Landsat 7 for reconstructing the dissapearance of Poopo lake

Climate change and water resources stress

In December 2015 new published several articles about the disappearance of the Poopo lake, which used to be the second biggest Bolivian lake. There is still uncertainty about the causes that dried the Poopo lake out.

Local residents claim that mining activities and erroneous water policies are the causes that dried the lake out. Local government prefers the easy solution and blames climate change, despite the fact that climate change is a process that takes several years. This article will not get into the detail of discussing why the lake dried out, but it will apply Landsat 7 data for reconstructing the last 17 years of the lake. 

Landsat 7 mission was launched on April 1999, as the seventh satellite of the Landsat programme. It collects and stores 532 multi-spectral and panchromatic images of the Earth per day. It has 7 multi-spectral bands and one panchromatic band (Image 1).
Landsat satellite NASA spectrum
Image 1. Landsat 7 bands
Source (What when how)

The normalized difference of the vegetation index (NDVI) provides a measure of the vegetation amount and condition of vegetation. The pigment in plant leaves, chlorophyll, strongly absorbs visible light (from 0.4 to 0.7 µm) for use in photosynthesis. The cell structure of the leaves, on the other hand, strongly reflects near-infrared light (from 0.7 to 1.1 µm). The more leaves a plant has, the more these wavelengths of light are affected, respectively. Thus, NDVI is estimated by a non-linear transformation of the visible (VIS) and near-infrared (NIR) bands of satellite information.

NDVI = (NIR - VIS) / (NIR + VIS)

If there is much more reflected radiation in near-infrared wavelengths than in visible wavelengths (High NDVI), then the vegetation in that pixel is likely to be dense and may contain some type of forest. On the other hand, if there is very little difference in the intensity of visible and near-infrared wavelengths reflected (Low NDVI), then the vegetation is probably sparse and may consist of grassland, tundra, or desert (Image 2). Water bodies are the opposite of vegetation because they reflect more visible energy than infrared energy; hence, water yields zero and negative NDVI values.
vegetation health satellite image
Image 2. NDVI concept

For this post we used Landsat 7 images from the Poopo lake area collected within the months of July - September (clear skies) between the year 2000 and the year 2016. NDVI was estimated for each image; thus, it was possible to estimate yearly extension of the Poopo lake. The video of the link  and the GIF animation show an animation of the last 17 years of the Poopo lake. After 2013 the lake experienced an accelerated retreat that ended in its disappearance.

The disappearance of Poopo Lake had serious socio - economic - environmental consequences. Local people had to migrate to other location, and the ecosystem of the region was destroyed. There are several doubts whether it will be possible to recover the ecosystem or not. The damage is done and it is urgent to look for recovery / adaptation policies. Unfortunately, bolivian government focuses on blaming others instead of acting.
Lago Poopo desaparece
Animation 1. Poopo lake drying out 2000 - 2016 (based on Landsat 7 data)


Desastre ambiental en Bolivia

Desastre ambiental Bolivia

Most expensive hydropower dams

hydropower dam spillway cost

Some time ago I read an article about the Zungero hydropower dam (Nigeria). The article pointed that Nigeria people was concerned because the total cost of the 700 MW project was 1.3 billion $US; that is 1.87 million $US per MW; a price they considered to be excessive (the title suggested the most expensive in the world).

I compared the Zungero dam price with the price of the hydropower dam project Chepete (Bolivia) which is also a very controversial project, not only because of its cost but also because of its potential disastrous consequences and several observations about its feasibility. According to the executive summary, Chepete will have a cost of 1.92 billion $US per MW; more expensive than Zungeru.

However, it would not be fair to consider Chepete as the most expensive hydropower because of one single comparison. The following table shows a comparison of some of the most controversial mega hydropower dams. Once again, Chepete is the most expensive hydropower project in world.

Hydropower Country Power [MW] Cost [B $US] Power cost [M$US / KW] Note
3 Gorges China 22 500 37 1.64 Source
Itaipu Brazil/Paraguay 14 000 19.6 1.40 Source
Belo Monte Brazil 11 233 14.4 1.28 Source
Xiluodo China 12 600 6.2 0.49 Source
Grand Ethiopia Renaissance Ethiopia 6 000 4.7 0.78 Source
Baihetan China 12 000 13.5 1.13 Source
Grand Inga Congo DR 40 000 80 2.00 Source
Chepete Bolivia 3 300 6.34 1.92 Source

There is a possibility that the Grand Inga Dam (Congo D.R.) may be more expensive (there is still not much data about this Mega project that would be the biggest hydropower dam in the world, with an installed power 12 times higher than Chepete).

