A significant stage in the development of this thesis is the way to obtain and manage the information required for our results to conform to reality. It is also relevant to use the sources and resources available for the analysis, and it is for this reason that the primary source of information that is the GIS Geographic Information Systems (GIS) is briefly introduced.
Geographic information systems (GIS) are, by definition, a structured database, which allows us to obtain, use, and analyze spatially referenced information. According to the type of information, qualitative data and quantitative data about the different aspects that are present on the earth's surface can be obtained, which allows us to digitally represent very relevant aspects for computational analysis such as streets, buildings, urbanizations, vegetation, bridges, and others.
37 GIS has been the leading providers of information for the study, management and analysis of hydrology processes or related to the circulation and transport of water on the earth's surface, and mainly they have been used in various aspects of flooding, as it allows us to connect between hydrodynamic modeling and the social component (Moody, 2010), where the results of hydrodynamic modeling over the areas studied or of interest are superimposed (Youssef, 2011) assures us that GIS tools provide an excellent platform to manipulate, combine and analyze the information for the determination of areas of potential floods very quickly and more efficiently.
GIS has had an evolution based on three pillars, the first of which is traditional cartography and technical procedures for the analysis of maps, which bases the need to obtain the information of the earth's surface digitally. The second is the identification, possession, and manipulation of geographic data, which allows us to identify urban areas, rural areas where the presence and absence of people in the area are determined. The third is the emergence of computer applications associated with computer-aided design, which finally allows us to achieve that compatibility of hydrological modeling with hydrodynamic models. Long before advances in computing capabilities permitted the use of GIS, cartographers had already developed sophisticated notions for spatial analysis and geographic information representation (Jones, 1997). Among their legacies we found methods to implement spatial functions such as overlapping, intersection and proximity calculations, it is indisputable that the computer processes have allowed an advance in the execution, the execution time of the various tasks. Information from a GIS, where we define a section of reality, can be represented in two conceptual models: 1. As a continuous medium defined by means of a set of quantitative variables or defined according to the surface. In this way, a section of territory can be described by the superposition of different surfaces that are considered of greater relevance. 2. As the definition according to defined limits with similar characteristics, where a value of them can only occupy space.
In the first case, the reality is defined as a set of surfaces that have a variable associated with each point in space (temperature, humidity, vegetation). If we consider the fact as in the second case, where objects are considered entities, they can be defined according to the dimension they occupy point type objects, linear objects, and polygonal objects.
Choosing a type of entity for the model depends on the scale of its use; these can be streets, buildings, bridges. Being the space our analysis variable, that is, the processes that occur
River Flood Modelling under Limited Data Acquisition using PWRI Hydrologic Model
38 at the level of the spatial scale are our object of study, it is very convenient how GIS stores information, so that the layers that are used are used. We need for the analysis, as well as the objects and entities that we are going to consider. In this way, we simplify reality to the variables that are relevant to the analysis. Real-world processes that are represented by geographic information systems allow the treatment of images obtained from satellites provided that such information contains the various layers where this type of data from the multiple sensors is stored. GIS has a large number of functions aimed at spatial analysis, including overlapping maps, proximity analysis, calculation of areas, perimeters, and volumes, route analysis, statistics, and algebraic maps (Burrough et al. ., 1998, Bernhardsen, 1999). The GIS fulfills the function of integrating the information.
However, more fundamental is the capacity in the spatial processing where the parameters that have to be included in the hydrological modeling are automatically extracted from the Digital Elevation Models (MDE), which we will present later. The geographical location aspect is solved with the geographic information systems since the information has systems that internally reference the data, automatically linking the thematic component (such as alphanumeric data), with the spatiality (which is the geographical location), and the spatial properties of objects. These maps constitute the foundational bases of GIS (Morad, 2001).
The GIS allows us to complement the information available with various sources of information in cases where the study focuses on an area where data is not available, making functions essential for entering data into the system. The GIS allows us to overcome this limitation with features of digitization, data verification, rasterization, and georeferencing (Clarke, 1997).
A factor that provides an advantage to GIS is the frequency with which the data that compose it is updated and entered since the origin of this information is from diverse sources. GIS also has the functions of data transport and data conversion between the same GIS programs and some features compatible with computer-aided drawing (CAD) models. Another advantage of using GIS is the integration of pos processes carried out with GIS information, which allows us to link the results of the developed model and return to the GIS environment.
After collecting the spatial information, for the storage process, there are two methodologies where the reality of the earth's geometry is structured to be used in a computational plane. This is because, for some authors, the conception of real space is
39 different between the two models that are: vector models and models in bit images (better known as raster images), where the first is considered continuous models and the second a model discreet.
Each data storage model has its advantages and disadvantages: The raster database is characterized by being very simple, and the calculations on them quite simple; however, vectors have complex structures that require sophisticated algorithms for analysis (Burrough et al., 1998). However, vector data can be stored compactly and displayed with high precision, unlike what happens in the raster format.
