVI profiles are useful in the cropgrowing conditioning [44]. Crop growing models methodIn this method the crop growing model isused to simulate the various stages in the crop lifecycle. The growing status and the estimated cropcondition are the result of the simulation [44]. Diagnosis modelThe Diagnosis model assesses the cropcondition using the characteristics of condition andenvironment that influence crop growth [44].3.1.3. Crop area and yield estimationCrop area estimation is very muchimportant for number of reasons; it is especiallyhelpful for yield estimation. Reliable and timelycrop acreage estimation could be helpful inagricultural planning and decision support systemto planners and policy makers for purchasing,storing, import export of food. Crop area estimationis the backbone of agricultural activities [5]. Thiscrop acreage estimation procedure is basicallydivided into following main steps are 1) Single datedata selection with maximum vegetative cropgrowth, 2) Identification of crop through imageusing ground truth, 3) Signature generation for thetraining site, 4) Image classification throughtraining statistics, 5) Crop area estimation usingadministrative boundary like district masks [41].There are three main methods of remote sensing toestimate the crop areas are followings [45]. Pixel counting and sub-pixel analysisUsing remote sensing imagery pixelcounting or sub-pixel analysis can be done toestimate area. But there are some limitations of thismethod for image classification due to same orderof the commission or the omission errors [45]. Regression, calibration and small areaestimatesThese methods are the combining catholicof regression, calibration and small area estimator’s
  6. 6. Amol D. Vibhute, Dr. Bharti W. Gawali / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.081-09186 | P a g ebut inaccurate information from remote sensingimagery with accurate information on a samplethrough ground surveys. Combination of fine ormedium resolution images provides findings ofarea estimation when ground surveys are notpossible due high cost [45]. Supporting area frame surveysThe different ways by which remotesensing imagery can support area frame surveysmay be stated as: to define sampling units, forlayer; as graphic documents for the ground surveyor for quality control [45]. Farmer haveinformation on crop yield at an early stage then itwill be profitable for farmer as well as countriesdevelopment which is dependent on agriculturalproduction, to propitiate the national requirementsfor the crops and also for income through exports[46]. The vegetation indices (VIs) generated fromremotely sensed images are usable for crop yieldestimation after analyzing the vegetation indicesand these indices correspond to the maximumvegetation stage of the crop. The average areaweighted vegetation indices calculated for eachdistrict is regressed with final yield at district.Direct weather variables such as rainfall,temperature and derived variables such as ET areused either independently or in combination inmultiple regressions. Sometimes, in the· absence ofweather data, purely statistical models such as AutoRegressive Integrated Moving Average (ARIMA)is fitted with time series data using standardtechniques. Crop wise production estimates arethus made based on the remote sensing basedacreage and yield estimates using a combination ofmodels [42].3.1.4. Mapping of soil characteristicsAn elevated understanding of the soil usedto increasingly better scale. It is beneficial inagricultural management and development.Traditional soil sampling and laboratory analysesmethods are very slow, valuable, and they couldnot reacquire all spatial as well as temporalchangeability of the soil quality, so this methodscannot efficiently provides the required information[47], [48]. In this condition Microwave (active andpassive) remote sensing plays a significant role inmapping of soil quality parameters. Mapping ofsoil characteristics is useful for different purposessuch as soil and crop management for improvingcrop yield estimation, sustainable land useplanning, soil erosion and runoff modeling inwatershed management, land – atmosphere gaseousexchange