Multi-decadal Landsat data and scanned copies of aerial photographs were used to investigate the spatial-temporal evolution of these 14 fluvial islands. Aerial photographs taken in 1945 at a scale of 1:250,000, were obtained from the U.S. Army Map Service, Series L500 (https://www.utexas.edu/). Landsat MSS (Multispectral Scanner), Landsat TM (Thematic Mapper), and ETM+ (Enhanced Thematic Mapper) data from December 1976, 1986, 1996, 2006, and 2016, were acquired at 10-year intervals (Table 2), with cloud-free images being selected to enhance the clarity and richness of the image information. Furthermore, to minimize the effect of the water stage on interpreting the change in the size and shape of the fluvial islands, all Landsat images used were captured during the dry season of November and December when the discharge was considered constant within the channel in the study area. The Landsat dataset was acquired from the Geospatial Data Cloud (http://www.gscloud.cn).

The annual runoff and sediment discharge data from 1950 to 2016 for the Datong hydrological station were obtained from the Changjiang Water Resources Committee (CWRC) (http://www.cjw.gov.cn/). Statistics indicate that from 1950 to 2016, the average annual runoff was 8,960 × 10⁸ m³, and from 1950 to 2016, the average annual sediment discharge, average annual sediment concentration, and average sediment delivery were 3.63 × 10⁸ t, 0.414 kg m^-3, and 216 t km^-2, respectively. The maximum runoff of 13,590 × 10⁸ m³ and minimum runoff of 6,671 × 10⁸ m³ were recorded in 1954 and 2011, respectively. The highest sediment discharge was recorded at 6.78 × 10⁸ t in 1964, while the lowest sediment discharge was measured at 0.72 × 10⁸ t in 2011. In this study, the runoff and sediment discharge data at Datong Station from 1950 to 2016 were used to analyze the evolution of the islands and evaluate the uncertainties associated with changes in the area of the fluvial islands, as inferred from the interpretation of the remote sensing data.

Multiperiod Landsat data and scanned copies of aerial photographs were used to perform a spatial-temporal analysis of the fluvial islands over approximately a 72-year time series. Only permanent islands were selected in the study area (Figure 1). The aerial photographs were mosaiced and vectorized, and the bank lines were digitalized in ArcGIS 10.2 (ESRI Inc., United States).

To assess the variations in the land area of the fluvial island, a geometric correction process was applied to each satellite image. This involved transforming the images from the Universal Transverse Mercator (UTM) projection to the Albers Equal Area Conic projection, ensuring a uniform and accurate representation of the island’s topography for comparative analysis. To correct the resolution discrepancies between TM and MSS images, we employed a combination of band synthesis, region of interest (ROI) subsets, and resampling tools in ArcGIS 10.2. Using the near-infrared bands available in both TM and ETM+, we can effectively differentiate between water bodies and vegetation cover. Additionally, to ensure accuracy, we refine the image correction process by selecting the ground control point (GCP) using ENVI 5.1 (ITT Visual Information Solutions Inc., United States).

Using a suite of remote sensing image processing techniques, the morphology and area of the selected fluvial islands were delineated for the period from 1945 to 2016. To elucidate the trends in area change, the following equation was applied to compute anomalies for both individual island areas and the aggregate total area (see Eq. (1)). ∆ A i j = A i j − A j ¯ (1) where the term j refers to a specific fluvial island, i denotes a particular year, ∆ A i j represents the area change of a specific fluvial island j in the year i, A i j is the area of a specific fluvial island j in the year i, and A j ¯ is the mean area of the fluvial island j over 72 years.

Historical aerial photographs and remote sensing imagery from the years 1945, 1976, 1986, 1996, and 2016 were utilized to delineate the boundaries of the fluvial islands in the lower Yangtze River. All images were scanned at a resolution of 600 dpi and geo-rectified using ArcGIS 10.2. During the image rectification process, each photograph was resampled at a resolution of 0.5 m to ensure consistency across all images. Rectification resulted in a root mean square error of <5 m for all images.

