Lab work 3 Geography Information System PDF

Preview

Summary

GLS 613Lab Practical 3 : Geostatical AnalysisStudent Name : NUR AZREEN BINTI MOHD ZAINI Student ID : 2017643126 Lecturer Name : DR. Nabilah binti Naharudin Group : AP2205A Date : 25/10/No. Lab Title Mark Mark each section 1 Introduction 52 Objective/s and study area 53 Procedures/ Results 154 Questi...

Description

GLS 613 Lab Practical 3 : Geostatical Analysis

Student Name

: NUR AZREEN BINTI MOHD ZAINI

Student ID

: 2017643126

Lecturer Name

: DR. Nabilah binti Naharudin

Group

: AP2205A

Date

: 25/10/2019

No. 1

Lab Title Introduction

Mark

Mark each section 5

2

Objective/s and study area

5

3

Procedures/ Results

15

4

Questions answer

25

Total mark

50

0

Table of Content

Tittle

Page

Introduction

2

Objectives

3

Procedure

4-8

Conclusion

9

1

INTRODUCTION By using polygon centroids from measuring test points placed in a point function layer or raster layer, you can easily generate a continuous surface or graph with ArcMap. Measures such as elevation, water table depth and levels of pollution can also be used in this reference. Geostatical Analyst in collaboration with ArcMap provides a large range of surface creation tools for visualizing, observing and interpreting spatial phenomena. The Geostatistical Analyst utilizes sampling points in a field at various locations and produces a consistent layer (interpolates). The sample points are measurements of some event, such as nuclear power plant radiation leakage, oil spill, and heights of altitude. The Geostatistical Analyst generates a surface to estimate values for each position in the landscape using the values from the calculated positions. The Geostatistical Analyst offers two classes of strategies for interpolation which is deterministic and geostatistic. To construct the surface, both approaches are based on the similarity of neighboring sample points. Deterministic approaches use interpolation of computational functions. Geostatistics is focused on both numerical and computational approaches that can be used to construct surfaces and quantify predictive uncertainty. In addition to providing various interpolation methods, the Geostatistical Analyst also offers a range of supporting tools Such resources allow you to properly explore and understand the data in order to build the best surfaces based on the information available.

2

OBJECTIVES 

To accessing Geostatistical Analyst and the process of creating a surface of ozone concentration using default parameter values to introduce you to the steps involved in creating an interpolation model.



To learn the process of exploring your data before you create the surface to spot outliers and recognize trends.



To creates a second surface that incorporates more of the spatial relationships discovered in exercise 2 and improves on the surface you created in exercise 1. This exercise also introduces you to some of the basic concepts of geostatistics.



To compare the results of the two surfaces that you created in exercises 1 and 3 and how to decide which surface provides the better predictions of the unknown values.



To learn the process of mapping the probability that ozone exceeded a critical threshold, creating a third surface.

3

PROCEDURE Exercise 1: Creating a surface using default parameters The purpose of this exercise is to demonstrate you the process of making sample data surfaces. This practice will bring you to the extension of the Geostatistical Analyst. You can use default parameter values to create a model in order to create an ozone concentration layer. Launch the Geostatical Analys. Attach the toolbar for Geostatistical Analyst. Use the Geostatical Wizard to build a surface. In this exercise, the Geostatistical Wizard and the method of developing an interpolation model have been implemented. The experiments will develop this approach by extracting the most relevant information from the data in order to create a better model.

Exercise 2: Exploring your data You're starting to explore the data in this exercise. Geostatistical analyst provides a variety of methods for information exploration. In this exercise, you will analyze your data in three ways, analyzing your data's distribution, finding, patterns in your data, and considering spatial autocorrelation and directional influences. a) Examine the distribution of your data using the Histogram tool

─ The methods of interpolation used to generate a surface give the best results when the data is normally distributed. You may choose to change the data to make it standard if your data is skewed The Histogram software plots frequency histograms for the data set attributes, allowing you to analyze the univariate distribution in each data set attribute. The ozone data is unimodal, but as seen in the histogram it might not be very similar to a normal distribution.

b) Create a normal QQ plot

─ A quantile-quantile plot (QQ) is often used to compare the data distribution to a standard normal distribution, offering a further measure of data normality. The closer the points are to the graph's straight line(45-degree), the nearer the sample data is to a

4

normal distribution. The generic QQ plot also reveals that the information is not normally distributed since the plot points are not a straight line.

c) Identify global trends in your data

─ If your data has a pattern, it is a non-random surface component that can be represented by a mathematical formula. You can add local variation to the surface by modeling the trend using one of these smooth functions, extracting it from the data and continuing the analysis by modeling the residuals, which is what remains after the trend is eliminated. You can evaluate the short-range variance in the surface when modeling the residuals. The Trend Analysis tool helps you to recognize the presence / absence of patterns in the input dataset and determine the polynomial order is best suited to the trend. Using the Trend Analysis tool you have seen that the data show a trend and that when refined, a second order polynomial would be the better pattern.

