We initially performed a detailed analysis of the river terraces along the Qingyi River, employing a 12.5-meter resolution Advanced Land Observing Satellite (ALOS) digital elevation model combined with high-resolution satellite imagery to comprehensively identify terrace surfaces on both riverbanks and to statistically assess their relative elevations. Fieldwork was subsequently conducted to validate the morphology and relative elevations of these terraces. During field measurements, we employed a Global Positioning System (GPS) equipped with network real-time kinematic (RTK) differential functionality, capable of achieving a planar accuracy of 5 cm and an elevation accuracy of 8 cm, enabling high-precision surveying. Besides, we conducted field measurements of several geological profiles across river channels along the Qingyi River to analyze the distribution, terrace elevations, sediment thickness, and other relevant characteristics. According to the field survey, the terraces of the Qingyi River are spatially discontinuous because of heavy erosion. Tectonic activities further changed the height of these discontinuously distributed terraces. Terraces of different ages are difficult to distinguish during field surveys. However, in the process of terrace profile analysis, it was found that there were specific differences in terrace sediments of different ages, mainly reflected in the hardness of gravel and the properties of matrix sand filled between gravels. The older the terrace, the stronger the degree of fragmentation, and the corresponding matrix is more cemented and clay-like. In addition, there are apparent differences in the color of matrix sand between gravels of different ages in terrace sediments. These characteristics probably reflect the attributes of terrace sediments becoming more weathered with increasing age. Therefore, we hope to compare the weathering degree of river terraces quantitatively. To realize the comparison work, a few problems must be solved.

The primary challenge in sampling major elements is determining the appropriate method for selecting the material to be analyzed. An ideal sample should accurately represent the average degree of weathering of a terrace surface. Through the features of terrace outcrops, it can be seen that the gravel’s degree of weathering reflects the terrace’s age. In the high terraces, the gravels are fragmented, whereas those in the lower terraces remain largely intact. The sandstones and mudstones in terraces T5 and T6 have weathered into mud (Jiang et al., 2016; Jiang et al., 2022). However, assessing the degree of gravel weathering is challenging, primarily due to the heterogeneous provenance of the gravels. Some are sourced from distant sedimentary deposits in the river’s upper reaches, while others come from proximal deposits in nearby tributaries. Consequently, there is no quantitative means of judging the degree of gravel weathering (Alberti et al., 2013; Nishiyama, 2002). However, due to the long-distance transportation and thorough mixing, the matrix sand filling the interstices between gravels is relatively uniform and can more generally represent the terrace sediments (Jennings and Shulmeister, 2002; Chambers et al., 2013). Therefore, we analyze the weathering degree of the matrix sand between gravels and further solve the problem of quantitative terrace comparison.

Another significant issue is determining the appropriate depth for sample collection. Most sedimentary terraces have a typical binary structure. Previous studies on terraces have found that mountain rivers often experience basin-wide floods, resulting in mixed flood sediments deposited on the top of the binary structure. This sediment type is called Slackwater Deposits and Paleostage Indicators (SWD-PSI) (Kochel and Baker, 1982; Baker, 1987; Baker, 2008). These flood deposits are often younger and more variable in age, and their sediments cannot represent the age of the original sediments of the terraces, which usually causes the problem of terraces being too young in terrace chronology studies (Lavé and Avouac, 2000) (Figure 3A). Therefore, we avoided flood and floodplain deposits during sampling. On the other hand, in our previous study, we conducted systematic sampling along the depth of a terrace near Qingshen County, collected sand filled between gravels as major element test samples, and analyzed the variation characteristics of the weathering degree of a fine-grained matrix with depth, to determine a suitable scheme for the sample depth of terrace weathering degree research (Jiang et al., 2021). We observed a gradual decrease in the weathering index from the surface to greater depths in the profile. Above a depth of 2 m, the weathering degree is high, and the weathering intensity tends to be consistent. Below 2 m, it begins to show a linear decreasing trend. At a depth of about 10 m, the weathering index decreases to 50, and no more extended changes (Figure 3B). This indicates that samples collected within 2 m from the ground surface can better represent the maximum weathering degree of the terrace (Jiang et al., 2021). In this study, samples were collected from the upper section of the gravel layer beneath the floodplain deposits, along with sand that filled the spaces between the gravels, to be used as the testing material.

