3.1. Basin-scale analysis
The watershed modeling by spatial analyst tools gives details about the basins’ morphometric features. Nine drainage basins of Aguarico, Branco, Cauca, Kuskokwim, Magdalena, Maritsa, Ohio, Sankuru, and Siikajoki were modeled in this study (Fig. 4). The basin characteristics derived from the watershed models were given in Table 2. The black dots in Fig. 4 indicate each watersheds’ outlet, and red dots inside the basins indicate the location of the first emergent MCI in the main channel upstream. The satellite observations show that MCIs exist abundantly between the specified red dots and outlet. It can be seen in Fig. 4 that MCIs are found in lowlands. The MCIs existing location on basin maps in blue areas is visible. The formation of MCIs in depositional lowlands was also confirmed in a study by Heidari et al. (2021).
Table 2
Basin characteristics (continued).
| Area (km2) | Basin length (km) | Basin width (km) | Form factor (-) | Relief (m) |
Aguarico | 21348 | 322 | 156 | 0.206 | 5674 |
Branco | 196573 | 730 | 570 | 0.369 | 5290 |
Cauca | 41680 | 683 | 95 | 0.089 | 8493 |
Kuskokwim | 126370 | 752 | 285 | 0.223 | 3518 |
Magdalena | 257681 | 1033 | 414 | 0.242 | 8518 |
Maritsa | 51480 | 284 | 227 | 0.638 | 2897 |
Ohio | 520485 | 1104 | 724 | 0.427 | 1954 |
Sankuru | 134397 | 696 | 266 | 0.277 | 837 |
Siikajoki | 22367 | 272 | 116 | 0.302 | 4285 |
Table 2
| Mean basin slope (-) | Mean basin elevation (m) | Main channel length (km) | Main channel slope (-) | Distance between the first MCI and outlet (km) | Number of islands in main channel |
Aguarico | 0.264 | 1417.6 | 390 | 0.00072 | 330 | 230 |
Branco | 0.093 | 221.1 | 560 | 0.00009 | 556 | 200 |
Cauca | 0.427 | 1679.3 | 1349 | 0.00101 | 130 | 80 |
Kuskokwim | 0.175 | 338.7 | 1130 | 0.0001 | 1130 | 300 |
Magdalena | 0.265 | 1144.3 | 1528 | 0.00093 | 700 | 560 |
Maritsa | 0.146 | 415.5 | 480 | 0.00046 | 410 | 320 |
Ohio | 0.133 | 350.5 | 1579 | 0.00008 | 1579 | 62 |
Sankuru | 0.109 | 708.8 | 1200 | 0.00011 | 1170 | 108 |
Siikajoki | 0.116 | 158.2 | 160 | 0.00048 | 50 | 42 |
The high areas in each examined basin have significantly different elevations. Consequently, in the Sankuru basin, which has the lowest elevation of all, MCIs exist on the blue, green, and yellow parts of the map. Also, the Ohio basin, with the second rank being low, has a similar condition to Sankuru. Comparing all the cases to each other demonstrate that the MCIs’ location is associated with the elevation distribution/values in the basins. The lower the basin elevation is, the earlier the MCIs emerge in the river (i.e., longer distance to outlet).
The examined rivers are located in the areas with dense tree cover in the continuous vegetation fields (VCF) maps (Fig. 2). The land cover figures are presented to quantify the type and percentage of vegetation cover (Fig. 5). Figure 5 confirms that vegetation is the dominant land cover of all studied basins. More than 70% of these basins are covered by various vegetation types (i.e., tree, grass, crop, scrub, and shrub), and they accommodate many MCIs in their main channel. This situation indicates the constructive role of vegetation cover in generating MCIs. More specifically, the vegetative fragments and large woody debris, which are transported during high flows and deposited in the active channel and their wake regions cause sediment deposition (Yagci and Strom 2022), the establishment of MCIs, and keeping these instream formations stable against peak flows (e.g., Fetherston et al. 1995; Gurnell et al. 2001; Wyrick and Klingeman 2010). These basic facts demonstrate the significant role of vegetation cover distributed over the basin in the aggregation of river morphology.
