 Today the technology is available the raw research has happened which had come for applications where they try to handle the data with the finest cell like you know maybe if some of your teaching GIS you may be knowing quatery we try to put the data in a quatery form in such a way wherever homogeneity is there you do not further divide whenever you have a smaller cell you further divide into quadrant, quadrant so that you can store the data to the resolution you need it. Quatery is very good as far as handling the thematic maps are concerned not for remote sensing data directly for handling it. Also there are run length encoding value point encoding these are all methods where people try to adapt for storing the thematic data and processing the thematic data. Good algorithms are available where they try to store so that one can even in raster go for the smallest or the finest resolution possible and then handle it but only thing is if you are a user of a GIS package you should know whether your GIS tool will it handle it because there are packages which give those kind of storage mechanisms there are packages which don't so which tool you are using based on that whether if it supports a quatery model finest resolution it can handle or whether it has some encoding methods you can still handle the finest resolution data without any difficulty one is raster next is vector vector is nothing as I said fundamentally x comma y is important for us when you put it on a ordered form of x y's you generate a line and I generate a line I go on close it I get generate an area how do we define an area if you go in mathematics how do we define us we have something called a Jordan curve we say a curve which closes a closed curve which has an interior and a boundary so whenever you have a curve and you have an interior area naturally it becomes a polygon okay so it can have a point it can be a line it can be a polygon as I said the basic units in vector which can be handled and you know when we handle it again try to handle point line and polygon it's a very good model you can you need not worry about the resolution of the pixel but the problems are whenever you are raster very easy to analyze very easy to process because as I said whatever is happening in image processing we can pull it and use it in raster vector has its own theory computational geometry plays an important role when you want to handle the data in a vector mode in a GIS okay so analysis processing they are not as simple as you do in raster okay so points we put lines we try to put polygons we try to put a three major categories whenever you handle the data in a vector representation easy to store accuracy can be very high but difficult to process so the when you come to the storage in a vector model the point that comes into picture is the first thing is how do we store this see raster you cannot do it you put a grid you store it only thing is I said quarter people can follow or run length encoding people can follow value point coding they can follow to compress the storage but in vector what we do you take the data you simply try to store it simply like this I take the point data position give the value x comma y the id 10 then you know I give a line x1 y1 x2 y2 x3 y3 I store it as an arc give the value okay one can store the polygons so this kind of structure only people started storing in vector in the initial stage to store the data but what we all had found as a limitations if you look at the data you look at there is a common boundary between 63 and 64 can just see there is a common boundary between two polygons we are only storing two one polygons here and there is a common boundary between both of them so what we try to store we store 63 as one entity 64 is another entity with a common boundary between them so the common boundary is stored twice now and you know even you store twice doesn't matter but you know a person who would have digitized who he would have done the same what he says what is the guarantee see you would have done this once he would have done this common boundary again so you know when you zoom it you will see this common boundary may not be sitting one over the other this may happen so what see of course many tools support still that kind of storage so the first one is we call it such non topological structure where you store as you see you do not bother whether I have to share a common boundary that kind of model was called as a kind of a non topological data structure to handle the vector data then came the topology into mind so when topology came what they did in fact you know topology the contribution topology is a theory in mathematics where we try to work on the relations between geometry projects because you know if you go a little deeper in topology what it does is topological relations are invariant they don't change whenever you do a skewing where whenever you enlarge it whenever you shrink it see I am in front of you whatever happens to this room whether they rotate it whether they enlarge it or whether you shrink it that's a relation a topological relation it is invariant whenever you do scaling or rotation or you know enlargement etc that holds good for GIS because sometimes we are going to enlarge the map and sometimes we are going to reduce the map so you know that topology concept has been picked up well used here to define the relations between the objects so what they have done nothing great what they introduced is in a simple term you wanted can tell you they take the common boundary instead of treating the whole polygon as a polygon you treated this is one arc this is another arc this arc is again common this is another arc can I not say this one two makes this polygon 64 minus one three makes this polygon 63 so what I am saying plus one and minus one so I am tracing once like this I am tracing this way other time so with the direction when I put it the common arc I can store the polygons with common boundary only once that becomes possible that is what the concept they introduced in topology what they started doing it is whenever they had to store the polygon features they started defining a polygon as a an arc or a chain of segments they started defining as a chain of segments defining the common arc only once telling which is the left polygon which is the right polygon how do I define the chain of segments