Major elements experience significant changes during weathering, making their analysis a crucial method for evaluating the degree of rock weathering (Nesbitt and Young, 1982; Singh, 2009). In Quaternary geological research, major elements are extensively applied to investigate the weathering processes of paleosols (Wang and Miao, 2006; Buggle et al., 2011; Song et al., 2018; Wang et al., 2023). In the studies focusing on riverbed and terrace sediment deposits, the major element is usually applied in analyzing river provenance, source area tracing, regional tectonic evolution, river system evolution, and river pollution (Potter, 1978; Potter, 1994; Sarin et al., 1989; Dupré et al., 1996; Canfield, 1997; Vital and Stattegger, 2000). Few studies have specifically explored the spatiotemporal distribution characteristics of major elements in river terrace sediments. This paper used the major element testing to ancient river sediments in terraces of different ages to estimate their weathering degree. The major element testing was completed at the Key Laboratory of Compound and Chained Natural Hazards Dynamics, Ministry of Emergency Management of China.

First, 0.7 g of powdered rock sample was mixed with 7 g of X-ray fluorescence flux, which consisted of anhydrous lithium tetraborate, anhydrous lithium metaborate, and lithium fluoride. Next, 3–5 drops of lithium bromide solution were added to the mixture. The samples were then preheated in a high-frequency furnace for 6 min, followed by heating to 950°C, where they were held for 7 min. After the sample is melted and mixed evenly, it will become homogeneous glasses after 9 min of self-cooling and air cooling. Then, samples were analyzed on a Panalytical Axios X-ray fluorescence spectrometer. The national standard GSR-1 was used for analysis, and the working conditions were 50 kV and 60 mA. Finally, 0.7 g of the sample was weighed into a crucible, heated in a Mabo furnace, then placed directly into a dryer. After drying, the sample was weighed again, and the loss on ignition was calculated. The loss on ignition refers to the mass lost by a sample when heated to high temperatures, primarily due to the volatilization of volatile substances (such as moisture and carbon dioxide) and the combustion of organic matter (Heiri et al., 2001). In geochemical experiments, the determination of loss on ignition helps to ensure the complete removal of moisture from the sample prior to testing, as well as evaluate the compositional changes resulting from high-temperature exposure (Santisteban et al., 2004).

To quantitatively assess weathering, the Chemical Index of Alteration (CIA) (Nesbitt and Young, 1982; Mclennan, 1993), which was shown to be effective in fluvial sediments (Singh et al., 2005; Caceres et al., 2013), was used to calculate the degree of weathering. The detailed parameters in this formula merit attention, as they are primarily based on the specific experimental results of previous researchers (Nesbitt and Young, 1982). The term CaO* represents the molar content of calcium oxide in silicate minerals, excluding contributions from carbonate and phosphate sources (Nesbitt and Young, 1989; Nesbitt et al., 1996). Given that CaO and Na₂O typically occur in similar proportions in silicates, the calculation of CaO* follows a specific method. If the molar concentration of CaO is lower than that of Na₂O, the value of m(CaO) is equal to m(CaO*). However, if the concentration of CaO exceeds that of Na₂O, then m(CaO*) is set equal to m(Na₂O) (Mclennan, 1993); this correction method is applied in the present study. In addition, several weathering indicators were applied to distinguish terrace samples of varying ages, including CIA, Na/K, A-CN-K diagram, A–CNK–FM diagram, S/10-CM-NK diagram, and Pettijohn classification. CIA = AL 2 O 3 / AL 2 O 3 + CaO * + K 2 O + Na 2 O × 100