The vector models were the first to develop; they present the information according to graphic primitives that are: the point, line segments, and circumferences. To build digital models using vectors, they are usually constructed by collecting points in a given area;
by joining vectors, we get the representation of reality digitally. Raster models are characterized by the construction of a matrix that can be square or rectangular where the values of the matrix are assigned according to a discretization of space and averaging the value of the property in that section of the space we call a cell, the This model's resolution is based on the size they give the cell to assign it the value. This type of information is usually presented in images, and the form obtained is through satellites. There are two types of raster models: Continuous (where there are variables that are present in a distributed form in the matrix) and discrete (where the limitation of the phenomenon is established within a specific region or area and wherein some cells they may be absent the variable, which can be quantitative or qualitative).
2.4.1 Advantages of Geographic Information Systems in Flood analysis
Below we present each of the advantages offered by information management through GIS with satellite image input data.
• Digitality: Most of the images from satellites are in digital format; this means that in many applications, there is no need to convert data, scan, or scan. The images are useful with a minimum of preparation that can be loaded into GIS. Due to digital information, satellite images are manipulated, enhanced, and processed to extract data and subtle details that other sources would not detect.
River Flood Modelling under Limited Data Acquisition using PWRI Hydrologic Model
40
• Speed: If we consider that satellites are in stable orbits, the maximum time to take pictures of the desired area is one week.
• Low cost: For large areas, satellite images are much cheaper than other techniques such as aerial photography, topographic bells.
• Globality: They are not limited by political limits, natural limits such as mountains or oceans, satellites can obtain an image of any surface of the earth.
• Updated: due to various phenomena and the variability of today's world, the conditions of the earth's surface are continually changing, satellite information allows us to update satellite information in a couple of days, which is much faster than when They use maps that take months or even years.
• Compact: The information coming from satellites presents information on the ground cover, infrastructure, roads, in an image.
• Accurate: If we consider that with this technique, the intervention of the human being is minimal during its processing since the information is acquired by means of a sensor.
• It is because of these aspects that we use GIS to store, use and organize information related to issues of our model such as the shape of the earth's surface, disturbing elements such as land use, vegetation, permeability and infiltration of soil, soil type, as well as the application of hydrological models. Our model is characterized by a model in a continuous medium, that is to say, that the information must be described by a constant variation; this requirement is fulfilled by the geographic information systems used; however, some aspects of a discrete type that require consideration are also presented.
2.4.2 Digital Elevation Models
An essential part of the analysis to predict Floods is the spatial component of the problem, where we have to define the form of the computational domain to be used in the model.
The computational domain represents the shape of the earth's surface where rainfall fluid interacts with the earth's surface. The most recommended way to define the domain is through the Digital Elevation Models (DEM) that we use through the Geographic Information Systems (GIS). GIS plays a fundamental role in this regard, not only in the integration of information but in the operational capacity in spatial processing that allows
41 the automatic extraction of the parameters to be included in hydrological modeling from the MOUs (Pusineri, 2004). Digital Elevation Models can be generated in various ways, and these are: through photogrammetry, interferometry, laser topography and classical topography, and other techniques. The models for obtaining spatial information, for representing the shape of the earth's surface, are models that allow us to consider the continuity of the relief information. These are functions that will enable us to obtain the value of z as a function of its spatial coordinates z = f (x, y).
The MOUs and the attributes we obtain from them, such as slope, drainage network, drainage area, curvature, and topographic index, are important parameters for any process where we need to perform a terrain analysis (Mukherjee, 2012).
Converting the real data of the land relief into digital information for analysis means moving from a complex spatial reality to a synthetic digital representation that can be used by current computational models for hydrodynamic modeling. The models that present the geographic information in the digital form start from the two-dimensional digital models that, when introduced from a third dimension, produce what we call the Digital Terrain Models (DTM).
Digital elevation models are used as a surface representation method. A DEM is a digital quantitative model of the topographic surface. Topographic data is crucial in flood modeling and is the best information to use due to the high accuracy of the information (Sanders, 2007).
In the case of raster models, the variable represented is the height of the terrain, each of the areas in which the image is divided is known as a pixel.
2.4.3 Raster data resolution
The spatial and temporal scale of the input data, as well as their availability, constitute a mathematical aspect in flood modeling and, therefore, in the Floods. Usually, the available topographic data have a low resolution for hydrodynamic modeling, being the most precise but more expensive photogrammetric techniques, (Burgos et al., 2012), We currently find platforms that have information for free, easily accessible, globally available and with High spatial resolution.
River Flood Modelling under Limited Data Acquisition using PWRI Hydrologic Model
42 The resolution of the input data of the model acquires value because the applications of the images increase more and more with the launch of new satellites that are added to those already in orbit. Increasing the number of satellites has a more significant amount of scene sizes, step frequencies, spatial details, and spectral resolutions. Also, with the increase in the types of space sensors, the images are more useful, which makes it easier for the user to choose the most suitable ones for their purpose.
2.4.4 Obtaining raster data
The data used as input for the construction of our computational domain is given by raster-type information from a satellite image to obtain a global MDE model provided by Advanced Spatial Borne Thermal Emssition and Reflection Radiometer of the Ministry of Economy, Commerce, and Industry of Japan, with National Aeronautics and Space Administration (NASA).