It is important to note that the cost of Chepete considered only the construction cost reported by the respective studies. There are additional costs that will certainly increase the total cost of Chepete, for instance:
  • Chepete will require a power transmission line longer than 1 000 km across the Amazonia.
  • Chepete is located on the Beni river, a river with a very high sediment load; Beni river could be considered as a river among the 10 rivers with highest sediment load. Thus, it may have high maintenance costs that will increase the total cost.
  • Big hydropower dams usually have high overcosts between 26% - 90% of the initial budget.
  • Chepete will affect a very fragile ecosystem and the environmental costs may be expected to be high with several additional negative consequences,
Although this data is not enough for defining the single most expensive hydropower dam, we can assume with high confidence that Chepete would among the most expensive hydropower dams (if built).

Bridge deck drainage is more important than we think: The Cau Cau bridge in Chile

Puente Cau Cau en Chile

The $30-million Cau Cau bridge was supposed to be the first drawbridge in Chile. It is a 90 m long bridge with two decks, each one 45 m long, two traffic lines and one bicycle line. However, in January 2014 when the bridge was finished and ready to be inaugurated, builders realized one small mistake; one traffic deck had been installed upside down.

They realized this error because the bicycle line of one deck ended abruptly and connected directly with the traffic line of the other deck, as shown in the figure 1.

errores  de ingenieria

Figure 1. Bicycle line ends abruptly (Source: Cau cau bridge Youtube video)

The Cau Cau bridge problem

Initially, one may think that this is a minor mistake and could be easily corrected by painting new lines. Actually, by looking at Google Earth images of the bridge, it looks as if they already painted new traffic lines and the bridge looks fine (Figure 2). However, the drainage design of the deck is the main problem and it cannot be solved so easily.

Thus, authorities announced that the bridge would be demolished and a new one would be built. But what is the problem with the deck drainage?

Error en el puente Cau Cau

Figure 2. Aerial view of Cau Cau bridge (Source: from Google Earth)

The Importance of bridge deck drainage

Deck drainage is usually considered a trivial task when building a highway or a bridge. However, it is a very important one and has its own requirements and challenges. In another post we will present an introduction to bridge deck drainage. 

The Cau Cau bridge deck was designed and built with a given longitudinal slope, so that water will flow in one direction (from the middle towards the edge). Such longitudinal slope creates a vertical difference between the two edges of the deck. Thus, there will be a sudden drop at both edges of the deck (Figure 3); besides, the water that was supposed to flow in one direction will flow on the other direction; thus flowing towards a no-outlet edge. 

I do not have the exact details of the designed slope but table 1 shows the vertical difference resulting from some possible longitudinal slopes ranging between 1 per thousand and 1 per cent, assuming a 45 m span. 

If you need  advice on deck drainage or if you have further questions, fell free to contact us.
Table 1. Vertical difference for different longitudinal slopes considering a 45 m span.
Slope [%] Vertical difference [cm]
0.1 4.5
0.2 9.0
0.5 22.5
1.0 45.0
bridge deck drainage
Figure 3. Sudden drop due to longitudinal slope for drainage
The following video from Discovery channel (from Incredible Engineering Blunders) describes this case. This is another example of bridge failure because drainage was underestimated.

2D flood model preferences

hydraulic modelling 2D

During the last years we witnessed an important increment in the availability of 2D flood models.
As a result, there were many discussions about "which one is the best 2D flood model?".