Individual image frames and mapping tables were mosaicked in ENVI to obtain georeferenced images that cover the entire study area. The image stitching process is influenced by two sources of errors: distortion errors between adjacent frames and inherent errors of the orthoimages. When GCPs are combined in areas that overlap between adjacent frames, distortion caused by topographical variations is minimized, thereby reducing these errors to a minimum.

The shoreline of the Yangtze River in Anhui Province, which stretches a straight-line distance of 250 km, was systematically divided into segments with equal intervals of 1 km. Cross sections were defined as perpendicular to the trend surface along the river segment. For this length, a 1 km interval is considered sufficient to characterize the morphological changes of the river channel within the study area. The displacement of both banks for each segment was determined using the methodology proposed by Kummu et al. (2008). Using GIS overlay analysis, the banklines of two periods were compared to quantify the displacement of each segment. Subsequently, the intersection points of the terminal lines of each end-member with the riverbank lines were calculated and their coordinates were compared to determine the variations in the riverbank line locations.

Remote sensing imagery and aerial photographs were used to extract the morphological characteristics of typical fluvial islands from different periods (Figure 2). As illustrated in Figure 2, the overall morphology of the fluvial islands appears relatively stable in the past 72 years. However, due to fluvial erosion and sediment deposition, some islands exhibit significant local changes. Specifically, HSZ and WZ have undergone erosion, while TMZ, FHZ, and YDZ have experienced significant sediment deposition. In the northern HSZ island, a diminutive island emerged around 1945. This island underwent a modest decrease in its expanse by approximately 1996 and ultimately integrated with the principal island by the year 2006. Concurrently, since 1966, a new island has been gradually accumulating at the northern edge of ZJZ, maturing into the more stable TMZ. Furthermore, WZ Island has been observed to exhibit convergence with the riverbank, while LZ and ZJZ have become morphologically stable.

The total area of the 14 selected fluvial islands experienced an expansion, growing from 373.3 km² in 1945 to 394.9 km² in 2016 (Figure 3A). Prior to 1986, the aggregate area of the islands witnessed a steady rise, with an average annual expansion of 0.32 km² yr^-1. In contrast, the period between 1986 and 1996 saw a swift decline in the total area, with an erosion rate of 1.5 km² yr^-1. Subsequently, post-1996, the total area resumed an upward trend at an average annual rate of 0.73 km² yr^-1.

During the last 72 years, the average areas of individual islands can be ranked as follows: ZJZ > TBZ > HSZ > LZ > XZ > FHZ > WZ > CSZ > TTZ > TMZ > QJZ > XHZ > YDZ > EMZ (Figures 3B, C) with the areas of the 14 islands varying significantly before the 2000s and then remaining relatively stable after that. Specifically, the island ZJZ is the main island of Tongling City and the largest of the 14 islands, with an average area of 69.1 km² and erosion that occurred during 1945–2016 (Figure 3B). The EMZ has the smallest area, with an average annual area of 3.5 km² (Figure 3C). TBZ and XHZ were also observed to show similar tendencies, with an increase from 1945 (42.6 km² and 8.8 km², respectively) to 1986 (61.9 km² and 10.9 km², respectively) and then a substantial decrease from 1986 to 2006, followed by a slight growth of 0.2 km² and 0.51 km² from 2006 to 2016.

As presented in Figure 8, the annual runoff at Datong Station did not show an obvious trend, except that the annual runoff for the years 1954 and 1998 was much higher than that in other years. However, the annual sediment discharge experienced a notable decline post-1986 (Figure 8), which had a significant impact on the sedimentary input and the total area of fluvial islands (Figure 3A), thereby affecting the formation and stability of the fluvial islands. This trend suggests that there was an insufficient sediment supply to support the expansion of the islands. Previous studies have found that natural factors such as runoff and sediment discharge play an important role in the evolution of fluvial islands or deltas in large rivers (Baki and Gan, 2012; Ve et al., 2021; Quang et al., 2023). The annual migration rate of high sediment concentration rivers is higher than that of low sediment concentration rivers, due to the rapid growth and migration of the bars that force bank erosion (Constantine et al., 2014). The average annual sediment discharge was 3.63 × 10⁸ t and decreased drastically after the construction of the TGD. The area of the 14 islands tended to decrease with the decrease in sediment content, while increasing the sediment content. This coincides with the findings of Li et al. (2019), and similar phenomena have been observed in the channel bars in the middle Yangtze River. River sediment discharge is usually considered one of the major factors controlling island dynamics (Moretto et al., 2014). In this study, it is inferred that sediment discharge is the primary factor responsible for the observed reduction in the area of the fluvial islands.

