Bridge collapse (Peru) & stream bank erosion

Last days a video of a bridge collapse in Lima (Peru) became viral. The video shows the moment when the scour of a bridge gets scoured and the bridge collapses. Discussions about the collapse of this bridge became a trending topic in several social networks, especially in Peru. Some discussions tried to find someone or something to blame and others focussed on whether the bridge was properly designed or not. In this post we will not discuss such topics. Whether there is someone to blame or not; whether it was properly designed or not, the fact is that the video a good example of stream bed erosion. Due to space limitation, this post will be an introduction to stream bank erosion. In a future post we may simulate in detail the erosion process of the video and analyse whether a properly designed rip rap would have prevented the erosion (and collapse) or not.
Fig 1. Bridge collapse due to stream bank erosion in Peru (Source youtube)

Rivers are dynamic systems that change over time. One of the processes defining such change is stream bank erosion. All rivers have stream bank erosion. Even the so-called stable rivers have eroding banks. Certain events such a flooding usually trigger stream bank erosion

Stream bank erosion becomes even more complicated when the steam has bends inducing secondary flows. Even small bends create a vertical velocities profile and a helicoidally flow known as centre region cell. Besides this helicoidally flow, the outer bank of the river also experiences a so called outer bank cell flow. This outer bank cell flow is a small helicoidally flow with a direction that opposes the main helicoidally flow. Although the outer bank cell is small and weak, it plays an important role in stream bank erosion.
Fig 2. Secondary flows (Source: Blanckaert and Vriend 2004)

The full solution of the stream bank erosion is quite complicated. Thus, simplified approaches were developed based on the main mechanisms controlling stream bank erosion. The mechanisms controlling stream bank erosion may be divided in two groups: Scour and Mass failure
  • Mass failure is the process when large chunks of bank material collapse into the river. Sometimes mass failure is a consequence of local scour removal.
  • Scour failure is the direct removal of bank material

There are two main modelling approaches for analysing bank erosion due to the mentioned processes:
  • Bank failure. Bank failure is a geotechnical based model that evaluates the bank stability and its critical failure plane. Once the soil gets saturated due to the flood, several theoretical failure planes are analysed considering its resisting forces and the driving forces, in order to define the one with the lowest safety factor. If the critical safety factor is lower than one, then all the soil above the plane will be eroded by mass failure.
Fig 3. Mass failure (Source: Carey B 2014)
  • Toe scour. Toe scour considers the bank material removed by the flow. The shear stress between the flow and the bank toe is evaluated considering the critical share stress and the erodibility of the bank. If the shear stress is higher than the critical shear stress, a portion of the bank will be eroded. After the toe gets eroded, a second failure mechanism may occur. Due to the toe erosion, the soil will look like a cantilever. Depending on the size of toe scour and the volume of soil above the scour whole, a cantilever shear failure may occur.
Fig 4. Toe scour (Source: Carey B 2014)

If we get more information about the physical characteristics of the bridge and the river, in a further post we may simulate the scour process that collapsed this bridge, and analyze in more detail this case.

Automating HEC-RAS tool

Some years ago, (early 2010) I found a way to break the RAS code and to automate some features of HEC-RAS. Thus, I was able to perform sequential modelling HEC-HMSàHEC-RASàSobek1D2D and to perform uncertainty analysis within cloud (Moya Quiroga et al., 2013). At that time there was no literature about the RAS code, so I had to try several options in the Controller class and find out how they work. In October 2015, “breaking the RAS code” was published (Goodwell, C. 2015) and I hurried into getting my copy (I was among the first ones, so it included a 4 colour pen).

Fig 1. HEC-RAS in sequential modelling (Source: Moya Quiroga et al., 2013)

Documentation in the book gave me important tips and I was able to develop several scripts for automating many pre-processing and post-processing tasks; not only updating parameters but also modifying geometry and post-processing results. HEC-RAS used to be a great tool, but now it became wonderful. However, all my scripts work under the console (no graphical user interface GUI). I developed them and I know how to use them. Thus, I decided to include a user interface in order to make more universal tools. Few years ago I released the first version with a GUI (AHYDRA), a tool for automating hydraulic analysis. Debugging and coding takes time, so I released and didn’t improve it. Few months ago I decided to update it and to release new improved version was released. This version includes new features and some bugs were fixed.