3.1.1. Slope effect
MCIs can emerge in any major river where a dynamic flow regime, sediment supply, and vegetation propagules are available (Gurnell and Petts 2002). Figure 6 exhibits the relationship between main channel slope and MCIs existence length in the main channel normalized by main channel length. According to the ecoregion map, the studied rivers were categorized into three classes: (1) Tropical and subtropical forests, (2) Temperate broadleaf/grassland, and (3) Boreal forests.
The correlation between main channel slope and MCIs existence length is noticeable in tropical and subtropical forests in Fig. 6. As the channel slope gets lower, MCIs emerge earlier upstream (i.e., longer distance to the outlet). Also, this relationship is similar in two other classes. Kuskokwim and Ohio rivers have lower basin slopes than the other rivers. For these basins, MCIs exist from the beginning of the main channel and continue to the outlet. However, the degree of this correlation is not comprehensible due to the scarcity of data. The study by Heidari et al. (2021), who examined three major rivers in Turkey, showed the same pattern for the location of MCIs. Although the fewer examined rivers left uncertainty, the data for different ecoregions in this study made the correlation clearer. This relationship implies that channel bottom slope induced gravity forces play a dominant role in the deposition process of the sediment particles and the dynamic behind the emergence of the midchannel islands. In Fig. 7, the variation of mean elevation of the basins with respect to mean slope showed a direct and distinct correlation for the examined cases. Therefore, it can be drawn that the higher the mean elevation of the basins, the higher the main channel slope. This basin setup (relationship between high elevation and a steep main channel slope) delays the emergence of MCIs toward the outlet.
3.1.2. Specific sediment yield effect
In order to understand the significance of sediment supply in the formation of MCIs, specific sediment yield (ton.km− 2.yr− 1) was calculated for each basin, as explained in the methodology section in a detailed manner. The relationship between specific sediment yield and the number of MCIs in the main channel normalized by their existence length (starting from upstream up to the outlet) in Fig. 8 shows that in tropical and subtropical rivers, increasing the specific sediment yield leads to the emergence of more islands in a unit length.
In temperate broadleaf/grassland class, the Ohio basin has a remarkably high urban area (Build area: 25%) according to Fig. 5. It is a well-known fact that urbanization negatively impacts on the generation of riparian vegetation in active river corridors (Porter et al., 2001; Moffatt et al., 2004; Pennington et al., 2010) and leads to channel degradation (Yousefi et al., 2017). Therefore, the number of MCIs in their existing length is significantly lower than Maritsa (Build area: 4%). In the boreal forests class, the ratio (i.e., number of MCIs/the distance between first MCI and outlet) is very high compared to Kuskokwim. However, this length is small, and MCIs exist only close to the outlet due to the steep main channel slope. On the other hand, Kuskokwim has more MCIs, and they exist all along the main channel due to the low slope.
The association of specific sediment yield with relief (i.e., the difference in elevation between the highest and lowest point of the basin) is quite distinct in Fig. 9 for all three ecoregion classes. Many researchers (e.g., Ali and Boer 2008; Hovius 1998; Jansen and Painter 1974; Milliman and Syvitski 1992; Pinet and Souriau 1988; Tamene et al. 2005) stated that the basin relief is one of the significant factors controlling the sediment yield. As can be seen in Fig. 9, the higher steepness of the basin leads to more specific sediment yield. However, the Aguarico river basin is not in alignment with the rest of the basins. It can be assumed that since the tree cover percentage is significantly higher in the Aguarico basin (Tree cover: 85%), the impact of the relief on the sediment yield becomes less pronounced. Instead, the excessive tree cover controls the sediment yield to a great extent.