study for climate change modeling,biogeochemical cycles study and precisionagriculture etc [48].Research has been carried out using microwave(active and passive) remote sensing for mapping ofsoil characteristics. Some soil property parametersusing microwave remote sensing data for soilquality assessment are explained below. Soil SalinityCrop growth and productivity that couldbe affected by soil degradation; soil salinity is amain problem of soil degradation. For meliorationof the salt affected soil it is essential to identify ofsoil type and intensity of soil with their exactlocation and areal extent. Due to unwanted spectralmixing with sand and little spectral reflectance inblack soils areas it is difficult to the illustration ofsalt affected soil in littoral areas, infertile areas aswell as black clay rich soils regions, hence usingoptical remote sensing imagery it is feasible toillustration of salt affected soils [48]. Ravine Erosion InventoryFor utilization of depiction and mappingof ravine affected areas we commonly use opticalmultispectral as well as panchromatic fineresolution imagery and its classification is mainlybased on manual interpretation of the imagefeatures as well as its depth categories, becausethese methods are qualitative in nature. Terrainruggedness and vegetation penetration capabilityare highly sensitive to microwave SAR, by takingadvantage of microwave remote sensing such asSAR and InSAR (Interferrometric SAR) forillustration and characterization of ravinesquantitative terms such as ravine density, ravinedepth, and ravine surface cover [48]. Sand Dune CharacterizationHuge infertile areas are covered by sanddune and these fixed, semi fixed sand dunes aredistributed around the desert periphery with mobiledunes in the inner part of the deserts. Normallydesert areas are difficult to access; hence tounderstand the environmental changes in dry areasinformation about the dune properties is very muchhelpful. So, remote sensing imagery are helpful formonitoring desert environments [48]. Soil texture and Hydraulic PropertiesAn accurate estimation of spatiallyvariable soil physical properties such as texture andhydraulic properties is necessary to developreliable models of water flow and the efficientmanagement of soil resources for improving cropproductivity with an environmental quality. ToMeasure soil physical properties are time-consuming, expensive but a large number ofmeasurements are necessary to quantify theirspace-time variability [48]. Soil DrainageAlthough the various characteristics, soildrainage which directly affect of crop growth,water flow and solute transport in soils. Drainage is
  7. 7. Amol D. Vibhute, Dr. Bharti W. Gawali / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.081-09187 | P a g emajor issue at every where so soil information willhelp to took decision at various levels. Therefore,radar remote sensing has the effective way to mapand analysis soil properties, such as soil drainage[48]. Soil Surface RoughnessSoil surface roughness (SSR) hasinfluenced soil thermal properties, infiltration rate,surface run-off and susceptibility of soil to erosion[48].3.1.5. Precision farming practicesPrecision farming or agriculture is nothingbut the management of farm which is depends onnoticing and giving response on various changes inthe intra-field with the goal of increasing returns oninputs without changing the resources. Precisionagriculture is helpful to locate the exact position ofthe farmer in the field which is totally based onremote sensing, GIS and GPS or GNSS technology[49]. To increase yield, quality, and profit andreduce waste it is essential to control the cropproduction using water, seed, fertilizer, etc. In thiscondition, precision farming will be helpful fortaking right decision, in the right situation, in theright position, in the right way, at the right time tobecome eco-friendly [50]. Hence precision farmingopposes to conventional farming practices whichinclude crop treatments like irrigation, use offertilizers, pesticides and herbicides wereequivalently applied to the whole area withoutnoticing variability within the area. Remote sensingtechnologies are becomes advanced