Welcome back. You’re now watching lecture
number 62 in a long series. This is the last lecture video you’ll need to watch. Last time
I talked a little bit about GPS and land surveying. You learned about GPS in earlier videos. I
went over the basics of what’s going on in land surveying. And I talked a little bit
about GPS enabled total station. And that’s a surveying instrument that allows you to
use traditional surveying technology in terms of laser range finding and things like that.
And when you can get access to satellite signals incorporate the satellite signals in your
surveys. So that’s where that’s all at. For the rest of this video I thought it would
make sense to go over the modules that we’ve covered over the course of this course. And
just hit the high points of each one of the modules. So module 1 way back at the start
of the course, I talked about systems, and that being a collection of components that
work together to accomplish a task. And a geographic information system has those information
system components, hardware, software, people, processes and data. And the geographic aspect
is an information system that answers questions that associate data with location on the earth.
OK. Then I talked about models. Models, remember, are simplifications, or generalizations of
some aspect of reality and in the GIS world the two basic models that we use are the vector
model and the raster model. Module two was about georeferencing, and if you remember
way back then, I talked about things like ellipsoids which are the smooth, mathematical
representations of the earth’s surface, and the geoid which is, you can think of it as
a bumpy gravity model representation of mean sea level. And so together with an ellipsoid
and an geoid, you’re going to be able to georeference locations on the earth that will give you
x, y and z locations on the earth. And once you get those 3-D locations, if you’re going
to put it on a 2-D map, you need to do projecting. And projecting is basically going from 3-D
to 2-D. And when you project, you’re going to distort some features of reality. And so
projection is about choosing which method to use to minimize the distortion in your
map that’s most important to you. And remember, there are like four different types of distortion
I talked about: shape, size, distance and angle. So depending upon your projection you
will minimize distortion in one or more of those aspects. Module 3 is about cartography.
The high level thought, thinking about cartography is that you’re working with a communication
model where you have a message that you’re trying to send to the person that’s going
to read your map. And in order to send the message in a way that will be understandable
to your map reader, you need to use a cartographic language. And there are standard components
of the cartographic language that you can put in your map that the reader will be able
to understand and then get a clear message from you about what you are trying to say.
So part of that cartographic language involves symbology. How do you symbolize things? And
how you symbolize things ends up depending upon the type of data that you’re trying to
symbolize. And remember, I talked about four different types of data: nominal, ordinal,
interval and ratio. And the first two, nominal and ordinal, those are qualitative data, so
you can’t put numbers to nominal and ordinal data. Interval and ratio are quantitative
numbers, and so you can do some math with those. So depending upon which type of data
you have, you’ll use different symbols in your map to send your message. You can also
take your data and you can group it into categories. And that’s called classification. There are
a bunch of methods you can use to classify your data, to again, not confuse your map
reader with too much detail, but provide the essence of the message you’re trying to send.
There are a bunch of other map elements that you need to include in your map part of the
cartographic language. For example, you might need a scale bar, you might need a north arrow,
you might need information about who made the map or where or what type of projection
the map is in. So those are all important pieces of cartography. Module 4 was about
data quality standards and metadata. And remember, metadata is the who, what, when, where and
why of your data. Data standards are rules by which people create their data sets so
that when someone shares data with you, that’s been created according to a certain standard,
you’re going to be able to look at that standard and understand how good the data are. The
way you determine how good data are has to do with the errors associated in the data.
And remember that depending on the type of data you going to have different ways of looking
at the error. With numeric data you’ll look at things like root mean squared error. And
with the qualitative data you’re going to probably use an error matrix to give you an
idea of what the error might be. There are two specific errors that happen in GIS a lot.