Personally, I consider that all models that satisfy some benchmarking tests (to verify their accuracy by comparing with some known cases) are good and will provide good results when properly used (adequate selection of cell size, coefficients, boundary conditions). I consider that the main differences between different 2D models are:
  • Computing cost. Some models demand more computing time and use bigger files.
  • Graphical user interface. This is the main difference in any computer program. Software developed by different developers have different GUI. Some are more friendly and some not.
  • User experience. Maybe this one is the most important one. When a user begins to use one model, he becomes familiarized with its interface and its performance. Thus, the users develops an intuitive knowledge about this model and becomes faster at using it.
This post is a poll about the preferences when selecting a 2D flood model. 10 of the most popular 2D flood models are presented for you to selecte the model that you prefer. The models considered in this survey were selected considering:
  • Their performance was verified considering at least 7 benchmarking tests of the UK Joint Defra Environment Agency (Reference 1, Reference 2).
The models considered in this survey are (they were listed in alphabetical order, no preference):
  • ANUGA: An Open Source Hydrodynamic / Hydraulic Modelling. Most of its components are written in the object-oriented programming language Python. Thus, it is possible to use it via Python scripts.
  • Delft3D: Developed and maintained by Deltares. Although the full potential of this model is for simulating 3D flows, it is also possible to perform 2D simulations by considering one vertical depth averaged layer.
  • Flo-2D: Developed and maintained by Flo-2D software. This model is based on finite elements.
  • Flood modeller pro: One of the most popular models in UK, and during the last years it got a fast growing number of users thanks to the release of a free version. It is the latest update of ISIS.
  • HEC-RAS: Developed and maintained by the USACE. This is one of the most popular hydraulic models. For many years it was limited to 1D flows. However, the HEC-RAS version 5 released in 2015 also includes 2D capabilities.
  • InfoWorks ICM: A very popular model from UK. This model does not only solves the 1D-2D overland flow, but it provides a complete catchment modelling including hydrology and underground drainage sewer systems .
  • ISIS2D: A very popular model in the UK. However, in was replaced by Flood Modeller Pro, which is the latest update of ISIS, and integrates bith the 1D and the 2D.
  • LISFLOOD: Developed and maintained by the University of Bristol. Unlike the other models, this model does not solve hydrodynamic differential equations, but it is based on cellular automata. Thus, it does not require much computing power and the computations are very fast.
  • MIKE21 - MIKE Flood: Developed and maintained by DHI. This is one of the 2D flood models with more cases studies in the world.
  • Sobek 1D2D: Developed and maintained by Deltares. This model was designed specifically for rivers. It was one of the first models to include the hybrid 1D2D capabilities; the main channel simulated as 1D, while the floodplains are simulated as 2D.
  • TELEMAC: Developed and maintained by the TELEMAC-MASCARET consortium. This finite element based model was initially developed by SOGREAH, but in 2010 it was released as a free model by the mentioned consortium
  • TUFLOW: Developed and maintained by BMT WBM. I think that this was the first model that included the nested grid capabilities.
  • XP solutions: It provides a set of models for water resources, drainage and flood hazard. It includes broader scope of tools for including the effects of civil infrastructure.

You can view the current results (percentage of preference) in the following link. Although I have my preference, I will not participate in this poll, as I do not considered it to be fair (I organize the poll). If there is another model that you would like to include in the poll, please let me know (the model needs to verify the benchmarking tests).

Scour protection at Akashi Kaikyo bridge

longest bridge Japan World
Some years ago, when I was in Kobe I had the opportunity to visit the Akashi Kaikyo or Akashi bridge, the bridge with the longest single span in the world (2 km). While I was looking at the documented experiences of the construction, I realized that the scour at the piers and the innovation in scour protection were among the most important and most challenging tasks.

The heavy load of the superstructure (the volume of concrete cast in each foundation exceeds 200,000 m3, totally amounting to more than 1.4 million m3 as a whole) and the harsh physical conditions of the Strait such as swift tidal currents and its heavy maritime traffic in which more than 1,400 ships navigate a day, made this project a challenge for the civil engineers.

The Akashi Kaikyo Bridge as shown in Figure. 1 is a suspension bridge with a center span of 1,991 m and an overall length of 3,911 m. Foundation work conducted were for 2 anchorages at both ends of the strait and 2 tower foundations on both sides of the navigation lane. Tower foundation located in deep-water zone is a cylindrical shape foundation 80 m in diameter and 70 m high.

After excavating the seabed from 45 m to 60 m below the sea level with a grab bucket dredger, a steel caisson fabricated on a dry dock was installed. Then, scour protection composed of filter units and rip rap was executed in a range of 240 m in diameter around the caisson, and desegregating underwater concrete of 270,000 m3 was deposited in the caisson interior.
suspended bridge puente colgante
Figure 1. Profile of the Akashi Kaiko bridge (Source: Kashima et al., 1998)
Scour protection
The towers are located in an area of strong tidal currents where water velocity exceeds the 7 knots (about 3.6 m/s). When the caisson was installed, accelerated currents (horse shoe eddies) were generated around the caisson and scour began to advance.