and the cost ofsensors are reduced, so we can easily use of thesefacility in the farming to identify the particular areain which the specific treatment is required for thatcrop as well as use of chemicals in required area, toreduce the amount chemicals used. So thesetechnologies will be helpful for protectingenvironment and saving the cost of the application[51]. Precision farming is an emergingmethodology which is designed to linkmanagement actions to site-specific soil and cropconditions, and place inputs of fertilizers,herbicides, and pesticides when and where they areneeded most to maximize the farm efficiency andminimize the environmental contamination. Thecore technologies which play an important role inprecision agriculture are GIS, GPS, and remotesensing. The importance of these technologies inagriculture was underscored when NASA (StennisSpace Center), in the early portion of the currentcentury, embarked upon the Ag 2020 program in aneffort to commercialize the geospatial technologies,develop practical tools for producers, andundertake projects with various types of crops toillustrate the utility of the technologies [31].4. USEFUL TECHNIQUES FOR LULCANALYSISTo analyze any particular area we need aGIS or remote sensing data. Those data are in theform of toposheet, maps and satellite or airborneimages. For remote sensing data there is lot oftechniques used by researchers to analysis theLULC in any area. The main techniques aresupervised and unsupervised classification, andothers are hybrid classification, box classifier,fuzzy classifier, Support vector machine,Normalized data vegetation index (NDVI), etc.These techniques are used for identification andchange detection in land cover and land use,classification of land use/land cover, agriculturalareas, patches, water bodies, as well asclassification of different crops and cropmonitoring etc.4.1. Supervised classificationWith supervised classification, we identifyexamples of the Information classes (i.e., landcover type) of interest in the image. These arecalled "training sites or different signatures" foreach class. For this method we require a prioriknowledge of the interested region [30]. There arenumber of supervised classifiers based on thestatistics, but the maximum likelihood classifier,minimum distance classifier, parellopipedclassifier, mahalanobis classifier are most usefulclassifiers [7].4.1.1. Maximum likelihood classifierThe maximum likelihood classifier is oneof the most popular methods of classification inremote sensing, in which a pixel with the maximumlikelihood is classified into the corresponding class.It is a statistical decision criterion to assist in theclassification of overlapping signatures; pixels areassigned to the class of highest probability.Gaussian maximum likelihood classifier classifiesan unknown pixel based on the variance andcovariance of the category spectral responsepatterns. This classification is based on probabilitydensity function associated with a particularsignature (training site). Pixels are assigned to mostlikely class based on a comparison of the posteriorprobability that it belongs to each of the signaturesbeing considered. Under this assumption, thedistribution of a category response pattern can becompletely described by the mean vector and thecovariance matrix [7], [52], [53].4.1.2. Minimum distance classifierThe minimum distance classifier is used toclassify unknown image data to classes whichminimize the distance between the image data andthe class in multi-feature space. The distance isdefined as an index of similarity so that theminimum distance is identical to the maximum
