基于多维风险模型与机器学习的深型水库沉积物磷释放风险评价及驱动机制

秦赫; 李扬; 张伊; 习刚正; 王超; 张潇月

doi:10.12404/j.issn.1671-1815.2507346

科学技术与工程 ›› 2026, Vol. 26 ›› Issue (18) :7951-7960 DOI: 10.12404/j.issn.1671-1815.2507346

环境科学、安全科学

基于多维风险模型与机器学习的深型水库沉积物磷释放风险评价及驱动机制

秦赫 ¹ ,
李扬 ² ,
张伊 ¹ ,
习刚正 ³ ,
王超 ³ ,
张潇月 ²^,^*

作者信息 +

Risk Assessment and Driving Mechanisms of Sediment Phosphorus Release in Deep-water Reservoirs Based on Multidimensional Risk Model and Machine Learning

QIN He ¹ ,
LI Yang ² ,
ZHANG Yi ¹ ,
XI Gang-zheng ³ ,
WANG Chao ³ ,
ZHANG Xiao-yue ²^,^*

Author information +

文章历史 +

摘要

为评估深型水库沉积物磷释放生态风险,保障南水北调核心水源地丹江口水库水质安全,构建了耦合活性驱动(active driver,AD)-迁移潜力(mobility potential,MP)-环境胁迫(environmental stress,ES)的三维风险指数模型,并联合薄膜扩散梯度技术、高分辨率渗析技术与机器学习,系统揭示了深型水库沉积物磷释放风险的时空格局与多维驱动机制。结果表明,枯水期全域呈低风险,而丰水期深水封闭库湾和入库河口深水区跃升为超高风险核心区。丰水期,较高的外源输入叠加热分层诱发的底层厌氧,增强了AD及ES指数放大效应,形成“缺氧胁迫+外源富集”的双重风险驱动,导致风险升高。随机森林量化显示底层水体溶解氧(DO_bottom)浓度为最核心因子并具有阈值效应,其浓度<3.5 mg/L直接触发沉积物磷释放高风险;而沉积物有效态磷(DGT_P)浓度>0.35 mg/L时,即使好氧条件仍可致中风险。DO_bottom-DGT_P及DO_bottom-ΔT的非线性交互效应是风险升级的核心驱动力,前者主要体现在生化协同作用,后者主要突出物理-化学耦合作用。抑制机制上,浅水区因强混合维持高氧和弱分层,显著削弱MP风险;过渡带则因AD/MP同步降低,形成双重抑制屏障。研究揭示了“地貌封闭性-水深梯度-外源输入”通过级联调控AD/MP/ES的空间机制,为深型水库内源磷精准防控提供科学依据。

Abstract

To assess the ecological risk of phosphorus release from sediments in deep reservoirs and ensure water quality safety in the Danjiangkou Reservoir, the core water source area for the South-to-North Water Diversion Project, a three-dimensional risk index model coupling AD (active driver), MP (mobility potential), and ES (environmental stress) was innovatively developed. By integrating the diffusive gradients in thin-films, high-resolution peeper techniques, and machine learning, this research systematically reveals the spatiotemporal patterns and multidimensional driving mechanisms of sediment phosphorus release risk in deep reservoirs. Results show a low risk across the reservoir during the dry season. In the wet season, however, the deep enclosed bays and the deep inflow river mouth areas shift into ultra-high-risk hotspots. During this period, increased external inputs coupled with anaerobic conditions induced by thermal stratification enhance the amplification of AD and ES indices. This forms a dual risk mechanism characterized by “hypoxia stress and external enrichment,” elevating overall risk. Random forest analysis identifies DO_bottom (bottom dissolved oxygen) as the most critical factor, with a threshold effect observed. High risk of sediment phosphorus release is triggered directly by DO_bottom less than 3.5 mg/L. Notably, concentration of DGT_P>0.35 mg/L can still cause moderate risk even under aerobic conditions. Nonlinear interactions between DO_bottom-DGT_P and DO_bottom-ΔT are identified as core drivers of risk escalation. The former interaction primarily reflects biochemical synergy, while the latter underscores physicochemical coupling. Regarding inhibition mechanisms, shallow areas maintains high DO_bottom and weak stratification due to strong mixing, significantly mitigating MP risk. The transition zone forms a dual inhibitory barrier through the simultaneous reduction of both AD and MP. This study reveals the spatial mechanism whereby “topographic confinement-water depth gradient-external input” regulate AD, MP, and ES through cascade effects, providing a scientific basis for precise control of endogenous phosphorus in deep reservoirs.

