An Automated Workflow to Find Areas at Risk of Flooding in a Cloudburst

Executive Summary

The rainstorm and flood disaster this year (in 2018) is the worst in the last six years after the 2012 floods that killed 363 people, displaced 2.1 million people and affected seven million people in 30 of the 36 states of the Nigeria country, according to the National Emergency Management Agency, NEMA. The economic losses in 2012 were put at N2.5 trillion. Those floods are attributed to the combination of many events, and one of the most important are very heavy local rainfall.

Standard climate models predict a rainstorm of such intensity once in a thousand years. In 2013, heavy rains and floods, which started in mid-July affected more than 81,500 people across Nigeria. Almost 8,000 people were displaced and more than 6,500 homes were damaged. As of September 11, 2013, 19 deaths had been recorded and 2,217 farmlands were destroyed. In 2015, heavy rains, compounded by the breakdown of dams in some states, caused floods in 11 states across Nigeria in August and September. According to the National Orientation Agency, NOA, 53 people died, 100,420 people were displaced and a lot of buildings and houses were swept away. In 2016, the Director, Disaster Risk Reduction Department of NEMA, said no fewer than 100 people died from flooding. According to the UN, 9,000 houses were completely destroyed. The number of livestock lost was about 26,000. At least, 38 people became victims of flood in Nigeria and more than 92, 000 were without shelters. And in 2017, NEMA said that 27 states experienced devastating flood disasters, in which no fewer than 90 people died. It seems that climate change has upset standard predictive models. Cloudbursts in Nigeria should probably now be expected with greater frequency.

This project seeks to create "blue spot" maps showing flood-prone areas during sudden rainstorms and map building in the Lekki Local Government Area (LGA) according to their flooding risk... by running a geoprocessing model using elevation data and building footprints. I expect that due to the low resolution of the Raster Layer (DEM) we will use, our result will be very coarse but in the model we inserted some parameters to correct it (our DEM) having a 30-m resolution (X,Y) and a 5m accuracy in elevation (Z). However, I expect all high risk buildings to be close to a water body.

The basic method includes:

I will attend to map blue spot, show flood-prone areas during sudden rainstorms and then map building and symbolize them according to their flood exposure using several a customized ArcGIS Modelbuilder. Broadly, I will

1.      Download and clean OSM data representing building footprint of our targeted area

2.      Download and correct DEM based on the Lekki Local Government Area (LGA) boundary

3.      A customized model builder that takes as input the DEM and building footprints

At the end of this project, a few questions are expected to be answered:

1.      Where are flood-prone areas during sudden rainstorms

2.      What is the exposure level of the LGA?

3. How many building can be considered as "High Risk" to flood? and Where are them?

Methods

For this analysis, DEM from ALOS (Advanced Land Observing Satellite) were retrieved. It is a Global DEM of 30m of resolution and a global accuracy of 5m in Z (elevation). We downloaded our data the 12/1/2018 for our Area of Interest. We also included building footprints of all the Lekki Local Government Area (LGA) from a private company.

Then, we automated the process by developing a model builder as shown below

Our model builder can be downloaded, since all wards and tools are not well visible.

Basically from our DEM we extract bluespots, we compute their area and length and we do an intersection with cleaned building footprint... and we used the area of intersecting bluespot-building to evaluate in percentage the degree of exposure of each building.

Results

Based on our result and on our classification method, we found that:

-         Very High Risk = 69.32%

-         High Risk = 25.99%

-         Moderate Risk = 3.67%

-         Risk = 0.78 %

-         Very Low Risk = 0.24 %

However, we see that the risk is concentrated in Urban Areas, where many residential developments were constructed in wetlands or small lakes and the drainage network is very poor. That is why homeowners living on such abandoned land face the challenge of frequent floods.

Discussion:

A number of factors limit the precision and certainty of the model results.

-         The models don't consider a building's elevation (its z-value) within a bluespot.

-         The models don't consider if buildings have elevated foundations or designs that raise their base height above the terrain.

-         The models don't remove interior bluespots; that is, bluespots that are completely surrounded by structures. In reality, water can't flow into these bluespots from their surrounding catchments, and they would therefore not fill up at the rate calculated.

Link to download the Model Builder

Sources (To know More) :

[1] : ArcGIS Learn

[2] : Geographic Information System specialization on Coursera

[3] : Geospatial analysis project

[4] : RAINS OF FURY: Nigeria loses 141 lives to rainstorm, flood in 2018

[5] : The problem looks inevitable in 2018 as Victoria Island, Lekki and Ajah might be submerged again

A propos de moi :

Je suis le Fondateur de geoAdvantage, une entreprise de développement de solutions logicielles combinant l’intelligence artificielle aux systèmes d’information géographique et dont le but est de générer de la valeur commerciale en prédisant la demande du marché, par l’extraction et l’analyse de données.

Suivez moi sur LinkedIn