  8. 8. Amol D. Vibhute, Dr. Bharti W. Gawali / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.081-09188 | P a g esimilarity. It uses mean vector in each classsignature, while standard deviation and covariancematrix are ignored [53].4.1.3. Parellopiped classifierA parallelepiped is a geometrical shapewhose opposite sides are straight and parallel. Theparallelepiped classifier uses the class limits andstored in each class signature to determine if agiven pixel falls within the class or not. The classlimits specify the dimensions (in standard deviationunits) of each side of a parallelepiped surroundingthe mean of the class in feature space. Theparallelepiped classifier is typically used whenspeed is required. The drawback is (in many cases)poor accuracy and a large number of pixelsclassified as ties (or overlap, class 255) [53].4.1.4. Mahalanobis classifierIt is based on correlations betweenvariables by which different patterns can beidentified and analyzed. It gauges similarity of anunknown sample set to a known one. It differsfrom Euclidean distance. It takes into account thecorrelations of the data set and is scale-invariant[54].4.2. Unsupervised classificationUnsupervised classification is a methodwhich examines a large number of unknown pixelsand divides into a number of classed based onnatural groupings present in the image values.Unsupervised classification does not requireanalyst-specified training data [7], [53]. The twocommon types of these classifiers are K-meansclustering and ISODATA (Iterative Self-Organizing Data Analysis) [7], [30].4.3. Hybrid classifierThe combination of supervised andunsupervised classifier is a hybrid classifier. Thisclassification produced more accurateclassifications than the supervised classification;however, it did not improve the accuracysignificantly in com-parison to the unsupervisedclassification [14].4.4. Fuzzy classifierFuzzy logic is a soft computing technique.These soft computing techniques are better thanhard computing techniques, because these softcomputing techniques like as fuzzy logic, neuralnetwork, machine learning, can gives better facilityto handle mixed pixels [26]. Fuzzy classification isthe process of grouping elements into a fuzzyset whose membership function is defined by thetruth value of a fuzzy propositional function anduses degree of membership functions in between1.0 and 0.0 for full membership and nonmembership respectively [55].4.5. Normalized data vegetation index (NDVI)NDVI is based on the principle of spectraldifference based on strong vegetation absorbancein the red and strong reflectance in the near-infrared part of the spectrum. NDVI (NormalizedDifference Vegetation Index) is mostly usefulmethod to classify the land cover vegetations [56],using this formula NDVI= NIR-RED/NIR+RED[57], [58], [59] and calculated using the RED (0.6to 0.7 µm) NIR (0.7 to 1.1 µm) wavelengths [29],[35], [59]. NDVI is a single band image having thevalues between -1 to +1, the value closer to +1 willtells about presence of healthy green vegetation[57], [35], [59].5. CONCLUSIONThere is lot of remote sensing techniquesused by many researchers for land use/land coverclassification, identification, classification andanalysis of any area or any region but Maximumlikelihood classification technique is most usefultechnique in supervised classification, because itgives greater accuracy than other techniques.Unsupervised technique is the second importantmethod for classification as well as identification.Fuzzy classification (also known as fuzzyclustering) algorithm is the best useful classifier forsame elements. And Hybrid classifier is the lowestuseful technique for classification. Normalized datavegetation technique (NDVI) technique is the mostuseful technique for land cover analysis as well ascrop condition monitoring. Soil characteristics arevery much useful for crop growth and crop yield.Precision farming is useful to enhance theagricultural production, to reduce the chemical usein crop production, to use water resources, toimprove quality, quantity & reduced cost ofproduction in agricultural crops with the help ofRS, GIS and GPS technology.REFERENCES[1] Meliadis Ioannis, Miltiadis Meliadis,Multi-temporal Landsat imageclassification and change analysis of landcover/use in the Prefecture ofThessaloiniki, Greece, Proceedings of theInternational Academy of Ecology andEnvironmental Sciences, 1(1): 15-25,2011.[2] D.K. Tripathi and Manish Kumar, RemoteSensing Based Analysis of Land Use /Land Cover Dynamics in Takula Block,Almora District (Uttarakhand), J HumEcol, 38(3): 207-212, 2012.[3] http://en.wikipedia.org/wiki/Land_cover/Land_use.[4] http://www.nrcan.gc.ca/sites/www.nrcan.gc.ca.earthsciences/files/pdf/resource/tutor/fundam/pdf/fundamentals_e.pdf.