One of those is called the ecological fallacy. And that’s the fallacy in which when people
look at your map and your data have been aggregated in one way or another. They tend to perceive
the single number that you report for the entire aggregation unit as applying throughout
the aggregation unit. So you can imagine if we had a map of world countries and it had
average income for each country. And you look at the average income number for the United
States, the ecological fallacy would be something that would tend to have map readers think
‘well everyone in the United States has that average income”. When in reality there are
lot of people that make more and less than the average. So that’s the fallacy, the ecological
fallacy. The other one is called the modifiable areal unit problem. And that is where when
you aggregate the data the result of your aggregation is going to depend upon the size
and the shape of your aggregation unit. So the larger the aggregation unit that you use
to aggregate your data, the less variation you’ll have in your data. So that washes out
the extremes of your data. And again, imagine using average income for the United States,
that’s going to give you the number that will be closer to kind of a smoothed average versus
having averaged income reported by individual states or by counties. You’ll have a lot more
variation in those smaller aggregation units. Aggregation by shape is gerrymandering. And
so if you choose a certain shape you will be able to accentuate certain aspects of your
data and send a message that accentuates the pieces of the data you would like to send.
Module five, big long module about the vector module. Remember the vector model says the
world is going to be represented by points, lines and polygons. Points are at the base
of that. Lines are points that are connected to each other. And polygons are lines that
begin and end at the exact same point. You can connect spatial information, the where
of those points, to other information about the data. And depending upon the type of relationship
between the spatial data and the non-spatial data you’ll use a different method. Sometimes
you’ll use what’s called a join to connect the data, sometimes you’ll use what’s called
a relate. You can ask questions of your data manually by just selecting subsets of your
data either in a portion of your map or representation of certain topological aspects, say all of
the points inside of a certain area. So you can select by attributes which are the ‘whats’
of the data, or you can select by location. And those things are called queries. So what
queries give you are a subset of your original data. The other thing you can do involves
what’s called geoprocessing, and part of geoprocessing is map overlay. And what that does is allows
you take different layers, or different themes, or different attributes, of your dataset and
overlay on top of each other. And depending upon the operation that you use, you will
create a new geometry that’s an output resulting from the operation of that geoprocessing tool.
So if you think of the operation that’s called intercept and you intersect two layers upon
each other, the output layer will be a new layer, with new geometry that represents the
common area between those two layers. There are other geoprocessing tools that you can
use frequently, like buffer, which allows you to create polygons at a distance from
certain types of data. Or clip, which lets you clip out or pull a piece out of your data.
Topology becomes important when you’re doing some of these geoprocessing operations and
it will be important when we talk about networks in the next module. But, topology is spatial
relationships between different types of data. So there are three aspects of topology that
we think about. There’s connectedness, whether things are connected to each other. There
is adjacencies, so you can imagine is something to my left, something to my right. Is it exactly
sharing a border with me? And there’s enclosure, whether things are inside of a certain polygon
or outside of a certain polygon. Module six was about the network model. In the network
model, what you’re doing is you’re combining different data into a model that allows you
to determine a path from one point on your map to another point on your map. And in the
network model points become what’s called intersections, lines become what’s called
edges. OK. The network model is going to use topology because you need to know if roads
are connected to each other, you need to know if intersections of two roads, which would
be the points, are located at the intersections of the two lines. Things like that. Different
kinds of networks, the two basic categories we talk about are the transportation type
networks, those are generally two-way networks as in car going both ways on a street, and
utility networks. Utility networks generally orient themselves in one direction, and usually
that direction is from high energy to low energy. Associated with let’s say a stream
that’s going to flow from high elevation to low elevation, or maybe an electrical grid
that’s going to go from higher voltage to lower voltage. And then the last thing I talked
about in the vector model was geocoding. And geocoding is using address information, textual
information, to create a location on the earth. To create a point on the map associated with
an address. Module seven was about the raster model. And the raster model is going to represent
or simplify the real world into a grid made up of cells, of uniformly sized cells. And