New automatization options
This version automatically updates data and simulates HEC-RAS considering:
·   Downstream boundary conditions (energy slope). Important for analysing the extent of the uncertainties due to BC and to select the best BC for our case.
·   Manning roughness. Important to analyse the range of potential manning roughness. Besides, it allows analysing a spatial variable Manning. For instance, the Manning range is between 0.03-0.04; we can simulate one cross section with 0.03, the next cross section with 0.034 and so on. Thus, we will be able to perform an uncertainty analysis of manning.
·     Upstream discharge boundary condition. Inflow is usually assumed as a fixed value. However, most inflow discharge data are based on steady state stage-discharge curves prone to errors. Thus, we can perform a sensitivity or uncertainty analysis of the inflow.

Fig 2. User interface of AHYDRA

Single and multiple simulations (Monte Carlo).
  •  The single simulations option (at the left of the GUI) allows to automate one simulation by changing one value of the mentioned possibilities.
  • Besides, this version includes the possibility to perform a Mont Carlo analysis by running several simulations and randomly updating the desired parameter within a user specified range. This option is located on the right side of the GUI.

Rip rap design

The single simulation option also calculates the rip rap size (d50 in SI units) for each cross section, by considering the simulated hydraulic conditions and the USGS rip rap design criterion.

This version includes an installer *.exe. Just double click it and follow the install wizard. After installation AHYDRA will be accessible via desktop and start up menu.

Scan the QR code with your tablet/smartphone to fill the free AHYDRA request form.

Bridge monitoring with satellite data SAR

Civil infrastructure is vital for supporting our life style and economy. However, the prolongued expossure to outdoor weather conditions and time deteriorates the infrastructure. Besides, after some years of decades, the infrastructure do not comply with new construction codes. This problem of infrastructure deterioration is becoming more important because of the increasing number of infrastructure collapses. Thus, it is important to perform a constant monitoring civil infrastructure.Remote sensing synthetic aperture radar (SAR) data provides a new and modern methodology for monitoring civil infrastructure.
How to monitor infrastructure with satellite data SAR?
The reader may be wondering "How satellite data can used to monitor civil infrastructure such as bridges?". The answer is Interferometric SAR (InSAR) and Differential InSAR.
InSAR takes advantage of the interferometric phase of simultaneous observations. Differential InSAR (SInSAR)  considers the interferometric phase of observations acquired at different times. Soussa and Bastos (2013) explain the concept of DInSAR with Figure 1. Two SAR observations at times T1 and T2 will show the deformation DR.
Figure 1. DInSAR concept (Source: Soussa abd Bastos 2013)

Several studies are beginning to take advantage of DInSAR for monitoring bridges. Soussa and Bastos (2013) analyzed ERS SAR observation of the Hintze Ribeiro centennial bridge in Portugal. The bridge collapsed in 2001 when one of its piles collapsed. The study showed that between 1995 and the collapse date the collapsed pile had vertical displacements up to -20 mm yr-1, where the negative sign means that it was sinking. Other example. In 2012, Cusson et al., (2012) made a similar study at larger scale in Vancouver and Montreal, Canada. More important than the number of examples however, is the fact that SAR data proved to be an important data source for monitoring civil infrastructure and finding hidden information in collapsed infrastructure.
Figure 2. Collapsed Hintze Ribeiro centennial bridge (Cource: StockClip)
Satellite monitoring or in situ monitoring?
Civil infrastructure can also be monitored by in situ inspections. Thus, we may wonder what are the advantages of SAR monitoring?; Is it better to use SAR monitoring or in situ inspection? Actually, there is no definite answer to the last question; it depends on the specific case. Nevertheless, Cusson et al., (2012) points some of the main advantages of SAR monitoring. Among the advantages of SAR monitoring we can mention:
  • Increased and constant monitoring frequency regardless weather or time
  • Possibility to monitor the whole structure
  • Monitoring regardless accessibility issues.
In further posts we will provide more details and more application cases of SAR data.In you want to know more about DInSAR and SAR application, feel free to contact us.

