2D flood simulation HEC-RAS vs Mike21: Hydrodynamic flood simulation comparison

Two of the most popular 2D hydrodynamic models are HEC-RAS and Mike 21. A common question among users is: which one is better? 

As a matter of fact, both models solve the same equations. They use different numerical schemes, but they solve the same set of equations; thus, when properly applied, both provide good results. In this post we will present a real case 2D flood simulation comparison between HEC-RAS and Mike 21.

Figure 1. 2D flood observed by satellite image and simulation by HEC-RAS Mike21

Flood study area

The comparison event is the 2014 Llanos de Moxos flood, in the Bolivian Amazonas. Llanos de Moxos (LM) are vast floodplains between latitudes 12.0º S and 17.0º S and longitudes 62.5º W and 67.0º W in the Bolivian Amazonia. They have a mean elevation below 150 m above sea level and a gentle slope lower than 10 cm km-1. The central part of Llanos de Moxos is a floodplain that floods more than 1 000 000 km2. The main river is the Mamore river which is also the most important Bolivian river.

The flood event of February 2014 was among the most severe ones. Between February 02 and February 22 the estimated discharge increased from 4 200 m³ s^-1 to more than 13 400 m³ s^-1.

A region of the flooded area located between 14.26°S, 65.24°W and 14.99°S, 64.62°W was selected for the simulation. Such area was selected because it covers the city of Trinidad which is the most important city in the Bolivian Amazonia and the location were river discharge was estimated. The simulated period was between February 02, 2014 and March 02, 2014.

More details about the study area and the simulations are described by

The 2D hydrodynamic models

The flood event was simulated using the new HEC-RAS version 5 and Mike21. Both models solve the 2D Saint Venant equations.

Figure 2. 2D Saint Venant equation

Where h is the water depth (m), d is the time varying water depth (m), z is the surface elevation (m), p and q are the flux in the X and Y directions (m3 s-1), C is the Chezy resitance, g is the gravity acceleration (m s-2), t is the shear stress (Pa), t is the time (s), r is the density of water (kg m-3) and Pa is the atmospheric pressure (Pa).

The HEC-RAS and Mike21 models were set up as similar as possible.

• Both models were simulated in a Dell laptop 64 bits with 2GB RAM and 1.4 GHz.
• Both models considered the same grid cell size,
• Both models considered the same time step
• Both models considered the same boundary conditions.
• In both models the topography was based on the Shuttle Radar Topography Mission Digital Elevation Model with a resolution of 90 m.
• Both models considered an equivalent roughness coefficient.

It is important to note that HEC-RAS and Mike21 have a small difference regarding the roughness. HEC-RAS considers the Manning roughness m; on the other hand, Mike21 considers the Manning roughness M, which is the inverse of m. More details about the set up of the models can be found in:

• Application of 2D numerical simulation for the analysis of the February 2014 Bolivian Amazonia flood: Application of the new HEC-RAS version 5
• Analysis of exposure to vector-borne diseases due to flood duration, for a more complete flood hazard assessment: Llanos de Moxos, Bolivia
• Application of a 2D flood model for defining evacuation zones for cattle in the Bolivian Amazonia: Mike21

The comparison HEC-RAS vs Mike21

The comparison focussed on: a) flooded area, and b) flood hazard.

Figure shows a comparison of the flooded area simulated by the models, and the flooded area observed by satellite image MODIS. 

Both models provide a good agreement with the observed flood. The flood extent simulated by Mike21 is some 2.65 % bigger than the extent simulated by HEC-RAS. Despite the difference in the simulated flood extent, when considering the measure of fit F1 and F2, both models provide similar results.

The flood evolution 

Figure 3a shows the daily evolution of the flooded area simulated with HEC-RAS and Mike21. Results from both models have the same trend. A steep slope until February 12, and then a gentle slope. 

Figure 3. Simulated daily flood evolution

Until February 12 the results are similar. After February 11 the Mike 21 model provides bigger flood extent. The maximum flood extent according to the HEC-RAS model occurs on March 01; then, in March 02 the flood extent shows a small decrease. 

The maximum flood extent from the Mike21 model (2 296 km2) occurs on March 02 and is about 1.3% bigger than the maximum flood extent simulated by HEC-RAS (2 124 km2). 

Figure 3b shows the daily evolution of the flooded volume simulated with HEC-RAS and Mike21. After February 12 the Mike 21 model simulates more flood water volume than HEC-RAS. 

The maximum flood volume simulated with HEC-RAS occurs in February 25. The maximum flood volume simulated with Mike21 also occurs on February 25. Thus, February 25 could be considered as a critical day with the biggest flood water volume. 

After February 25 some flood water moves to areas outside the study region. Although the following days still show some increase in the flood extent area, those new flooded areas have shallow waters. 

Although both models simulate the maximum flood volume on the same day, the flood volume is different. The total flood volume simulated with Mike21 (7.69 km3) is about 8.3% higher than the flood volume simulated with HEC-RAS (7.10 km3).

Flood hazard

This comparison considered 5 flood hazard levels depending on the flood depth. 

Figure 4 shows the evolution in time of the flood hazard and the flood hazard map from the most hazardous day. Figure 3 also shows the overflowing locations number in the order that they begin to overflow. 

Figure 4. Flood hazard simulation HEC-RAS and Mike21

Both models show that the West margin of the Mamore river is the most hazardous one. Hazard levels H2, H3 and H4 are almost constant. Until February 12 the extent of flooded area categorized as extreme hazard increases rapidly. During this period both models have similar predictions with an area about 1.2*E9 m2. 

After February 12, 2014, there is little change in the flood. At February 15, the flood extent classified as H2 is almost equal to the flood extent classified as H1. 

The most hazardous day according to Mike21 is February 24, as flood hazard H5 reaches its highest extent. The most hazardous days according to HEC–RAS are between February 24 and February 26.

Numerical models provide similar results

HEC-RAS, Mike21, and any other 2D flood model must simulate certain base marks. Different agencies have compared and published the comparison of such bench marks., and show that they all have small differences, but overall provide similar results.

This short real case comparison demonstrated again that both models have similar performance, and both models provide good and similar results.