where the vector model was best for modeling discrete entities, the raster model usually
is best for modeling continuous entities. You can convert from vector to raster and
raster to vector, but one or the other model will generally be preferred, depending on
what you’re trying to look at in the real world. So this grid is made up of cells. The
cells are the same size and the way you locate something on the map is how many cells away
you are from one of the corners of the map. So you might be four rows down and three columns
across. If you know how big the rows and the columns are, the cell size, then you know
where that cell is located. Each cell is going to have a single value that applies throughout
the area enclosed by that cell. If the cell has an integer value, you’re going to be able
to connect the information in your raster model to non-spatial information, just like
you could with the vector model. But if your raster model has decimal data, you’re not
going to be able to connect it. So all you’re going to be able to get out of your raster
model is a number. You can image that might be temperature or precipitation or elevation
or something like that. But if your raster model has integer number associated with,
let’s say, different types of land cover, you can connect the number in your raster
model to attributes about the land cover. Well, this is forest, this is cropland, this
is urban, things like that. The way that do the equivalent of overlaying in the vector
model is to use what’s called map algebra in the raster model. And the raster model
allows you to compare, to combine raster datasets using map algebra to get an output that represents
some combination of your input datasets. In the vector model, there are like three different
types of geoprocessing operations you could use. There’s local, there’s neighborhood,
and there’s global. In the raster model, there’s also local, neighborhood and global operations.
But there’s a fourth type of operation, which is zonal. So you should be aware of what those
different operations mean. Sometimes when you are combining rasters you’ll have different
rasters with different cell sizes or different orientations. In order to line these raster
datasets to get a common output dataset, you do what’s called resampling. So you should
know what that’s about. And then raster models are used for terrain analysis, because terrain
is generally modeled as elevations or slopes or aspects or curvature. And so you’re going
to be able to use a raster model to look at the terrain. And you can also use a vector
model to look at the terrain, and that would be called a TIN. A triangulated irregular
network. You can do this equivalent of buffering in the raster model and that would be what’s
called distance. And you can weight the distance in the raster model. A weighted buffer might
be something that you could use to simulate the equivalent of a raster network. That was
another big module. Module five and module seven were pretty big modules. Module eight,
I talked about interpolation and spatial statistics. There were four kind of basic methods of interpolation
that you can use. Interpolation again, is estimating values at unknown points, based
upon know values at known locations. So there’s four basic ways to do that. And one is called
trend, another is called inverse distance weighted. Inverse distance weighted relies
upon Tobler’s first law of geography, which says everything is related to everything else,
but things that are close to each other are more related than things that are further
apart. So Tobler’s first law has a lot of impact when you’re doing interpolation. You
give heavier eights to things that are closer to each other than you do to things that are
further apart. And that’s what inverse distance weighted interpolation is about. There’s also
what’s called a spline method of interpolating. And that’s basically the equivalent of putting
a rubber sheet over your data values and having the rubber sheet smoothly contour between
the values that are known. And the fourth one is called natural neighbors. And what
that does is it makes polygons and within a certain polygon the value that you get will
be closer to the value at the center of that polygon than at the center of any of the other
polygons. Then I talked about statistics. Traditional statistics, and remember there’s
two different types of statistics when you’re thinking of descriptive statistics, statistics
that describe the central tendency in your data. So for traditional statistics that would
be the mean, the median or the mode. And then there’s descriptive statistics that give you
an idea of dispersion in your data. So for traditional statistics, you might think of
the variance or the standard deviation. Or maybe the skewness or kurtosis. Spatial statistics
take into account not only the values of your data, which is what traditional statistics
consider, but also the locations of your data. And when you take into account some of the
statistical behavior or your data, you’re able to do a fifth type of interpolation,
which is called kriging. Kriging is going to be the best method of interpolation if
your data are able to meet a whole series of assumptions that are behind the statistical
assumptions in the kriging process. In addition to the descriptive statistics that I’ve talked
about, the central tendency and the dispersion that you can get from spatial statistics,
in spatial statistics you can also get an idea of clusters. Where there might be a lot
of points or not very many points. And where there might be high values and low values.