to form a polygon it is not only that when the concept of topology came they were able to even define a boundary which is interior within a boundary like you know there may be an island there may be a water body within a village when I woke out about the land use and this water body area I may like to subtract so what happens it becomes a hole the polygon inside of polygon is a hole in that even that they were able to define and a polygon with multiple interior boundaries it can have many holes in it that was also defined so topology was very well picked up used so they introduced topological concepts in a GIS and you know which improved a lot the method of storage because good contributions had been done again by the people who worked on research in US especially first topology was implemented in their census data they came with a coding called you know dime with that coding they introduced to store their census tracks Canada some people worked to store their data and which was adopted later by the GIS packages so they introduced topology but today if you ask me what is the latest development going on is they are not much insisting on topology but what they are insisting is the common boundary must be same they don't mind repeating the boundary but they don't want unwanted polygon in between okay common boundary will be if you digitize today's in our GIS you will see it will talk you talk you talk you and fill it so common boundary can be same they don't mind repeating it they don't want to introduce noise in that common boundary that's the way the trend has gone so once you have a polygon with an arc of segments naturally gets defined even with an interior polygon in it and then all attributes can be stored and you know you can get the data you can immediately put the whatever the add-on attributes put it in different tables etc all these things are possible but whenever the topology has to be built one has to keep in mind whenever you digitize the data once vector databases created cleaning editing and building topology have to be created done it has to be done in case you want a clean common boundary between two of them because errors such these things may happen see suppose you have a segment which is hanging which is not touching the boundary how can it become a common boundary between two polygons when you have an overshooting segment where it will go this part so you know normally there are procedures to clean the data if you are using a GIS tool it provides it so cleaning editing a must because see in a GIS what we are working see if you talk about accuracy in GIS we talk about precision also see we work on tenth decimal and eleventh decimal of a meter on the ground that much we try to store the data so even in a small segment hanging is wrong it won't accept it because it will come bar both the nodes up to tenth decimal accuracy so you know when you zoom zoom zoom where your data may look very good when you zoom you will see there will be a segment hanging system will not allow you to go to the next step so you know cleaning becomes a must editing becomes a must if your data is cleanly done without any error topology gets built up automatically in the GIS tools which support topology okay and you know once it is done one can create the database etc etc and just to talk about since i spoke about the two data models when do we go for raster when do you go for vector because you know both have advantages both have limitations there are books which talk again you know putting big tables they write compare what is raster what is vector why should we go for raster why should we go for vector again it depends on your case study if you work for a i'll say for an impact analysis of element or you are going for a natural resources application which are the area where i have to cultivate paddy that kind of question you have to go for a raster you are working for an alignment problem a transportation problem your location-based services problem a service center problem a telecommunication problem you must go for vector okay so wherever you try to use remote sensing data try to go for raster wherever you go for a linear feature where your application is there better go for vector so you know both models are plus points and minus points and today every GIS handles both the models not like you know 12 15 years back your GIS means it's a vector GIS it's a raster GIS today everything is hybrid you can sit out from here to that that you hear so you know both have greatness so based on the application user can choose and of course another model is TIN TIN is year and extension of vector model where people try to store the elevation data triangulated irregular network we say this is basically used for topographic data elevation data from which people can generate grid DEM digital elevation model can be generated contours can be generated slope and aspect can be generated etc TIN also is one of the models used for elevation data okay and then comes after the spatial objects then naturally our non-spatial attributes come into picture where we try to store the non-spatial data so we try to store in using any one of the rdbms tools see database management system is a software tool which handles the normally the tabular data so in GIS we try to pull the dbms tool for handling the non-spatial data only thing is there should be a link between your spatial object and the tabular data which you are storing if there is a unique ID to your point there will be that unique ID will be stored in your record in the table so that one can link it and query it so rdbms tools are used one can go for the cheap rdbms tool like you know ms access one can use the tool like article based on the application one can use for handling the non-spatial data so whenever you try to store it how do we link whenever you create let me say a point layer i digitize the points system will automatically create the table for you based on the dbms you select that point x comma y it will give a system an ID and whatever you are given as a user ID well one well two if you say it will also store moment that well one well two is there you get all other attributes and store in it so whether you have point or a line or a polygon system will create a default table where you can append your own attributes and store it okay and as far as the data collection for