Flooding is also considered a significant factor that causes morphological changes in fluvial islands (Asaeda and Rashid, 2012). Within the time span from 1945 to 2016, there were two historic crests. The data presented in Figure 8 also show that annual runoff and sediment discharge reached a relatively high level, while the total area of the fluvial islands decreased to the lowest in 1954 and 1998. In 1954, a huge flood occurred with the water stage reaching 16.6 m and a maximum flow of 92,600 m³ s^-1 at Datong Station. In 1998, another catastrophic flood occurred with a maximum flow of 81,000 m³ s^-1 (CWRC). The flood in 1998 reduced the total area to its lowest area after 1950. Bertoldi et al. (2009) also observed similar phenomena in their study of the Tagliamento River, a braided river where water flow and sediment transport are largely uncontrolled, with extreme floods being a major cause of erosion on fluvial islands. Annual runoff also influences the flow velocity along the fluvial islands, with the river segments of the study area exhibiting relatively stable flow velocities. It was observed that narrower stretches of the river exhibit higher velocities compared to wider ones and that velocities are higher upstream than downstream. Consequently, fluvial islands situated in regions with higher velocities, such as EMZ and HSZ, experienced more erosion (Moretto et al., 2012).

The evolution of fluvial islands is the result of long-term interactions between river boundaries and sediment transport (Shi et al., 2017). In this case, the evolution of fluvial islands in the lower reaches will be affected by riverbank erosion and deposition. The left bank shows more deposition than erosion, while almost all sections of the right bank show deposition. This may be due to the fact that a large number of shore protection projects effectively prevent shore erosion. The area with the largest erosion distance on both banks happens to be along with the merging of fluvial islands, and a similar island merging phenomenon also occurs in the Zhen-Yang reach of the Yangtze River (Yang, 2020). Fluvial islands with large morphological changes are found around the bank with evident displacement, suggesting that the evolution of fluvial islands in the Anhui section of the Yangtze River may be closely related to the erosion and deposition process on both banks of the Yangtze River. Comparing this study with the findings of Baki and Gan (2012) and Hudson et al. (2019), we suggest that the riverbank migration plays an important role in the island dynamics. Specifically, in this study area, the narrowing of the river channel will hinder the development of fluvial islands, while the widening of the river channel will promote the silting and growth of fluvial islands. Further investigations should be carried out to detail this point in the future.

The evolution of fluvial islands is also influenced by other natural factors, such as precipitation, vegetation cover, and vegetation types (Baubiniené et al., 2015). In the present study, the area changes of ZJZ and TBZ are relatively minor, and they are situated under similar hydro-sedimentary conditions. However, they have experienced different erosion and deposition processes, which is hypothesized to be due to the overflow of water from the lakes within the fluvial islands during the rainy season, leading to the loss of soil and shallow banks after the river scavenging. Furthermore, the higher the vegetation coverage and the richer the vegetation types, the less soil erosion on the island. Among them, trees are superior to shrubs and grasses, mixed forests are better than single-species forests, and the dense reeds along the outer edge of the island also serve as natural flood barriers. Short shrubs and trees that grow in the floodplain can also play a good role in stabilizing the embankment (Corenblit et al., 2015). In a study on channel bar morphodynamics in the middle Yangtze River, Wang et al. (2018) reported that the bars with high vegetation coverage tend to be stable during floods, while the bars with less vegetation are prone erosion. It can be seen that the vegetation can enhance the anti-erosion ability and thus promote the stability of fluvial islands.