Through interpretation and field surveys, we found that the upper reaches of the Qingyi River feature a narrow valley and a steep gradient, with terraces mainly developed in the wider sections of the valley, such as Zhonggang Village, Wulong Town, and Lushan County. In the middle reaches, the valley widens, and the river gradient decreases, allowing terraces to develop in both the wide valley and gorge sections. In the lower reaches, the river flows through a relatively flat basin, where the valley widens significantly. The Qingyi River basin contains six terrace levels, labeled from lowest to highest as T1, T2, T3, T4, T5, and T6. Terraces T1–T4 are extensively distributed across the basin and are primarily sedimentary or piedmont terraces. Terraces T5 and T6, which are mostly located in wider sections of the river valley, include erosional terraces in the upper reaches, piedmont terraces in the middle reaches, and are less common in the lower reaches. The terrace development at key locations along the Qingyi River, from the upper to the lower reaches, is summarized below.

Wulong Town is located in a relatively wide river valley area in the upper reaches of the Qingyi River, where six levels of terraces have developed (Figure 4). Among these, T1 is a sediment terrace, T2 and T3 are piedmont terraces, and T4, T5, and T6 are erosional terraces. The elevation of the terraces ranges from 4 to 5 m (T1), 15–16 m (T2), 48–55 m (T3), 85–90 m (T4), 117–120 m (T5), to 148–155 m (T6). The surface of the T1 terrace is flat, approximately 100–200 m wide, with a sediment thickness of about 4 m. It features a river facies binary structure, with well-rounded gravel and a lithology primarily consisting of limestone and quartzite. The T2 terrace surface is about 20–30 m wide, with a sediment thickness of approximately 3 m. The gravel is moderately rounded, and the lithology is mainly limestone. The T3 terrace surface is about 100 m wide, with a sediment thickness of around 5 m. The gravel is moderately rounded, and the lithology is primarily limestone (Figures 4A, C).

In the Lushan County river valley, located in the upper reaches of the Qingyi River, four levels of terraces are present (Figures 4B, D). The T1–T4 terraces are all piedmont terraces, with elevations ranging from 8 to 13 m (T1), 18–33 m (T2), 85–97 m (T3), and 131–152 m (T4). The T1 terrace forms a narrow strip along both banks of the river, with a sediment thickness of about 4 m and well-rounded gravel. The T2 terrace has a sedimentary layer about 14 m thick, characterized by a typical binary structure, with well-rounded gravel. The T3 terrace has a sediment thickness of 12 m, with well-rounded gravel. The T4 terrace features a sedimentary layer with a visible thickness of 12.7 m, where the gravel, composed of sandstone and mudstone, is weathered and fragmented (Figures 4B, D).

In the Ya’an river valley of the middle reaches of the Qingyi River, six levels of terraces are preserved, making it the section of the Qingyi River basin with the most complete terrace preservation (Figures 5A, C). The terrace elevations are as follows: T1 is 4–8 m, T2 is 15–18 m, T3 is 48–59 m, T4 is 88–92 m, T5 is 120–130 m, and T6 is 201–205 m. The T1 terrace has a sediment thickness of about 5 m, with gravel primarily composed of granite and diorite. The T2 terrace has a sediment thickness of about 11 m, with well-sorted and well-rounded gravel. The T3 terrace has a sediment thickness of about 9 m. The T4 terrace has a sediment thickness of approximately 12 m, with lithology dominated by sandstone and quartzite, and heavily weathered sandstone gravel. The T5 terrace has a sediment thickness of about 18 m, with gravel consisting of quartzite, sandstone, mudstone, and minor igneous rock, showing a high degree of weathering. The T6 terrace has a sediment thickness of about 6 m, with gravel composed of quartzite and mudstone (Figures 5A, C).