Although the peripheral of the caisson was overlaid with rip raps (1000 kg each), it was confirmed that a sucking phenomenon of ground soil through gaps of rip rap would occur around the caisson. The challenge was to find a way to prevent the scour. Several scour protection measures were analyzed, and a new practical anti-scour filter unit concept was developed (Figure 2).
Akashi-scour pile scour
Figure 2. Scour protection measures considered for Akashi Kaiko (Kitagawa et al., 1991)
The selected measure included the installation of a filtering layer with a thickness of 2 m in a range of 10 m around the caisson. Then, the filter was covered with rip raps of 8 m thick. The filter unit consisted in a netbag weighing 1 metric ton and filled with crushing stones with diameters between 30 mm and 150 mm. Also, a surrounding area of 240 m was covered with rip rap. Then, a continued monitoring/investigation continued for the next 7 years.

After the 7 years period, it was found that scour occurred at the out-most peripheral, but is currently stabilized. Figure 3 shows the caisson without the protection and with the protection.
Figure 3. Caisson without protection (a) and with protection (b) (Source: Kitagawa et al., 1991)

Photovoltaic solar farms

Nowadays it is normal to hear about renewable energy. Solar energy is a renewable energy becoming very popular. Moreover, some people even call it “The edge of sunshine”. In this post we will provide a short introduction about solar power.

Figure 1. Solar farm
Source: PV magazine

Solar energy can be captured by thermal systems and photo-voltaic systems. In this post we will introduce photo-voltaic systems. Solar power is based on photo-voltaic, the conversion of light into electricity at the atomic level. When a given material absorbs photons of light, it releases electrons; thus creating an electric current that can be used as electricity. This property of capturing photons and releasing electrons is known as the photoelectric effect.

Figure 2. Photovoltaic system
Source: DitrolicSolar

Sunlight is composed by ultraviolet rays, the visible spectrum rays and infrared rays. All of them are ray lights but with different wavelengths. The so called Solar Constant is the total energy flux outside the Earth’s atmosphere and it is equal to 1367 Watt per sq m. Considering a cross sectional area of the Earth’s surface of 127.4 million sq km, the solar energy flux received by the earth is about 1.7 *10^17  Watt. Thus, in one day the Earth receives more solar energy than the energy consumed by world’s population in 1 year. Unfortunately, not all this energy is converted to electricity. However, in order to estimate the potential solar electricity we have to consider some limitations such as available sunlight hours, angle of incidence, atmospheric absorption and efficiency. 

Astronomical movements
Because of astronomical movements from the Earth, the solar insolation varies constantly. For instance, because of the earth’s rotation no energy can be stored during the night; only daylight hours can be used to store energy. This sunlight hour’s variation is more critical at northern and southern latitudes because of latitude and the Earth’s tilt. During winter season it is normal to have Septentrional and Austral latitudes with several months of only a few sunlight hours (or none).  Besides, it is important to consider that the solar intensity changes within the day. At noon the sun is at its highest point in the sky and the energy reaches the ground perpendicularly. Thus, we have the highest insolation, also known as full sun. This is a very important point for designing solar farms; the inclination given to the panel will define the energy capacity.

Angle of incidence
The optimum angle of incidence is the one that allows a solar panel to be perpendicular to the Sun’s rays. Thus, the whole solar panel will receive energy. Otherwise, when the panels are not perpendicular the effective collecting area decreases according to the non-perpendicular angle. 

Figure 3. Incidence correction]
Source: Solar power

Atmospheric absorption
It is important to consider that the atmosphere is a mass that absorbs energy. Thus, when the light passes throughout the air, some energy is absorbed by the air. Meteorological conditions such as clouds increase the absorption rate. Locations at high altitude have more solar potential because less energy is absorbed by the atmosphere. Thus, it is important to consider meteorological conditions and the elevation of the proposed location for a solar farm.

Unfortunately, any solar panel and any transformation device has an efficiency far from 100 %. Most of the solar panels have an efficiency between 15% and 22%, depending on the material and manufacturer. Material Science Engineering is performing important research for creating new improved solar panels. Thus, every year the solar panels have higher efficiency and their price tends to reduce.

How many Watts can be produced by solar farms?
The capacity of a solar farm depends on many factors such the geographical location, the size, the number of solar panels and the type of solar panel. Thus, it is not possible to give an average capacity. In this post we provide the capacity of the 5 largest photo-voltaic farms in the world.