Einsele and Hinderer (1997) mentioned that beside the basin relief, other characteristics could influence the specific sediment yield since the whole area of basins is not mountainous. The watershed shape is another influential parameter on the sediment yield (Zhang et al. 2015). The form factor (Horton 1932) characterizes the basin shape. The form factor value varies between 0 and 0.785 (i.e., a perfect circle). When the value is closer to 0, the watershed is more elongated. Figure 10 exhibits the relationship between specific sediment yield and the form factor of the examined basins. The specific sediment yield decreases when the basin is more elongated. The trend degree is different for Aguarico, Cauca, and Magdalena than for the other basins. According to the Table 2, the mean slope for Aguarico, Cauca, and Magdalena basins is significantly higher, while the values are close to each other in the rest of the basins. It can be inferred that the higher slope increases the impact of basin shape on sediment yield.
3.2. Morphometric analysis of the individual MCI units
The morphometric analysis of MCIs in the examined rivers was carried out on the reach scale. A large number of MCIs exist in the lowland reaches of every size, from pioneer islands to mature one. Figure 11 shows that MIP increases with a strong correlation and linearly during the growth in MCI width from the preliminary stage to the advanced stage. This strong relationship between island width and MIP is quite similar to the relationship between the diameter of the conceptual circle and MIP, as can be seen in Fig. 11a. Likewise, MIP increases when the island length grows as well (Fig. 12).
However, island length correlation with MIP is weaker than island width. This situation could be explained by natural objects adjusting their form in a streamlined manner when exposed to a flow condition (Schoelynck et al. 2012). Enhanced blockage strengthens the contraction effect and shear stress during the island growing process, as Heidari et al. (2021) demonstrated quantitatively. Increased shear stress and contraction effect constitute the main dynamics behind the streamlining effect on the island and lower correlation between the length of the island and MIP. Variation of MIP with blockage ratio (Fig. 13) demonstrates a similar correlation for the conceptual circles and MCIs. In the preliminary stage of MCI formation, the blockage ratio (i.e., contraction effect) is low, while the contraction effect becomes more prominent with the island growth. However, it is worth highlighting that for a given blockage ratio (W/D or W/Wb), while the cylinder takes much higher MIP values, in the MCI case, MIP takes significantly lower values. This difference becomes particularly pronounced in the advanced stage of MCI. This fact can be interpreted in the way that for a unit perimeter of an island, the area of the MCIs becomes less due to the streamlining effect.
By definition of MIP, the island area is associated with form drag, and the island perimeter is associated with skin friction. It is a well-known fact that when a body is exposed to flow, the total drag force acting on the body consists of the sum of the form drag and skin friction components (Sumer and Fredsoe 2006). According to Fig. 13, during the development of the MCI, the ratio between the form drag and skin friction changes depending on the island geometry. More specifically, in an advanced stage of an island, the form frag becomes a more influential factor compared to that of skin friction. The decreased skin friction in the mature stage of an island can be explained by the process of island streamlining due to the formative role of prolonged base and infrequent peak flows. Obviously, the streamlining process, which emerges in mature stages of the island, plays a significant role in decreasing skin friction.
The satellite imagery belonging to different years showes that the MCI size can change over time. Investigation of sample MCIs revealed that the island growth was not observed in all the studied rivers (Fig. 14). In Branco, Kuskokwim, Ohio, Sankuru and Siikajoki rivers the MCI morphometry remained constant in different years. It can be inferred that the incoming sediment supply to MCIs and sediment output are in equilibrium in these cases. The MIP increases over time in Aguarico, Cauca, and Magdalena rivers. According to the results of the previous sub-section, Magdalena, Aguarico, and Cauca basins have higher specific sediment yield, respectively, and the amount is remarkably more than other basins. It can be seen in Fig. 14 that the MIP growth in the Magdalena river, which has the highest specific sediment yield, is more significant than in Aguarico and Cauca. This result implies that the sediment supply exceeds the transport capacity of these three rivers. On the other hand, the MIP decreases in Maritsa’s MCIs. Therefore, the incoming sediment and transport capacity of the Maritsa river is not in balance, and there is a deficit in the sediment supply.