In the Hongya Valley, the elevations of the terrace fronts relative to the current riverbed vary as follows (Figures 5B, D): Terrace T1 lies between 6 and 8 m, Terrace T2 between 21 and 25 m, Terrace T3 between 74 and 81 m, and Terrace T4 between 127 and 132 m. These terraces (T1–T4) are distributed across the basin, with T1 and T2 extending throughout the valley. The exposed gravels mainly consist of sandstone and mudstone, with a reddish-brown sand and silt matrix. Terrace T3 has a more limited distribution, characterized by a gravel layer approximately 16 m thick. The gravels in this layer are primarily quartzite, sandstone, and mudstone, with a brown, clayey sand matrix. Terrace T4 is concentrated around the valley’s highest hills, where the gravel thickness above the bedrock strath reaches up to 20 m. The sediments here are predominantly quartzite and mudstone gravels, with diameters ranging from 10 to 30 cm, embedded in a clay-rich matrix (Figures 5B, D).

To evaluate the weathering extent of the terraces, samples were obtained and analyzed for their major element composition. The sampling sites were located beneath the overbank deposits and at the top of the gravel layers, with the matrices filling the interstitial spaces between the gravels also being sampled. The sample locations are shown in Figure 2, and typical profiles for terraces T1–T6 are provided in Figure 6. According to the profiles, the matrix between the gravels of T1 is black-gray, while T2 is brown-gray, T3 is brown, while T4, T5, and T6 are all reddish-brown. These color variations likely indicate the increasing weathering of terrace sediments with age.

In our previously published article (Jiang et al., 2016), we conducted major element testing on the Qingyi River terrace, analyzing 12 samples which were reported in that article. These 12 samples are labeled as m13-m18 and m23-m28 in Table 1. In the current study, we tested the major elements of 33 new samples from the Qingyi River terrace. These samples are labeled as m1-m12, m19-m22, and m29-m45 in Table 1. All these data are listed together in Table 1 for convenience in further data analysis in the following sections. In addition to these results, several representative materials are included for comparison, such as terrigenous shale (Taylor and Mclennan, 1985), world sediments (WS) (Mclennan, 1995), Luochuan loess (Gallet et al., 1996), Xiashu loess (Zhang et al., 2007), western Sichuan loess (Yu et al., 2020), and Xuancheng loess (Hong et al., 2010).

The ratios of average major element concentrations normalized to the upper continental crust content (Taylor and Mclennan, 1985) reveal the variability in elemental distributions across the different terraces (Figure 7). The content of major elements in the samples has the same tendencies, indicating that the samples may have an extensive material source. We also found that the TiO₂ and total content of iron oxides (TFe) in samples were more significant than in the content of upper crust materials (UCC). This may be because of the transport of minerals as fine particles, such as biotite. These minerals are more likely to be suspended in the river and thus to be carried into the terrace sediments (Vital and Stattegger, 2000; Singh et al., 2005). Additionally, from terrace T1 to terrace T6, a gradual decrease occurs in the MgO, CaO, Na₂O, and K₂O contents, which reveals an increasing degree of weathering. As the main rock-forming mineral, the decomposition of feldspar dominates the weathering process. Plagioclase is more easily weathered among the feldspars than orthoclase (Nesbitt et al., 1980). Correspondingly, the weathering process can be divided into three stages: an early stage characterized by the leaching of Na and Ca, a middle stage with the leaching of K, and a late stage marked by the leaching of Si. As the terraces age, the loss of mobile elements becomes more pronounced. Overall, the terrace sediments in the Qingyi River basin appear to have undergone weathering primarily during the early and middle stages (Figure 7).