both the models are concerned see in gis one is hardware second is okay you buy the gis software either you can go for you know millions of dollar software you can go for cheaper software also okay and tool gives you all facility to do analysis etc but we all spend a lot of time and money on data because you know if you put nice system will give you nice if you put a good data system will give you good result because it cannot think so the thing is much of time money labor everything we spend in a gis on data okay see once you do a primary survey naturally see i work for bombay i say i buy the latest remote sensing data spending so many lakhs of rupees naturally my land use map will be very good i go and borrow okay you have this mmid i go take some data somebody i go take some data naturally they would have done it using some method sometime i have to be satisfied so one can do a primary survey can do a secondary source data but it depends again on the time and cost and of course if you go for raster naturally remote sensing becomes the best to source today as far as the ready raster is concerned similarly vector what we are all now trying to use is either a total station r a gps a geode degree sever you can take it go and collect your locations you can map the boundaries take the data readily in vector form so the data is possible as a primary source also as a secondary source also people can pull the data the question is quality one has to keep in mind if you depend on secondary sources the certainty and the accuracy people who are using may not okay so we saw the data model about the spatial objects now we come to the analysis okay so in analysis when you say normally what gis can do it is it can answer you what is at a given location very simple go to a place pick it just it will say what is that and you ask a query show me the area where the rainfall had been greater than 700 millimeter slope is less than 10 degrees and you know there is a water body nearby etc etc it can see or you are at this place and you want to find out the where is the closest to market available closest to bank available it closest to bank or closest to market when i say it need not be euclidean distance it will be actually on the row distance which you have to walk it is not you know crowfly distance system can actually tell you which is the distance available where you can and you know you can do spatial temporal modeling trends how it has changed how was earlier how is today how it may change tomorrow one can do the analysis and the wonder is one can add any model with gis see earlier days you know adding models in gis was difficult we used to take the data put it out and do it and bring it back and all today embedded models are possible why because you know the way the software has changed is today gis has become not only as a full-fledged tool they give you as components they give you as like kind of you consider like libraries so you customize the whole thing for your application using those libraries plus whatever the model you want to do you integrate it and make it as a full sdss for you so you know what happens if any model can be put so that the gis with this functionality and your model can give the answer okay so broadly speaking on spatial analysis it can be categorized as queries distance calculations overlay what you do neighborhood operations and the connectivity function connectivity function we normally use what for network analysis shortest path finding etc so you can broadly categorize them and to these categories first let us see the basics of raster operations you have raster data let's say remote sensing data or you have digitized you are scanned you have a map first thing is what you don't need to do any analysis simply you can use this as a display like a digital cartography simplest is display the data when i say display the data there are many ways one can display i show a map with colors with a legend we have not done anything we have just plotted a scale show it put a north arrow with a legend and you can show or you can show in the display in the visualization as a graded color suppose you are working on temperature you can say high to low or you are working on elevation dm low to high or ranges of elevations can be stored or you know the third way is you can store just the contours alone or you can show a subspective view today again in gis there are modules available where you can fly through you can animate it you have the elevation data you drape over it actual land use and virtually you can fly through the area that much is possible okay nothing they all can be raster so the first type is simple display you do nothing you can display the data second is broadly speaking about raster the way they categorize the operations are in three local focal zone if you see the literature they'll say that three broad categories they run the operations as far as the raster is concerned what is local local means i have supposed the layers of data if i run through pixel by pixel one below the other if i take fifth row eighth column in the next layer also fifth row eighth column next layer also fifth row eighth column i am going vertically down among the layers that is a local operation i am taking a layer i'm not bothered who is adjacent to me i am going vertically down with multiple layers pixel by pixel down like you have a multi band data in remote sensing you don't bother who is adjacent you take a four band as a vector and you process it the same idea here whenever you try to do the same position and go below on multi layers it is called as a local operation okay in local operation plenty can be done in GIS almost every raster GIS tool gives a lot of analysis simplest i'll tell you is map algebra map algebra if you have seen that is the simplest which is very very common in almost all raster GIS tools people might have used probably Idrisi ilvis gram plus plus anybody has used see normally we all use it the word map algebra map algebra is nothing you have multiple maps one sitting below the other if it is 100 rows by 100 columns this also must be 100 rows by 100 columns put one below the other you can run the query as i said show me the areas where slope is less than 10 degree rainfall is greater than 300 millimeter the area has water body within one kilometer you run all with an expression like if then