This digital topographic map provides the information obtained by the radiometric instrument ASTER, which is located on the NASA TERRA satellite that is 705 kilometers away, this information is the largest and most accurate in the world, covering 99% of the surface of the planet (Saccro, 2009). The main objective of the availability of this information is to be able to improve the understanding of the processes at the local and regional level that occur, and that happens on or near the earth's surface and in the interior atmosphere, including; the interaction that exists between the atmosphere and the surface.
ASTER images correspond to the visible spectrum and wavelengths of infrared thermal radiation, with spatial resolutions ranging from 15 to 90 meters.
ASTER is composed of 3 VNIR, SWIR and TIR subsystems, each of them has specific characteristics such as VNIR has three (3) bands in the near-infrared spectral and region, with a spatial resolution of 15 meters; SWIR has six (6) bands in the spectral region of the shortwave infrared, with a resolution of thirty (30) meters and TIR has a resolution of five (5) bands in the thermal infrared with a spatial resolution of 90 meters. The error in this model is the deviation in height with respect to the height in the real surface; for obtain the true value of the data, other sources are sought that have a higher degree of accuracy. The data from the ASTER GDEM are derived from a map with more exact values, field values, GCP Ground Control Points, or elevations obtained by photogrammetric means.
43 In the spatial case of MDT extraction, the system requires that before obtaining the MED, the stereoscope pair must be transformed into a couple of standardized epipolar images, which are parallel to the XY plane of the object space. ASTER stereoscopic data has been used in different topographic conditions of the terrain, so its importance in the application of them in our modeling. The main uses of this data are:
• For smooth and rough topography
• To reproduce the MED of a scene (Cheng et al., 2003).
• The difference of MEDs for a volume of changes (Vignon, 2003)
• To produce an ortho-rectified image (Marangoiz et al., 2005)
• Simultaneous stereoscopic pair information is acquired from ASTER, with a radiometric advantage (Toutin., 2002)
• Ability to reduce the radiometric differences of the stereo pair, since the time elapsed between the acquisition of the two images is usually a few seconds, minimizing atmospheric and optical effects.
2.4.5 Hydrological Model (rain-runoff)
One of the main advantages of GIS in modeling is that they accelerate modeling with the development and implementation of hydrological models. This input data, the data generated during the process, and the output data will maintain georeferencing and topological relationships.
Hydrological applications have been categorized according to (Maidmente, 1993) as follows:
• Evaluation and hydrological inventories.
• Determination of hydrological parameters.
• Construction of simple hydrological models
• Development of integrated hydrological models.
The most used tool for natural risk assessment and location studies are GIS hydrological applications, these two applications have similar things, but each one has a different treatment.
River Flood Modelling under Limited Data Acquisition using PWRI Hydrologic Model
44 For those of us who use GIS for modeling, we use the calculation procedures that already exist, and that were developed outside of GIS, which makes GIS considered a tool for geographic information management.
Geographic information systems are used to determine the main drainage networks, where the fluid when contacting the earth's surface will take a particular route, the hydrological model used is the model of eight (8) neighboring cells. Where according to the raster information contained in a satellite image, the model is applied. The propagation of the rising wave towards downstream is considered. Model considerations assume that the entire surface can be adequately represented by rectangular cells, where we can adequately characterize the hydrological processes involved in flash flooding - This discretization in rectangular cells has an adequate size to make this representation conform to morphometric characteristics of the basin. The drainage network corresponds to a topology obtained from the connection structure of the cells
According to Giraldo et al., 2005, after processing the MDE and assigning addresses to the drainage network according to model D8, a very satisfactory approach to the actual structure of the drainage network is taken. It can also be defined that the flow follows the direction of the local potential energy gradient and that this gradient is equal to the slope of the terrain (Ramírez, 2002). The definition of the drainage network is made according to the direction of the flow, and the computational algorithm analyzes the number of cells that pour into the same cell; hence it is determined that it is a shed surface. The process by which the preceding is identified is mainly two: the simple flow models and the multi-flow models. In the first one, it is considered a single direction where the fluid can be directed, while in the second one, it is found that it can go in multiple directions. The deterministic model of eight (8) solutions is the most commonly used method, where the flow can be directed to the eight principal directions that limit each cell and to the direction of the maximum negative slope (Martz et al., 1992) found that the D8 addressing model reproduces well the drainage network in large tributary area basins and well-formed streams, even if the divergent flow in convex areas is not considered. The drainage network is a vector data that we obtain to determine the directions to which the flow is directed, and we delineate the flow lines and sub-basins from the information provided by the MDE models.
• There are three (3) situations that the D8 model does not consider when defining the drainage network: Cell i is surrounded by eight cells taller than it.
45
• Cell i is part of cells with the same height. These areas are considered flat or flat areas. If this flat area is surrounded by higher cells, this is called a sump.
• One of the cells adjacent to i, with a slope equal to the maximum. This problem is known as indeterminacy and also occurs when cell i is at an edge or boundary of the MDE. The D8 model assigns each pixel or each cell of the maximum slope down between the eight addresses considered, determined by the eight adjacent cells.