Table 1. Top 5 largest photo-voltaic solar farms (Source:Imeche
Solar farm Location Capacity Area
Solar Star California USA 579 MW 13.0 sq km
Desert Sunlight California USA 550 MW 15.4 sq km
Topaz California USA 550 MW 24.6 sq km
Longyangxia Dam Solar Park Qinghai China 320 MW 9.16 sq km
Golmud Solar park Qinghai China 200 MW 5.64 sq km

Further reading:

HEC-RAS evolution

4 years  ago, the new version of HEC-RAS version 5 with 2D capabilities was released. HEC-RAS is one of the most popular hyodynamic models. it is a freeware that includes a DLL to develop your own script/ codes. In this post I will provide a fast summary of the different versions of HEC-RAS.
HEC RAS  window
Figure 1. User interface of HEC-RAS v1 (Up) and HEC-RAS v5 (Down)

HEC-2 or version 0

In 1968, the U.S. Corps of Engineers released HEC 2, the first American computer program for computing water surface profiles in channels with irregularly shaped cross sections. This program was written in Fortran and was able to calculate water surface elevations by solving the conservation of energy equation with the same iterative method used by HEC-RAS. The user interface was just a console screen with numbers and letters. Alothough it was not officially a HEC-RAS, I mention it as version 0 because its algorithm is the core of HEC-RAS.
The origin of HEC-RAS river modelling
Figure 2. User interface of HEC-2. 
Source: HEC-2 manual

HEC-RAS Version 1

The first version of HEC-RAS was released in 1995, with a very basic user interface. I have the idea that this basic user interface is a key element for the high popularity of HEC-RAS. It is a simple, yet easy to understand interface. A comparison between the first interface of HEC-RAS and the interface of the newest HEC-RAS shows that there was little change in the interface (Figure 1).

HEC-RAS Version 2

HEC-RAS 2.0 was released in 1997. Then, in the late 1998 HEC-RAS version 2.2 was released. The Federal Emergency Management Agency states that the most significant changes of this version were the corrections in bridge and culvert modeling.

The previous versions of HEC-RAS incorrectly calculated conveyance through bridges and culverts in pressure flow situations. Specifically, the problems occurred in the momentum equation calculations for bridge analysis and in distinguishing between inlet and outlet control situations for culvert analysis.

In general, culverts analyzed with HEC-RAS Version 2.2 under pressure flow will show lower water surface elevations than previous versions of the model. However, bridges analyzed with Version 2.2 under pressure flow generally show higher water-surface elevations using this version compared to previous versions of the model.

HEC-RAS Version 3

In January 2001 the HEC-RAS version 3 was released (The first HEC-RAS that I have used). The main innovations of this release were the capabilities to perform unsteady flow simulations and the introduction of GIS with GeoRAS.

HEC-GeoRAS changed the methodology for entering the geometric data and presenting results. Unsteady flow capabilities and GIS capabilities increased the range of applications and changed forever the way of showing results. Now, it was possible to show flood maps instead of just water profiles.
Figure 3. HEC-GeoRAS

HEC-RAS Version 4

In March 2008 HEC-RAS version 4 was released. The most significant modelling improvements of this version were the capabilities for sediment transport modelling and water quality modelling.

New buttons were included to the graphical user interface: 1 button for quasy-unsteady flow conditions, 2 buttons to enter sediment and water quality boundary conditions and 2 buttons to perform the water quality and the sediment transport simulations. Besides, this was the first version that included the RAS Mapper, which is the RAS own GIS tool. Now, HEC-RAS does not need any external GIS software for showing flood maps.

HEC-RAS Version 5

In February 2016 HEC-RAS version 5 was released. This was a very expected release. Not only because of the 2D capabilities, but also the release of the beta version in 2015 increased the expectation and the rummors about this new version; the social media and the internet forums were fulled with comments about this new version.

The capability for 2D simulations is a great improvement that increases the range of applications for HEC-RAS. Now HEC-RAS can be used for detailed 2D flood studies and dam break flood simulations.
Figure 4. 2D Dam break simulation with HEC-RAS. 

The first scientific publication using this new 2D capabilities became available also in February 2016 (using the beta version). Personally, it is personal honor to be the author of such publication and I hope that it can be useful for future studies.
This video shows an  animation of a 2D flood simulation and a hypothetical dam break simulation with HEC-RAS v5.
Now I wonder what will be the next big improvement of HEC-RAS? what will be the most important innovation for HEC-RAS 6? The current version 5.0.6 includes precipitation capabilities. Will version 6 become an integrated hydrologic-hydraulic model? Version 4 included lateral scour. will version 6 go for a fully morphodynamics approach? Give us your opinion on what would you like for a future HEC-RAS 6 (link to form).