The CIA index can directly quantify the degree of weathering of the sample (Figure 8). Besides, the Na/K ratio serves as a key indicator for assessing the weathering of plagioclase in the samples and can also provide insights into the overall chemical alteration of the sediments. Plagioclase is characterized by a high concentration of sodium, whereas potassium feldspar, illite, and mica are enriched in potassium. Due to the significantly higher weathering rate of plagioclase compared to potassium feldspar, the Na/K ratio in weathering profiles exhibits an inverse relationship with the degree of weathering. Combining the CIA index and Na/K ratio, we can further deepen our understanding of the sample’s degree of weathering. (Figure 8). For reference, several representative materials are included for comparison alongside the major element analytical results, such as terrigenous shale (Taylor and Mclennan, 1985), world sediments (WS) (Mclennan, 1995) (serving as a global reference for various sediment types), Luochuan loess (Gallet et al., 1996) and Xiashu loess (Zhang et al., 2007) (representing sediments that have undergone significant early-stage weathering), as well as Chuanxi loess (Yu et al., 2020) and Xuancheng loess (Hong et al., 2010) (which are in the intermediate stage of weathering, showing notable potassium loss). It can be seen that Na/K and CIA show an apparent negative correlation, reflecting the trend of increasing weathering degree of river sediments from new to old. Both T1 and T2 are lower than UCC, indicating that due to provenance reasons, sorting, and mixing when the sediments were formed, feldspar was enriched to a certain extent compared to UCC, resulting in higher contents of Ca and Na. In the terraces T4 and T5, the weathering of the terraces reached a level equivalent to that of terrigenous shale, but overall it was less than the weathering of laterite. Only T6 has almost reached the weathering degree of laterite. Since few T6 terraces are preserved, we only collected one sample in the Ya’an River Valley.

The degree of weathering can be further represented by an A–CN–K ternary diagram, in which A denotes Al₂O₃, CN represents CaO + Na₂O, and K stands for K₂O, in combination with the CIA. This diagram serves as a model for illustrating the weathering progression of the UCC. In the early stage of continental weathering, the weathering of anorthosite follows a trend parallel to the line from A to CN, with a direction toward A. This stage typically reflects the weathering process from the UCC to terrigenous shale. As weathering advances to the A–K line, the process enters the middle stage, characterized by the weathering of orthoclase and illite. Subsequently, the weathering trend aligns parallel to the A–K line. In the final stage of continental weathering, the weathering products undergo a silicon runoff process, positioning near point A on the diagram (Figure 9). The data from each terrace sample are plotted on the A–CN–K ternary diagram (Figure 9), alongside the positions of major rock-forming minerals, the UCC, and natural water. It shows the terrace data in the diagram span a broad area (Figure 9), revealing significant differences in the weathering characteristics of terraces of varying ages along the Qingyi River. The weathering trends of T1 and T2 deviate from the UCC and shift towards the natural water direction, suggesting that these younger sediments are enriched in river solutes, with higher calcium and sodium contents compared to the original crust. Conversely, the weathering trends of T3, T4, and T5 align with the UCC weathering pattern, indicating a more consistent progression. The geomorphic surface connections between terraces suggest a clear progression from fresh to older sediments, with pronounced separation and distinct weathering trends.

The A–CNK–FM ternary diagram provides an alternative perspective for examining the weathering trends of river sediments. When the Qingyi River sediment data are projected onto this A–CNK–FM diagram (Figure 10), in which A denotes Al₂O₃, CNK represents CaO + Na₂O+ K₂O, and FM stands for Fe + Mg, they show distinct patterns in comparison to standard materials such as UCC, terrigenous shale, and loess. Specifically, the Qingyi River sediments are generally displaced towards the FM corner, indicating a higher concentration of iron and magnesium oxides. This shift suggests that, throughout the river’s development, fine-grained minerals like iron hydroxides and biotite are more readily transported by flowing water and become suspended in the river sediments. Previous studies of modern rivers and riverbeds have similarly demonstrated this behavior (Vital and Stattegger, 2000; Hong et al., 2010). These suspended fine-grained minerals eventually settle and accumulate in the riverbed, and as the river continues its erosional activity and forms terraces, these minerals are preserved in the sediment layers. The diagram further reveals that as the river sediments age, there is a gradual enrichment of Fe and Mg (Figure 10).

