GAUGING RIVER MORPHODYNAMICS

What if we were to trace the change of a riverine morphological vitality over the span of decades and even hundreds of years? What if we were to assess river behavior influenced by various natural conditions and human influence? What if we were to contrast river morphological dynamics? Well, my method does exactly this and, as a matter of fact, it is fairly simple.

At first, let us have an impression of the existing magnitudes of an annual river morphological change using the examples of Lena and Brahmaputra rivers located in Northern and Southern Asia, respectively. As demonstrated by the pair of images below, the difference in the intensity of change can be striking. However, a method is needed to reasonably quantify it; the quantification is the key to the understanding of how different rivers really are and to what extent they are influenced by alternating conditions.

Introduction

In a nutshell, the method is comprised of three steps, which are the sorting of chosen geospatial data for a considered river sector, the comparison of water surface areas, and the creation of graphs. As a data source, Landsat Collections were used.

It has to be mentioned that, although for the analysis I chose four rivers (the Mississippi, the Lena, the Brahmaputra, and the Ganges), I am going to cut down on most intermediate results to enhance the digestibility of this article.

Step 1. Data Sorting

The analysis starts with the creation of a table of data availability for a chosen reach or, as it can also be called, sector of an analyzed river. The tables I used were created by the full scanning of the remote data collections in search of imagery of suitable quality. At this stage, captured-on-different-dates water surface areas within a sector are computed and their values are put in appropriate cells of a table. After the scanning, cells with similar values are highlighted using different colors, and series for similar water surface areas are formed. The combination of closely standing series can also be used to minimize the time gaps and increase the temporal resolution.

By the way, the interest in creating series of similar values is based on the assumption that the equality in water surface areas for different points in time within the considered sector denotes the equality in water levels; therefore, we can say that each series attributes to a certain imaginary at-the-very-water-surface horizontal “cut” of a river channel. Thus, the plan-view change in the configuration of this “cut” can be correlated to the morphological change of a river channel.

Following the above-described steps, here is what I got for the Mississippi and the Lena.

As we can see, the amount of available and appropriate data is dependent on both location and climatic conditions of an analyzed region. The latter can cause no-data seasons which substantially hinders the analysis. Specifically, for the Lena such absence of data is explained by harsh winters during which the river is frozen, whereas it is the monsoon summer seasons, with their excessive cloudiness, that influence the Brahmaputra and the Ganges. 

Steps 2 and 3. Comparison and Creation of Graphs

Generally speaking, the image comparison can be described as a two-phase process where both phases complement each other. As the identifier of a morphological change, I used the rate of change, which is the relationship between the area of an overall change happened per unit time within a considered reach, which includes both erosion and accretion areas, and the initial water surface area of the same reach.

One of the phases is based on the consecutive comparison of one chosen reference image with other following images of the same series. This comparison builds historical channel migration curves, which give a good representation of variations in the intensity of morphological change.

The next graph, for instance, depicts the changes in the rate of morphological change within a reach of the Mississippi river over the span of around 45 years. The three curves here represent the plan-view morphological change at three different levels of a river channel. The blue, red, and green curves correspond to low, intermediate, and high levels of the channel, respectively. It’s seen from the graph that the intensity of morphological change is happening at different rates depending on the considered level. Moreover, the abrupt changes in the upper two curves signify a significant change in the rate of morphological change that occurred in the 1980s.

The following two graphs are attributed to the Ganges River. For the convenience of visual comprehension, I shifted the three channel migration curves of the same series to one reference point in time. As a result, the decrease in the morphological change intensity after the 1990s of roughly 20 % is well distinguished. This slowdown, as it happens, is the result of the construction of the controversial Farakka Barrage, which was commissioned in 1975.

By bringing the channel migration curves of four rivers together in one graph we can now contrast them and get a feeling of the river liveliness. The noticeable difference between the Mississippi and the Lena, and the Brahmaputra and the Ganges is driven by anthropogenic influence, as referred to the Mississippi, and a strong permafrost influence in the Lena’s case.

The second phase of comparison represents variations in the rate of morphological change over time. For that, only those images from one series were compared that had the same chosen time period in between.

The two graphs below illustrate the alterations in the intensity of change for the Brahmaputra and the Mississippi rivers. From the first graph, which represents the Brahmaputra, it is seen that the trendlines created for 1-, 2-, and 3-year fixed periods are practically parallel meaning that there have been no variations in the rate of morphological change within the last four-five decades. On the other hand, by looking at the graph for the Mississippi it is clear that the rate of change has varied significantly over the years with an overall downward trend. With the help of Google Maps, I figured out that the high values of the rate of change during the 1970s and 1980s had been caused by the construction of dikes that temporally intensified the sediment accumulation within the analyzed reach.

Coming from the trendlines to values, I put together the intensities of an annual rate of morphological change to get this interesting chart. 

According to the values, the dynamics of a river morphological change can vary greatly depending on both natural and anthropogenic factors. Remarkable, however, that even under natural conditions the morphological dynamics can differ on average by 14 times, as shown by the example of the Lena and the Brahmaputra rivers.

Summary and Outlook

To sum up, the method has proved to be a reliable and universal tool for riverine morphological change analysis. Mainly the robustness of the method, with respect to remote sensing data, is based on the data interconnection and the variety of possibilities for data comparison. Thus, faulty and missing data can be deductively identified. With the introduced method, not only it is possible to quantify and assess the intensity of change, but also to trace the periods of significant morphological changes and to a certain degree characterize them. Furthermore, prognoses can be made with the assumption that the conditions of considered systems stay the same. The digitalized historical data can be effectively used for the approximation of the intensity of change in the distant past. Using the method it is possible to cross compare different types of river channels because values of the rate of change can be determined for all of them.

From my point of view, this method can also find a practical application in short- and long term observations that follow the implementation of new riverine modifications and various river restoration projects. Although, in this case, geospatial data of higher temporal and spatial resolutions will most likely be required.

P.S. ....and there is still so much more about the method I'd like to tell.

I thank Professor Markus Noack for all of the support of this work.