And so what you do in those types of statistics, which are inferential statistics, is you assume
that your data are just randomly spaced or randomly valued, and then you do a statistical
test. And if you statistical test is far enough away from what would be a random distribution,
you can conclude that there’s something going on behind your data. There’s something that’s
making that data be not just random data. And once you have an idea that there’s something
in your data that may be making it to be not random then you’re able to ask a question
of ‘why is it that my data aren’t random”. And one of the things you can do to explore
why your data aren’t random, you can do what’s called a regression. So there are global types
of regression, which do the regression over the entire map and then there are local regressions,
which do the regression in smaller areas of your map. So that your regressions can vary
across the surface of the map. Module nine, I didn’t do, it was about geo-databases and
data management. I’m not sure what you covered in that module, but I’m sure it was a god
and exciting module. And then the last module, module ten was about remote sensing and GPS.
Remember, remote sensing gets categorized under active and passive remote sensing. Remote
sensing that’s active generates its own signal and senses what reflects from the object you’re
studying as its data input. Passive remote sensing senses what is reflected from that
object generally and usually by the energy from the sun. When you look at remote sensing
there are a bunch of different types of resolution you need to consider. There’s the regular
spatial resolution that you might consider in other types of data, but there’s also spectral
resolution which is which piece of the spectrum your remote sensor is going to be sensitive
to. There’s temporal resolution, which is how frequently that data can be collected
because it is generally from satellites that don’t hover over the same surface spot of
the earth. And there’s radiographic, or radiometric, which is how many pieces your dataset can
broken into. So, does it go from 1 to 100 in units of one, or 1 to 100 in units of then?
The active remote sensing methods are LIDAR, which uses lasers for light, RADAR used radio
waves, and SONAR uses sonic waves. And so those generate their own signals and still
sense what is reflected from their signal. GPS is about satellites that are orbiting
the earth sending out signals that have time information. And your GPS receiver can receive
those signals and knowing what time the GPS receiver gets the signal and when it’s sent,
it is able to determine how far away that satellite is. Once you have a signal from
multiple satellites you can triangulate your location and find a specific location on the
earth. There are bunch of things that can go wrong with your remote sensing and your
GPS signals, and those might be errors that you might need to correct. And so with remote
sensing a lot of those errors that might be in your data might be associated with things
that happen to the signal as it is coming through the atmosphere. Similar things can
happen to the GPS signal as it is coming through the atmosphere. There are geometric corrections
that you need to make to remote sensed data because things aren’t straight and orthogonal
in remote sensing. They’re at off angles and in GPS, there are other types of errors associated
with things like the clocks on the satellites and the clocks in your GPS. The ephemeris,
which is knowing exactly where that satellite is, so you can tell how far away it is. And
by doing orthometric corrections to your remotely sensed data you’re able to get better and
more accurate data. And by doing what’s called differential corrections to your GPS data,
you’re able to get more accurate data. OK, so that was as quick as I could make it fly
through the very high level aspects of all nine modules that I’ve covered in this course.
And each one of those was made up from lectures, 61 lectures, up to now that went into some
more detail about those things I just talked about. And if your going to go our and use
this information daily or frequently in your life or in your career, you will end up learning
much more about all the things that I’ve covered in the 61 videos. But if you’re only going
to use it infrequently or maybe not at all, what I hope you go away from this course with
is an awareness of spatial importance of data and an awareness of the vocabulary and the
basic concepts associated with geographic information systems. And at the very least,
if there is only one thing you remember from this course, it’s Robischon’s first law of
geography, which is ‘if it makes a difference where, geography matters’. And we know that
if geography matters, you’ll be able to use GIS concepts to actually flesh out and send
a message to someone to explain why geography matters in this case. So I want to thank you
for having the patience to watch all these videos. I know it’s really frustrating at
times to sit and watch videos, but I hope you are able to learn something and I hope
you’ll be able to take something away from having to watch the videos. Thanks again and
good luck down the road.