d) Explore spatial autocorrelation and directional influences

The cloud of semi-viariograms / covariances showed that the unusually high semi-variogram values are largely represented by the coastal lines. This analysis tool shows that anisotropy should be taken into account in the interpolation model. The surface of the semivariogram shows spatial autocorrelation in the results. You can begin surface interpolation with confidence, realizing that the data set contains no outer sample points. The surface you generate in exercise 1 can be more reliable using default options and parameter values since you now know that patterns and anisotropies occur in the data and can be optimized for this in the interpolation. The prediction model can also be improved by a data transformation.

5

Exercise 3: Mapping ozone concentration During this exercise, the Ozone concentration map generated in exercise 1 is enhanced and some basic geostatistical concepts are included. You will once more use the ordinary method of interpolation, but this time you will use the trend and anisotropy to create better predictions in your model. The ordinary kriging model is the simplest geostatistical since it has the lowest number of assumptions. The global trend in the data set is mapped default by Geostatistical Analyst. The area shows the quickest change in the direction from Southwest to Northwest and the slower change in direction from Northeast to Southeast. The increase in air quality in the southwest and northeast can be related to the ozone development from mountain to coast. The elevation and the prevailing direction of the wind contribute to relatively low mountain and coastal values. The high human rate also causes high rates of mountain-coast emissions. Therefore, these patterns can properly be eliminated. I.

Semivariogram/Covariance modelling

The aim of the semi-variance / covariance modeling is to establish what is best for a model that passes through the semi-variogram points. This is a graphical representation that provide a picture of the data set's spatial correlation. You can fit models into spatial relations in the dataset by using the semi-function / covariance modeling dialog box. The geostatistical analyst decides appropriate lag sizes for the semi-diagram groupings first. Lag size is the size of the distance category in which pairs of positions are combined to minimize the huge number of possible combinations. The empirical semivariogram values of red points shall be grouped according to the separation range with which they are correlated to give a clearer picture of semivariogram values. The points are split into bins, and the size of the lag defines how large each interval is. This method is referred to as binning. The parameters for the omnidirectional stable semivariogram model are nugget, range partial sill and shape.

6

II.

Directional semivariograms

The semivariogram points and the fit model will be affected by a directional influence. In some cases, items similar to each other may be more equivalent than in other directions. In a semi-functional system, Geostatistical Analyst may find spatial and anisotropic effects. Wind, runoff, geological structure or a wide range of additional processes may contribute to anisotropy. In making your map, the directional effect can be quantified objectively and accounted for. With the Search Directorate device, you may analyze the disparity in data points for a certain path. It allows you to check the spatial results on the semi-diagram. The output surface is not impacted. III.

Searching neighbourhood

You may specify the number of points, the radius and the number of circle sectors to be used for forecasting by using the Searching Neighborhood dialog box. The points chosen in the information view window display the weights associated with each calculated value to estimate a crosshair value for the position IV.

Cross-validation

Cross-validation is aimed to help you make informed decisions on which model makes the most specific predictions. It makes you aware of the expectation for the uncertain values of the model. Manually, cross-validation omits a point in the dataset, calculates a value for the location value of that point using the rest of the information, then contrasts the measured and forecast values The forecast error calculated statistics act as diagnostics showing whether the template is appropriate for making decisions and generating charts.

7

Exercise 4: Comparing models Exercise 4 compares the two models in order to find out which one better predicts unknown values. You can compare the predictions in two or more mapped surfaces by using the geostatistical analyst. It lets you determine which method offers more detailed ozone concentration estimates based on statistical cross-validation. In this exercise, you will compare the Trend Removed layer created in Exercise 3 with the Default Kriging layer you created in Exercise 1.

Exercise 5: Mapping the probability of ozone exceeding a critical threshold In exercise 5, you will use indicator kriging to measure the likelihood of crossing a critical ozone threshold and produce a final map displaying all the surfaces you have generated in this tutorial. In decision making it is necessary that using a map of the projected ozone to identify unsafe areas, as the uncertainty of the predictions must be understood. You can use Geostatistical Analysis to chart the risk of ozone values crossing the threshold to support the decision-making process. Although Geostatistical Analyst has a number of techniques that can accomplish this function, the simplest system available, indicator kriging, will be used for this exercise. This approach does not allow a specific distribution to match to the dataset.

8

Conclusion In this project we have used the Geostatistical Wizard, the Histogram Tool, Normal QQ Map, Semivariogram / Covariance Cloud, and Ordinary Kriging for the ozone value prediction, and a probability map indicator showing that the ozone concentration is above the critical threshold value. In the Geostatistical Analyst, there would be any other form of interpolation which could be used to predict spatial data.

9...