Oroville dam disaster observed by satellite images

Last days we had several news about the Oroville dam. In this post I will not describe the details of this natural disaster because there are already several sites about that that. Instead, I prefer to provide a visualization of this disaster so that we can understand its magnitude and visualize it. For such purpose I used remote sensing satellite data collected before the disaster and during the disaster. In all the images you can get a full screen image by clicking the image.

The figure shows 2 images observed by the French satellite Spot. The left images was observed on November 11, 2016 (before the disaster), while the right image was observed on February 14, 2017 (during the disaster). The water changes are evident. Before the event we easily visualize the shoreline. Besides, we can also visualize 2 details during the event (right image). A) At the dam we can see a white spot which the water over the spillway, and B) the water in the reach between the dam and Oroville city has different colours because of the mud and earth; you can get a full screen image by clicking the image
Fig 1. Spot image of Oroville dam before the event (left) and during the event (right) Click the image for full screen image

Thanks to Digital Globe, ESA and Gizmodo, we can visualize in more detail what really happened in the auxiliary spillway. The left image shows the spillway before the event, while the right image shows the spillway during the event. We can observe the spillway at full capacity and the auxiliary spillway already collapsed.

Fig 2. Image of Oroville spillway before the event (left) and during the event (right) (Source: Digital Globe, Gizmodo). Click the image for full screen image

 The following image shows in more detail the collapse of the auxiliary spillway and the road.
Fig 3. Image of Oroville spillway before the event (left) and during the event (right) (Source: Digital Globe, Gizmodo). Click the image for full screen image

A spanish version of this post was published at iAgua

10 countries that deforest the most

A forest is an ecosystem; a community of plants and animals interacting with one another and with the physical environment. We depend on forests for our survival, from the air we breathe to the wood we use. Forests provide habitats for animals and livelihoods for humans. Forests also offer watershed protection, prevent soil erosion and mitigate climate change. However, every year thousands of forest hectares are lost due to agriculture, lumber and fires.

Which country deforest the most?
A few days ago an infographic listing the countries that deforest the most was published. However, the infographic focuses on raw numbers and does not consider that fact that the Brazilian forest is bigger than any of the other mentioned countries. Here we rearrange the list considering the forest area of each country.

You can visualize the interactive geo-chart of the results (Click here).

10. Congo DR
In 1990 Congo DR had 1 603 630 km2 of forest. By 2015 its forest area was reduced to 1 525 780 km2. It lost 4.8 % of its forest.

9. Colombia
In 1990 Colombia had 644 170 km2 of forest. By 2015 its forest area was reduced to 585 017 km2. It lost 9.2 % of its forest.

8. Brazil
In 1990 Brazil had 5 467 050 km2 of forest. By 2015 its forest area was reduced to 4 935 380 km2. It lost 9.7 % of its forest.

7. Bolivia
In 1990 Bolivia had 627 950 km2 of forest. By 2015 its forest area was reduced to 547 640 km2. It lost 12.7 % of its forest.

6. Tanzania
In 1990 Tanzania had 559 200 km2 of forest. By 2015 its forest area was reduced to 460 600 km2. It lost 17.6 % of its forest.

5. Argentina
In 1990 Argentina had 347 930 km2 of forest. By 2015 its forest area was reduced to 271 120 km2. It lost 22.1 % of its forest.

4. Indonesia
In 1990 Indonesia had 1 185 450 km2 of forest. By 2015 its forest area was reduced to 910 100 km2. It lost 23.2 % of its forest.

3. Birmania
In 1990 Birmania had 392 180 km2 of forest. By 2015 its forest area was reduced to 290 410 km2. It lost 25.9 % of its forest.

2. Zimbawe
In 1990 Zimbawe had 221 640 km2 of forest. By 2015 its forest area was reduced to 140 620 km2. It lost 36.5 % of its forest.

1. Nigeria
In 1990 Nigeria had 172 340 km2 of forest. That is 12.5 % of its territory. By 2015 he forest area was reduced to 69 930 km2. It lost 59.4 % of its forest.