  9. 9. Amol D. Vibhute, Dr. Bharti W. Gawali / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.081-09189 | P a g ehttp://geography.huji.ac.il/personal/Noam%20Levin/1999-fundamentals-of-remotesensing.pdf.,p.p-3,4,5.[5] Goswami S. B., Dr. Aruna Saxena and Dr.G. D. Bairagi, Remote Sensing and GISbased wheat crop acreage estimation ofIndore district, M.P., InternationalJournal of Emerging Technology andAdvanced Engineering, Volume 2, Issue 3,March 2012.[6] http://www.seosproject.eu/modules/agriculture/agriculture-c00-p01-a.html.[7] Sandeep Kr. Soni, Crop Area Estimationfor Bundi Tahsil of Rajasthan usingRemote Sensing and GIS Technique,Geospatial World Forum, Hyderabad,India, 18-21-janeuary 2011.[8] Amber L. (Williams) Coleman and JohnM. Galbraith, Using GIS as anAgricultural Land-Use Planning Tool,Virginia Agricultural Experiment Station,Bulletin 00-2, Department of Crop andSoil Environmental Science College ofAgriculture and Life Sciences VirginiaTech, December-2000.http://www.vaes.vt.edu/research/publications/index.html.[9] Ujoh Fanan, Kwabe Isa Dlama andIfatimehin Olarewaju Oluseyi, Urbanexpansion and vegetal cover loss in andaround Nigeria’s Federal Capital City,Journal of Ecology and the NaturalEnvironment Vol. 3(1), pp. 1-10, January2011.http://www.academicjournals.org/jene.[10] Lakshumanan.C, Pradeep Kishore.V,Viveganandan.S, Krishnakumar.P,Muthusankar.G, Landuse / Land coverdynamics study in Nilgiris district part ofWestern Ghats, Tamilnadu, InternationalJournal of Geomatics And Geosciences,Volume 2, No 3, 2012.[11] Emilio Chuvieco and Alfredo Huete,Fundamentals of Satellite Remote Sensing(CRC Press, Taylor & Francis Group,2010).[12] Ian Heywood, Sarah Cornelius and SteveCarver, An Introduction to GeographicInformation Technology (Third edition,Pearson, 2010).[13] Arun Mondal, Subhanil Guha, PrabhashKumar Mishra, Sananda Kundu, LandUse/Land Cover Changes In HugliEstuary Using Fuzzy C-Mean Algorithm,International Journal of Geomatics AndGeosciences, Volume 2, No 2, 2011.[14] Offer Rozenstein, And Arnon Karnieli,Comparison Of Methods For Land-UseClassification Incorporating RemoteSensing And GIS Inputs, EarselEproceedings 10, 1/2011.[15] BHAGAWAT RIMAL, Application ofRemote Sensing And GIS, Land Use/LandCover Change In KathmanduMetropolitan City, Nepal, Journal ofTheoretical and Applied InformationTechnology, JATIT & LLS, 2005 – 2011.[16] Dr. Lazarus M. Ojigi, Analysis of SpatialVariations Of Abuja Land Use And LandCover From Image ClassificationAlgorithms, ISPRS Commission Vii Mid-Term Symposium, Enschede, TheNetherlands, 8 – 11th May 2006.[17] Sudeesh S and Sudhakar Reddy C,Vegetation and Land Cover Mapping ofNagarjunasagar-Srisailam Tiger Reserve,Andhra Pradesh, India using RemoteSensing and GIS, International Journal ofGeomatics And Geosciences Volume 2, No4, 2012.[18] Atiqur Rahman, Sunil Kumar, ShahabFazal and Masood A Siddiqui,Assessment of Land use/land coverChange in the North-West District ofDelhi Using Remote Sensing and GISTechniques, Indian Society of RemoteSensing, DOI 10.1007/s12524-011-0165-4, 05 January 2012.[19] T. V. Ramachandra, Uttam Kumar,Geographic Resources Decision SupportSystem for land use, land cover dynamicsanalysis, Proceedings of the FOSS/GRASSUsers Conference - Bangkok, Thailand,12-14 September 2004.[20] Daniel Ayalew Mengistu and Ayobami T.Salami, Application of remote sensing andGIS inland use/land cover mapping andchange detection in a part of south westernNigeria, African Journal of EnvironmentalScience and Technology Vol. 1 (5), pp.099-109, December, 2007.[21] Seyed Zeynalabedin Hossein, SayedJamaleddin Khajeddin, Hossein Azarnivadand Seyed Ali Khalilpour, Land UseMapping Using ETM+ Data (CaseStudy:Chamestan Area, Iran), College OfNatural Resources & Desert Studies,University Of Yazd, P.O.Box 89195-741Yazd, Iran.http://www.isprs.org/proceedings/XXXV/congress/comm7/papers/75.pdf.[22] Haruhisa Shimoda And Kiyonari Fukue,Eastern Asia Land Cover ClassificationUsing Modis Surface ReflectanceProducts, ISPRS Archives Xxxviii-8/W3Workshop Proceedings: Impact OfClimate Change On Agriculture, Tokyo151, Japan.http://www.isprs.org/proceedings/XXXVII