In the S/10-CM-NK projection diagram, which S denotes Si, CM represents CaO + Mg, and NK stands for Na₂O+ K₂O, the young sediments from the Qingyi River are positioned towards the CM vertex, while the older sediments trend towards the S/10 vertex. This pattern reflects the enrichment and loss of quartz in the material composition of the sediments (Figure 11). The overall trend from younger to older terrace samples follows a path parallel to the S/10-CM axis, suggesting that the migration of sodium and potassium remains relatively consistent throughout the weathering process. However, the S/10-CM-NK diagram primarily captures variations in silicon content, and as a result, the early stages of the weathering process are not clearly discernible. Moreover, the diagram does not show a distinct separation between the terraces of different ages (Figure 11).

The chemical indices of major elements, such as the SiO₂/Al₂O₃ ratio, provide insights into the relative abundances of quartz, feldspar, and clay minerals (Potter, 1978). Similarly, the Fe₂O₃/Al₂O₃ ratio offers an indication of mineral stability and can be used for classifying arkosic compositions (Herron, 1988). Furthermore, comprehensive indices are useful for assessing sediment maturity (Pettijohn et al., 2012). In this study, the samples were classified using the Pettijohn lithological classification scheme (Figure 12). The analysis reveals a progression in the elemental characteristics of the terrace sediments, shifting from arkosic and subarkosic compositions to litharenite with increasing age. This trend demonstrates a clear distinction between terraces of varying ages, with older sediments exhibiting greater maturity. However, terraces T1 and T2 are not clearly differentiated within the Pettijohn framework, suggesting that these younger terraces are still in an early stage of weathering.

We applied several weathering indicators to distinguish terrace samples of varying ages, including CIA, Na/K, A-CN-K diagram, A–CNK–FM diagram, S/10-CM-NK diagram, and Pettijohn classification. The CIA and Na/K indicators can distinguish terraces of different levels according to their weathering degree in a more efficient method. Still, the only problem is that the samples of low terraces are relatively discrete. In future work, this issue can be improved by optimizing the methods for selecting sampling targets. In the A-CN-K diagram, these terraces are in the early weathering stage characterized by the removal of Ca and Na. In the age of terrace T6, which is about 300 ka, the sediments begin to shift towards the A–CN axis, marking the transition into the middle weathering stage, where potassium is depleted. The A-CN-K diagram shows a relatively comprehensive superiority in distinguishing the different terraces. The A–CNK–FM diagram highlights a clear separation of samples from different terrace levels, which effectively illustrates the early weathering process, particularly the removal of calcium and sodium. This trend mirrors the weathering progression observed from the UCC to terrigenous shale. However, the elements of iron and magnesium exhibit irregular distribution characteristics in the terrace sediments, making the A–CNK–FM diagram not stable in differentiating terrace samples. The S/10-CM-NK diagram does not show a distinct separation between the terraces of different ages, indicating the terraces samples are not sensitive in the distribution of Si element. The Pettijohn classification appears to be particularly effective for distinguishing between the older terraces rather than young terraces.

Furthermore, we performed a detailed comparison of several weathering proxies with diagrams to assess their sensitivity in distinguishing samples from river terraces of different ages (Figure 13). For the CIA, NA/K, and Pettijohn classification, we directly constructed box plots to compare the data from different terrace samples. For the A-CN-K, A-CNK-FM, and S/10-CM-NK diagrams, due to the challenge of determining the relative weights of the three vertex indices, we applied the geometric mean method to combine the three sets of data and generated box plots for comparison (Nicolli et al., 1989; Burt et al., 2003; Pekey et al., 2004). The result shows that, depending on the specific contextual requirements, the three indices derived from major elements—CIA, Na/K, and the A-CN-K diagram—are highly suitable for distinguishing the weathering degree of terrace samples from different ages. In contrast, the A–CNK–FM diagram, S/10-CM-NK diagram, and Pettijohn classification are more sensitive to terrace sediments older than 100 ka, but less so for younger sediments, making them less effective for terrace classification (Figure 13). Based on the data presented in this paper, we hypothesize that the A–CNK–FM diagram, S/10-CM-NK diagram, and Pettijohn classification indicators may have greater application potential for terraces or ancient alluvial fans on timescales exceeding 100 ka, or even extending to 1 Ma time scale.

