1 of 1

Introduction

Research Goal

What is Geo Analytics?

  • It is the ability to visualize spatial and geographic information to gain critical insights and make better business decisions.

  • It involves combining location information with various other data sources such as demographic, geo-economic, and behavioral data.

ArcGIS: Analysis

Study Area

Summary and Conclusions

Acknowledgments

I would like to thank Mr. Menhennet (Dept. Head of Geography, Waterloo Collegiate Institute), my geography teacher for his guidance and support.

References

  1. Thomas, A.W. (2004). Meeting Challenges with Geologic Maps, American Geological Institute.

  • Huang, H (2007). Geomorphologic investigations on karst terrain : a GIS-assisted case study on the island of Barbados,  Dept. of Bioresource Engineering, McGill University

  • Matthew, D.C., & Milewski, A.M. (2018). Sinkhole formation mechanisms and geostatistical-based prediction analysis in a mantled karst terrain, CATENA, Volume 165, Pages 333-344

Using Geo-Analytics to Predict Sinkhole Occurrences

Simha Kalimipalli, Waterloo Collegiate Institute

Research Sample

  • Florida is chosen as a sample study area as 81% of land surface is karst
  • Florida’s sinkhole data is publicly available on United States Geological Survey (USGS) website
  • Contains information about 3919 sinkholes from year 1948 to 2018.
  • Some of the attributes are latitude, longitude, county, dimensions, shape and comments.
  • Sinkhole volume and category were calculated using raw data
  • Meta cause and Meta effect was deduced using semantics.

Can we use Geo-Analytics to predict where potential sinkholes are formed?

  • The overall goal of this study is to investigate the development and distribution of sinkholes using ArcGIS online by
  • To determine
    • Geological features that can affect sinkhole formation
    • Human activities that influence sinkhole formation

What are Sinkholes?

  • Sinkholes are large depressions or holes in the ground caused by some form of collapse of the surface layer.
  • Sinkholes can be shallow or deep, small or large, but all are a result of the dissolving of the underlying limestone.
  • Sinkholes are most common in “karst terrain.” or a region where bedrock can be dissolved by groundwater.
  • 20% of earth’s surface has Karst Terrain

How do they form?

  • When rainwater seeps into the bedrock. It dissolves the bedrock underneath and forms a depression

  • Human activities, such as groundwater extraction, quarrying, and land development alter the surface drainage and lower the water table which can further increase the frequency of sinkhole formation.

Sinkholes are Dangerous!

  • Loss of Human Lives
  • Damage to residential, commercial, industrial, and agricultural infrastructure
  • Destruction of highways and railway tracks
  • Contamination of Groundwater

Walt Disney World Resort, Florida

August 12, 2013

Florida Sinkhole (FSH)

Layer

Population

Density (PD)

layer

Karst

Topography (KT) layer

Intermediate Aquifer

System (IAS) layer

Road (RD) layer

Nearest Neighbor (NN) layer

Although, sinkhole formations are widespread across the globe, they are one of the most under-evaluated geologic hazards!

Sinkhole Data

Correlation FSH & IAS

  • Most sinkholes occur in areas where with Green and Red areas

Acquire Sinkhole data

Perform data cleaning /

manipulation

Upload data to ArcGIS

Design Layers

Create Sinkhole Polygon maps and Point maps

Apply geometric and proximity filtering

Conduct Validation

Perform

Geo-Analysis

Methodology

List of variables that drive sinkhole formation

    • High Population Density as evident in urban areas

Population Density

(PD)

    • Whether Bedrock formation is made of water-soluble evaporate rocks (salt or gypsum) or carbonate rocks ( limestone or dolomite).

Karst Topography (KT)

    • Presence of Intermediate Aquifer system

Intermediate Aquifer System (IAS)

    • Proximity to roads and highways

Road (RD)

    • Proximity to previous sinkholes

Nearest neighbor (NN)

Correlation Analysis

FSH Layer

Future Research

  • We intend to use more powerful software for analysis.
  • We also plan to use the variables from our study to estimate a robust statistical model that can best predict the sink hole occurrence.
  • We also plan to extend the study to other data sets

Limitations

  • ArcGIS online is limited in analysis compared to ArcGIS desktop.
  • Availability of other data sources such as aerial imagery for further analysis
  • Interpreting textual data
  • Sinkholes represent one of the most widespread and under-evaluated geological hazards.
  • Study used GIS-based sinkhole mapping methodology to analyze primary factors that influence sinkhole formation.
  • Areas with the following characteristics have a higher sinkhole formation probability within the study area: a) population density of 500 people per mile, b) underlying rock is long karst, c) intermediate aquifer less than 200 feet and d) history of sinkholes within 2km. About 55% of sinkholes in the study areas have the above mentioned characteristics.
  • Some of the strategies to use to minimize damages are: maintain the drainage of retention ponds (especially during storms), maintain the water pipes, minimize groundwater pumping and quarrying in areas where karst is exposed, and find alternate ways to frost-freeze method for protect crops when temperatures drop.
  • Policy makers, city planners and engineers can use the findings of this study analysis to develop strategies for minimizing damages.

ArcGIS Layers

Areas

IAS depth

% of sinkholes

Green

0 to 1 feet

53%

Red

1 to 101 feet

33%

Yellow

101 to 201 feet

11%

Rest

More than 201 feet

3%

Correlation FSH & PD

  • Most sinkholes occur in urban areas where PD is greater than 500 people per mile.
  • 79% of sinkholes occur in urban areas

Correlation FSH & KT

  • Most sinkholes (92%) occur in areas where karst has fissures, tubes and caves over 300 meters long and with 15m to 75m vertical extent
  • More prone where karst is exposed

Correlation FSH & RD

  • Most sinkholes occur in areas close to roads

road buffer

% of sinkholes

within 500m

73%

within 250m

53%

within 100m

36%

within 50m

28%

Correlation NN

  • Most sinkholes occur in areas where there is history of sinkholes
  • Using distance analysis a buffer was constructed.

nearest neighbor

% of sinkholes

within 4.9kms

95%

within 1.3kms

75%

within 0.53kms

50%

within 0.18kms

25%

First

    • We examine how each variable correlates with sinkhole occurrence

Next

    • We examine how different combinations of variables together imply the sinkhole occurrence

Water component (WC)

    • Both steps use ArcGIS analytics and spatial data sources from USGS, FGS, and FDEP

Findings from multiple layers: FSH, PD, KT, IAS & NN

  • Most sinkholes (55%) occur where

Layer

Characteristics

PD

500 people per mile

KT

rock is karst

IAS

less than 200 feet

NN

history of sinkhole within 2km

Findings from multiple layers: FSH, PD, KT, IAS, NN & RD

  • 42% of Sinkholes occur where

Layer

Characteristics

PD

500 people per mile

KT

rock is karst

IAS

less than 200 feet

NN

history of sinkhole within 2km

RD

proximity to road within 500m

Florida’s Insurance Claims half-billion per year (approx.)