if then like we write a command an instruction in a programming code you run a long instruction but instead of variable here everything is a map so you give a big algebraic expression with the maps system will give you the answer which packet satisfy your query so you know map algebra is something which runs fully on a local operation any number of layers possible and you know not only you run if then query it can be algebraic five star a plus three star b where a is a map b is a map it can be algebraic it can be arithmetic it can be relational it can be logical it can be all operations can be run in a map algebra when you have all layers sit in your thing simply pull that map algebra module give the expression the operation is done using the local operations because it runs through the layers of pixels i have simple example i'll tell you i have let us say monthly rainfall for this city january february march like that 12 months i have i want to get the map showing maximum it will go through all the 12 pixel values in that whichever is the maximum value it will put it into your output pixel for every pixel it will maximize find out and put it in your output pixel so maximize is a common operation minimize average so any number of layers you can run it map algebra supports it okay so local operation the best is map algebra which people can use it to run n number of analysis functions okay it gives a generic tool which people can use it and of course another thing is normally we use the word overlay isn't it in gis all of us use the word overlay because we all are pick this gis from the idea of cartographer cartographer picks a map puts one or the other maps it what our draftsmen used to do take that map again put another map and map it same thing we are trying to do digitally so overlay is again a common operation which people do it so take a map how many polygons are there take another map put over it both will combine and give more polygons that is also a local operation why local operation say i have a polygon here with four squares and another polygon with the diagonals when i put one over the other i get now many small small polygons and that area it will say from the first map i have taken soil one from the second map slope one so my new polygon contains the category soil and slope okay so overlay is a common operation where people run it with two layers and combine the themes here and here and create a new map and whenever it combines it it creates a table also along with that telling which theme and which theme had been combined to create the new so you know system gives that detail also so when you have local operation well known is one is as i said bap algebra second is two maps overlay third is there is something called conditional overlay see suppose you have two maps let me take again soil and slope you have two categories of soil two categories of slope you know what you want to combine and give a new value as per your analysis before hand itself you can give if this combination and this combination occurs you give me new output value one or you give me in the new map value 10 what you give accordingly also it can so it can do overlay conditional overlay map algebra all these things are all well known operations and local okay this is an example which shows us a local i mean overlay done unconditional overlay done between a taluka boundary map and a lithology map and we saw local next i said is focal focal is nothing i take a map i single layer here it is not multiple layer you have to keep in mind in a single layer i try to use my closest neighborhood to compute something okay i don't know whether you add a you might have had a lecture on remote sensing what we do in remote sensing we try to work on edge operator how do we do edge operator i run some filters am i gray value i take my surrounding gray value i have a convolution filter with which i try to run it and get the edges whenever there is a change of pixel value it delineates the edges same way here we can when you have a neighbor you try to use the adjacent 8 pixels or probably one more level it can be 3 by 3 it can be 5 by 5 take it compute it and get your value so in a gis what can be the best application for neighborhood it can be suppose i have the elevation data i can use for calculating slope so your focal operation the best example is of course you can smooth an image you can run filtering and all plus slope aspect etc are computed in a gis using focal operation where we take 3 by 3 neighbor okay how a slope is computed on all i'm not covering the algorithm but you can refer in any book they give the formula you have an elevation when you have nine eight elevations around you you find out where your gradient is maximum among the difference of height get the tan theta put that in your center pixel or take the theta put it you get your slope map okay for elevation finding a slope is a focal operation in the perspective of gis so neighborhood operation can be filtering smoothing the images or you know you can calculate slope and what is aspect moment you get a slope map in which direction your slope is following that is what you try to store it can be zero degree 45 or 90 or whatever it comes you store so a slope map can be generated from elevation data an aspect map can be generated a relief map can be generated they are all neighborhood operation which we call it as focal operation so local is possible focal is possible and then comes extended neighborhood instead of just a three by three pixel you can use why not suppose I have a hundred rows by hundred columns I am standing here this is a location of a school that is again a location of a school you would like to find out in your map who are all closer to you so that this school can be used by this district boundary and that school can be used by those people living there so you would like to make it instead of only using the closest pixel you are trying to use more pixels so that you can make a proximity map so we are not only sitting with only one level pixel we are going more another application of extended neighborhood is common operation we use in GIS is buffering buffering you never do with only one pixel I have a road I would like to protect one kilometer around the road or half a kilometer around the road and your pixel maybe your resolution maybe 30 meter so how many