  10. 10. Amol D. Vibhute, Dr. Bharti W. Gawali / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.081-09190 | P a g eI/8-W3/B5/1-6-125-----------------12%20Dec%20formatted.pdf.[23] Edwin Martínez Martínez, RemoteSensing Techniques For Land UseClassification of Rio Jauca WatershedUsing IKONOS Images, Agricultural andBiosystems Engineering Department,University of Puerto Rico-Mayagüez.http://gers.uprm.edu/geol6225/pdfs/e_martinez.pdf.[24] Anil Z Chitade, Impact Analysis Of OpenCast Coal Mines On Land Use/ LandCover Using Remote Sensing And GISTechnique: A Case Study, InternationalJournal of Engineering Science AndTechnology, Vol. 2 (12), 7171-7176, 2010.[25] Vemu Sreenivasulu and PinnamaneniUdaya Bhaskar, Change Detection inLanduse and landcover using RemoteSensing and GIS Techniques,International Journal of EngineeringScience and Technology, Vol. 2(12), 7758-7762, 2010.[26] Vijaya Musande, Anil Kumar andKarbhari Kale, Cotton CropDiscrimination Using Fuzzy ClassificationApproach, Indian Society of RemoteSensing, DOI 10.1007/s12524-012-0201-z, 6 January 2012.[27] Carlos Antonio Oliveira Vieira, PaulMather and Paul Aplin, Agricultural CropClassification Using The Spectral-Temporal Response Surface, Anais XISBSR, Belo Horizonte, Brazil, INPE, 05-10 April 2003.[28] Tippathoo TEEKAPAKVISIT, CropIdentification Using Radarsat-2 Multi-Polarization Data : A Case Study OfLopburi Province, Thailand, Geo-Informatics Scientist, Institute of SpaceKnowledge Development, Geo-Informaticsand Space Technology DevelopmentAgency (Public organization), 120(Building B) Chaeng Wattana Road, LaksiDistrict, Bangkok 10210, Thailand.[29] G. S. Jang, K. A. Sudduth, E. J. Sadler andR. N. Lerch, Watershed-Scale Crop TypeClassification Using Seasonal Trends InRemote Sensing‐Derived VegetationIndices, American Society of Agriculturaland Biological Engineers, Vol. 52(5):1535-1544, 2009.[30] Cankut ormeci, ugur alganci and elifsertel, Identification of Crop Areas UsingSPOT – 5 Data, TS 4H – Remote Sensingand Imagery I, Sydney, Australia, 11-16April 2010.[31] M. Duane Nellis, Kevin P. Price andDonald Rundquist, Remote Sensing ofCropland Agriculture, Papers in Natural,Resources. Paper 217, The SAGEHandbook of Remote Sensing, (SAGEPublications. 14 Apr. 2010), pp.4, 5, 13,14., 1-1-2009.[32] Sebastian Mader, Michael Vohland,Thomas Jarmer and Markus Casper, CropClassification With Hyperspectral Data ofThe Hymap Sensor Using DifferentFeature Extraction Techniques,Proceedings of the 2nd Workshop of theEARSeL SIG on Land Use and LandCover, University of Trier, Germany, 28-30 September 2006.[33] Shwetank, Jain Kamal and Bhatia K. J.,Review of Rice Crop Identification andClassification using Hyper-Spectral ImageProcessing System, International Journalof Computer Science & Communication,Vol. 1, No.1, January-June 2010, pp. 253-258.[34] G. Camps-Valls, L. G´omez-Chova, J.Calpe-Maravilla, E. Soria-Olivas, J. D.Mart´ın-Guerrero and J. Moreno, SupportVector Machines for Crop ClassificationUsing Hyperspectral Data, InformationSociety Technologies (IST) programme ofthe European Community, Spain.[35] Md. Rejaur Rahman, A.H.M. HedayutulIslam and Md. Ataur Rahman, NDVIDerived Sugarcane Area Identification andCrop Condition Assessment, Dept. ofGeography & Environmental Studies,University of Rajshahi,Rajshahi-6205,Bangladesh.http://www.ida.liu.se/~746A27/Literature/NDVI%20derived%20sugar%20cane%20area%20identification.pdf.[36] Abhishek Kumar Maurya, Estimation ofAcreage & Crop Production ThroughRemote Sensing & GIS Technique,Geospatial World Forum, Hyderabad,India, 18-21-Janeuary 2011.[37] Jesus Soria-Ruiz, Yolanda Fernandez-Ordonez and Heather McNairn, CornMonitoring and Crop Yield Using Opticaland Microwave Remote Sensing,Geoscience and Remote Sensing, Pie-GeePeter Ho(Ed.), ISBN:978-953-307-003-2,In Tech, Mexico, Canada,01,October,2009.[38] Çağatay TANRIVERDİ, A Review ofRemote Sensing and Vegetation Indices inPrecision Farming, KSU. Journal ofScience and Engineering 9(1), KSÜ. Fenve Mühendislik Dergisi, 9(1), 2006.[39] Kamaruzaman Jusoff, Land Use andCover Mapping with AirborneHyperspectral Imager in Setiu, Malaysia,journal of agricultural science,vol.1,No.1,December-2009.