The chronological results of the terraces examined in this study have been reported in our previously published work (Jiang et al., 2016; Liu et al., 2015; Jiang et al., 2022). Furthermore, the sampling profiles and locations for the major element analysis in this paper are consistent with those used for age analysis in previous terrace studies. The materials used for major element testing and age determination are identical, consisting of sand or clay filling the spaces between gravels in terrace deposits. For major element analysis, the samples assessed represent the overall major element content of the infill between gravels within the sedimentary strata. In chronological analysis, radiocarbon dating was used to determine the age of organic carbon in the sedimentary layers, rather than the age of discrete carbon particles, while optically stimulated luminescence (OSL) and electron spin resonance (ESR) dating were employed to measure the signal age of quartz particles within the sediment. The compiled terrace age results are shown in Table 2.

We combined the major element data of river terraces in this study with the river chronology data in previous studies to estimate the direct relationship between the weathering degree and age of the terraces (Figure 14). The fitting results of the CIA index and the terrace age show a linear relationship between the two as a whole, and the value of Pearson’s r is 0.99. It shows that within the age range of 300 ka of the T6 terrace, the sediments in the terrace show a linearly increasing weathering trend. In other words, most terraces can be compared by weathering indicators obtained by major elements. In addition, by projecting the CIA index and age of the terrace onto the climate curve, it can be found that the abandonment age of the terrace has a good correspondence with the interglacial period (Figure 14). The deep-sea oxygen isotope (Lisiecki and Raymo, 2005) and the Chinese loess sequence (Ding et al., 2002) were plotted in Figure 11 to show the paleoclimate information. On the deep-sea oxygen isotope curve, T1 corresponds to the Holocene warm period, T2 corresponds to the interglacial MIS3, T3 corresponds to the interglacial MIS5, T4 and T5 correspond to the interglacial MIS7, and T6 corresponds to the interglacial MIS9 (Figure 14). On the magnetic susceptibility curve of the loess sequence in Jingbian, T1 corresponds to the Holocene warm period S0, T2 and T3 correspond to the interglacial period S1, T4 and T5 correspond to the interglacial period S2, and T6 corresponds to the interglacial period S3 (Figure 14). This correspondence shows that during the interglacial period in the Qingyi River Basin, the amount of river water increased and the cutting ability became more powerful, resulting in the abandonment of the riverbed to form terraces. We believe that when the terraces were formed during the interglacial period, they began to experience intense weathering under the environment of heavy precipitation. In the glacial period, precipitation decreased, and weathering weakened. Higher-level terraces often experience one more interglacial weathering process than lower-level terraces, and a new-level terrace develops during the interglacial period. Therefore, this stable terrace formation pattern and weathering process leads to a linear increase in the degree of terrace weathering.

In this study, the height of terraces above the river bed is also considered to influence the weathering process. We collected the terrace’s heights in previous studies and compared them to the sites of major element samples in this study (Jiang et al., 2016; Liu et al., 2015; Jiang et al., 2022). Results show that the CIA index and the terrace’s height above the river seem to own a linear relationship on the plot (Figure 15). According to the calculation, the value of Pearson’s r is 0.97. We think the chemical weathering of the terrace’s sediment is influenced by precipitation more than riverbeds. Once the terrace is abandoned and relatively higher than the riverbed, it is no longer directly affected by the river erosion and transport system. The weathering process of the terraces is more like feedback to ancient climate changes.