pixels you need come say come 30 pixels so whenever you do a buffering whenever you do a distance calculation whenever you make a proximity map they are all extended neighborhood operation extension of focal okay because among that buffering is very very common distance operations also so buffering is one of the extended neighborhood operations in raster and of course again if you go through that you know they say you know visibility analysis sitting on the tower or putting a signal how far the signal will go you have an elevation data every pixel as an elevation it is a digital elevation model and you are starting on this standing on the top of the tower how much area is visible it goes on all the eight directions and if there is a hindrance beyond it it is not visible so even view shed mapping is an extended neighborhood operation so one can do buffering visibility etc using that then comes the zonal operation zonal is you know a lot of operations are possible I just brought only one view graph to you zonal means you know what it does see I said neighborhood focal and all done in a single layer when I said local operation number of layers can be many when you come to zonal it operates on two layers one is a zone another is a team you keep in mind one is a zone another is a team I can take an example of zone map maybe a thaisal boundary map of a district I have a map well say 10 thaisals are there each thaisal level of one one one two two three three thaisal must have raster map a thaisal boundary map and next is let us say rainfall map it is a thematic map this is a zone map when I run a zonal operation what it does for every zone or for every cluster it takes this value finds minimum maximum average whatever you want to calculate and it stores in a tick so that moment you give a thematic map on an administrative boundary map your table directly gives all the things about this theme in this zone okay so automatically the zonal operation does it so whether it is a infall you put or a elevation you put or whatever the attribute you put the thematic map it gets quantified it is stored in a table for every zone you can pull anything and make it a map or use it for any computation okay that is what zonal operation does zonal operations compute a new value for each location of the existing values from a specific layer that are associated with the look not only with the location itself but with all the locations that occur within its own so it's not a local a zone this always a theme it works okay so I have a zone map a thaisal boundary map is shown that I think elevation data is shown so that moment this zonal map works with the elevation map for each zone it calculates the maximum value minimum value average etc stores in a table for this theme whatever the theme it stores and then it puts on a table you can pull it and make a map for any attribute okay so that is the concept of a zonal operation can we process the vector as well as with the rascal you see normally if you take officially speaking topo map we are not supposed to digitize that is one thing you have to keep in mind no ananna forces authorized to digitize a toposheet except survey of india okay let us assume not toposheet any thematic map okay see normally whenever you use a thematic map the best way to do it is you vectorize it don't use it directly as a remote sensing data scan the data vectorize it after you cleaning building and all you rasterize to the resolution unit that is a way you should use whether you are taking land use or whether you are using contour or whether you are using road network you must digitize first in a vector mode and then bring it to the in case you are integrating remote sensing data you come to the raster domain in case you are using for a network analysis just keep it with the vector itself and use it okay yes okay please now you know you look at in india andamanic over how many islands are there thousands of islands are there okay are they connected no they are all as far as jaw mantry is concerned they are all disjoint polygons agree but are they government of india will they give us the data for every small polygon no there is only one headquarter port player if you want to store the non-spatial data you have to store only as a single record fine so what that is the reason where they try to handle what is called as a multi polygon a multi polygon you have to see hierarchically upper than a polygon where a multi polygon is treated as a region a region is consist of multi polygon you give a tree that what id is of those polygons come to a region region becomes one for that region you store the record okay that is the way either a multi polygon is treated even a multi segments are treated say i may be working on a national highway and that high i am working in gujarat and karnataka may not be in marashra the highway may be passing somebody in gujarat somebody in karnataka after treated so there comes what multi arcs or multi segments so you know these are all got evolved by the people when they were doing applications see now like i'll show you afternoon vector models and vector analysis everything you know when people go to an application that time you find okay our GIS doesn't handle it we have to introduce it in fact multi polygon concept was introduced by ESRI when they were handling hawai maps because hawai is part of us when they had to handle naturally they couldn't handle the data because there is again a single capital so this concept was introduced which now as i said latest OGC specification they have taken care of very well not even that's what i said they have introduced something called multi geometry where points are possible lines are possible polygons are possible so you know you build it you give the link and then treat it as a special object it is different region means you it is a separate category of objects you see you cannot mix it point you store it you say it is the polygon as a feature okay what is the geometry it will ask you to treat it polygon when you are saying region you have to have a separate link hierarchically which polygons belong to them moment you define the region that option is available absolutely it may be joined it may be disjoint moment you use the word multi polygon it is a collection of polygons may be connected need not be connected that is the way it is generated