With the development of the social economy, anthropogenic influences play an increasingly significant role in the evolution of fluvial islands (Luan et al., 2018; Yang et al., 2018). Due to overexploitation and unreasonable land use, bank collapse often occurs. The evolution of the fluvial islands within the lower Yangtze River is also related to anthropogenic engineering and planning. The construction of the dams, the submerged banks, bank reinforcements, as well as guard works, have been significant factors in preventing shoreline collapse, improving the channel conditions, and stabilizing the branching pattern. These projects have weakened the scouring in the channel and protected the river channels and dikes along the two banks (Hudson et al., 2019).

In the lower reaches of the Yangtze River, the significant reduction in sediment discharge and external material input, which are necessary for the development of islands, is closely related to the construction of the TGD (Shi et al., 2017). As reported by Chen et al. (2016), the construction of over 50,000 reservoirs in the Yangtze River basin since the 1950s has had a minimal effect on the annual runoff, yet the seasonal fluctuations in dam storage could influence the monthly distribution of runoff. The completion of the TGD has been associated with a notable reduction in the annual volume of sediment transport, serving as a compelling indicator of the significant impact of human activities on sediment dynamics (Zhao et al., 2017). Typically, the increasing runoff correlates with the increasing scouring, leading to a decrease in the area of fluvial islands due to erosion. However, since 2006, the total area of these islands has paradoxically expanded along with the annual runoff (Figures 3, 8). This trend could be attributed to the government’s initiatives to reinforce the embankment system along the Anhui section of the Yangtze River, which has mitigated island erosion and facilitated ongoing sediment deposition.

Previous studies have revealed that ongoing enhancement of embankment systems has been instrumental in mitigating erosion and submersion of fluvial islands by river waters, fostering their stable development and progressive accretion, which in turn has led to a gradual expansion of their surface areas (Hudson et al., 2019). Following the completion of the TGD, both the individual and cumulative areas of the fluvial islands exhibited a slow growth. This indicates that the significant decrease in sediment supply from upstream, has hindered the expansion of these islands due to the construction of the TGD (Li et al., 2016). Additionally, the dam has exacerbated the scouring effects during periods of high water, further eroding the islands and impeding their sedimentation and expansion. This process has also been observed to induce changes in the morphology of the fluvial islands (Luan et al., 2018).

Human activities such as sand mining, reforestation, and water regulation projects have a profound impact on the development of fluvial islands (Gurnell et al., 2005). Since the end of the last century, the evolution of these islands in the Lower Yangtze River has been notably influenced by human interventions. Practices such as riverbed dredging have incised the riverbed, leading to deeper and narrower channels, which in turn affect the morphological development of the fluvial islands (Lou et al., 2018). Previous studies have demonstrated that the fluctuation in the area of the fluvial islands of Yangtze River is a direct result of changes in the hydrological and sedimentary regimes. Sand and gravel extraction can fragment the riverbed, fostering vegetation growth and accelerating the amalgamation of fluvial islands (Moretto et al., 2012). Afforestation efforts on these islands, coupled with increased vegetation cover, have a stabilizing effect on the riverbanks and the islands, thus influencing the dynamics of sediment transport. This intervention limits erosion and channel migration, leading to channel incision and enhanced stability of the fluvial islands. These anthropogenic influences indirectly modify the distribution of water flow and can even alter the main distributaries, thereby affecting the conditions of exogenous material for the formation of fluvial islands.

Furthermore, the formation and evolution of fluvial islands not only influence their morphology, area, and direction of their development, but also profoundly influence their geochemical characteristics and regional variations. The age of island formation is correlated with the degree of soil development and the extent of soil utilization, which, in turn, markedly affects the accumulation and migration patterns of the physicochemical properties and elements of the soil (including the major and trace elements). This inevitably impacts their geochemical characteristics and regional differences. In contrast, variations in land use and soil properties also indirectly reflect the surface processes within the watershed, to some extent affecting the stability of the morphology of the fluvial islands. The interplay between these factors and their implications for the morphology and geochemistry of fluvial islands remains an area that is ready for further investigation.