  11. 11. Amol D. Vibhute, Dr. Bharti W. Gawali / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.081-09191 | P a g e[40] Sudeesh S And Sudhakar Reddy C,Vegetation And Land Cover Mapping OfNagarjunasagar-Srisailam Tiger Reserve,Andhra Pradesh, India Using RemoteSensing And GIS, International Journalof Geomatics And Geosciences, Volume 2,No 4, 2012 .[41] Shesha Sai MVR, Ramana KV andHebbar R, Remote Sensing Applications inAgriculture, (National Remote SensingCentre, ISRO, Department of Space,Government of India, Balanagar,Hyderabad-500 625, A.P., India), 1-21.[42] R. Kalpana, S. Natarajan, S. Mythili, D.E.Shekinah and J. Krishnarajan, RemoteSensing For Crop Monitoring - A Review,Agric. Rev, 24(1):31-39, 2003.[43] http://www.sc.chula.ac.th/courseware/2309507/Lecture/remote18.htm.[44] Meng Ji-Hua and Wu Bing-Fang, StudyOn The Crop Condition MonitoringMethods With Remote Sensing, TheInternational Archives of thePhotogrammetry, Remote Sensing AndSpatial Information Sciences. Vol.XXXVII., Part B8., 2008.[45] F.J. Gallego, Review of The Main RemoteSensing Methods For Crop AreaEstimates, ISPRS Archives Xxxvi-8/W48Workshop Proceedings: Remote SensingSupport To Crop Yield Forecast And AreaEstimates, Agriculture Unit, IPSC, JRC,21020 Ispra (Va), Italy, 2006.[46] http://www.seos-project.eu/modules/agriculture/agriculture-c02-p01.html.[47] Mehrez Zribi, Nicolas Baghdadi andMichel Nolin, Editorial Remote Sensingof Soil, Applied and Environmental SoilScience, Volume 2011, Article ID 904561,2 pages, France, 21 November 2011.[48] S K Saha, Microwave Remote Sensing InSoil Quality Assessment, ISPRS ArchivesXxxviii- 8/W20; Workshop Proceedings:Earth Observation For TerrestrialEcosystems, Agriculture And SoilsDivision, Indian Institute Of RemoteSensing (Iirs), Indian Space ResearchOrganization(Isro), 4, Kalidas Road,Dehradun – 248001, India.[49] http://en.wikipedia.org/wiki/Precision_agriculture.[50] http://www.ncpahindia.com/v2/precision-farming.[51] http://www.seos-project.eu/modules/agriculture/agriculture-c04-p01.html.[52] http://wgbis.ces.iisc.ernet.in/energy/paper/MODIS/intro.htm#gmlc.[53] http://www.sc.chula.ac.th/courseware/2309507/Lecture/remote18.htm.[54] http://en.wikipedia.org/wiki/Mahalanobis_distance.[55] Md. Rejaur Rahman and S.K. Saha,Multi-resolution Segmentation for Object-based Classifi cation and AccuracyAssessment of Land Use/Land CoverClassifi cation using Remotely SensedData, J. Indian Soc. Remote Sens. (June2008) 36:189-201, 30 November 2007.[56] Hojat Ahmadi And Kaveh Mollazade,Application of Remote Sensing In TheAgricultural Land-Use Assessment,Journal of Theoretical and AppliedInformation Technology, Jatit, 2005 -2009. Www.Jatit.Org.[57] Ramchandra T.V., Regional land covermapping using remote sensing data,Journal of Agricultural, food andEnvironmental sciences, Volume 2, Issue1, 2008.[58] J.T. Aranha, H.F. Viana And R.Rodrigues, Vegetation Classification andQuantification By Satellite ImageProcessing. A Case Study In NorthPortugal, International Conference AndExhibition On Bioenergy, April 6th – 9th2008.[59] Guifeng Han and Jianhua Xu, Vegetationclassification in eastern china using timeseries NDVI Images, Proc. SPIE 6790,MIPPR 2007: Remote Sensing and GISData Processing and Applications; andInnovative Multispectral Technology andApplications, 67901N,doi:10.1117/12.749124, 14 November2007.