Graphical abstract

关键词

沉积物 / 磷释放风险 / 薄膜扩散梯度技术 / 高分辨率渗析技术 / 风险指数模型 / 丹江口水库

Key words

sediment / phosphorus release risk / DGT (diffusive gradients in thin-films) / HR-Peeper (high-resolution peeper) / risk index model / Danjiangkou Reservoir

引用本文

引用格式 ▾

秦赫, 李扬, 张伊, 习刚正, 王超, 张潇月. 基于多维风险模型与机器学习的深型水库沉积物磷释放风险评价及驱动机制[J]. 科学技术与工程, 2026, 26(18): 7951-7960 DOI:10.12404/j.issn.1671-1815.2507346

QIN He, LI Yang, ZHANG Yi, XI Gang-zheng, WANG Chao, ZHANG Xiao-yue. Risk Assessment and Driving Mechanisms of Sediment Phosphorus Release in Deep-water Reservoirs Based on Multidimensional Risk Model and Machine Learning[J]. Science Technology and Engineering, 2026, 26(18): 7951-7960 DOI:10.12404/j.issn.1671-1815.2507346

登录浏览全文

4963

注册一个新账户忘记密码

水库作为关键的水资源调控枢纽,其水质安全关乎供水保障与流域生态健康。富营养化是水库面临的重大环境挑战,而沉积物内源磷释放则是驱动水体磷循环、维持富营养化状态的核心过程,尤其在深水型水库中表现更为复杂^[1-3]。经典湖沼学研究早已确立溶解氧(dissolved oxygen,DO)是控制沉积物-水界面磷行为的关键因子^[4],厌氧条件通过还原性溶解铁锰氧化物削弱其对磷的束缚,导致溶解性活性磷(soluble reactive phosphorus,SRP)释放增强^[5-7]。近期研究进一步揭示,深水水库中强烈的季节性热分层通过物理隔离阻碍底层水体复氧,加剧并延长缺氧期,为内源磷释放创造了有利环境^{[1,8 -9]}。同时,高分辨率原位监测技术(如薄膜扩散梯度技术DGT和高分辨率渗析技术HR-Peeper)的应用,显著深化了对沉积物有效态磷浓度(ρ_DGTP)和孔隙水磷浓度(ρ_SRP)空间异质性及其动态迁移过程的理解^[10-11]。基于此,研究者们尝试构建风险评价模型以量化沉积物磷释放潜力^[12-13]。

尽管取得重要进展,当前对深型水库沉积物磷释放风险的认知仍存在显著局限。首先,现有模型多聚焦单一驱动因子(如缺氧)或静态的“源强”评估,对“地貌封闭性-水文过程(热分层)-外源输入”等多维空间属性如何通过级联效应(如物理驱动、迁移潜力、环境胁迫的协同)动态调控风险格局,缺乏系统性整合与量化表征^[14-15]。其次,关键环境因子及其交互作用对释放风险的阈值效应和非线性放大机制仍不清晰^[16],限制了风险预警的精准性。此外,作为南水北调中线工程核心水源地的丹江口水库,具有库容巨大、水深梯度显著、入库支流多且外源输入空间异质性强等典型深水水库特征^[17-18],其内源磷释放风险格局及驱动机制的系统研究仍相对薄弱,特别是缺乏融合高分辨率原位观测与空间建模的综合评估。

针对上述问题,现聚焦丹江口水库,联合DGT与HR-Peeper,通过不同水文期原位观测、风险评估模型构建及机器学习解析,构建沉积物磷释放多维风险指数模型,实现释放风险全面动态评估;深入揭示丹江口水库沉积物磷释放风险的时空分异规律,识别高风险敏感单元;定量甄别影响沉积物磷释放风险的主导因子及其交互作用的阈值与非线性格局。以期突破传统单一缺氧驱动的认知框架,深化对深水水库内源磷释放复杂驱动链条的理解,更为丹江口水库及类似深水水库的内源磷精准防控提供关键科学依据和空间靶向管理的理论支撑。

1 材料与方法

1.1 研究区概况

丹江口水库(32°36'N~33°48'N,110°59'E~111°49'E)作为南水北调中线工程的核心水源地,是全球规模最大的战略调水水库之一。水库由汉江库区与丹江库区构成,总库容达290.5×10⁸ m³,水域面积1 050 km²,具有防洪抗旱、水资源调配和生态保护等多重功能。研究区地处鄂豫陕三省交界,属北亚热带向暖温带过渡的季风气候区,年均气温15~16 ℃,年均降水量800~1 000 mm(80%集中于6—10月丰水期),雨热同期且多短时强降雨,水土流失风险显著。土地利用空间异质性是影响沉积物磷循环的关键背景,入库河口区受流域农业活动主导,沿岸耕地占比超35%,化肥施用导致外源氮磷负荷密集输入;库湾淹没区(如远河库湾)原为丘陵河谷,淹没后形成枝状水域,沉积物以陆源有机质为主,易积累内源污染物;深水核心区毗邻生态保护区,人类干扰较弱但受热分层效应控制;浅水过渡带毗邻城镇,受生活污水等人类活动影响,沉积物再悬浮频繁。这种“河口-农业、库湾-有机质、深水-分层、浅水-扰动”的空间分异,通过外源输入强度与内源迁移过程的耦合^[19-20],共同影响沉积物磷释放的潜在风险格局。

1.2 样品采集与处理

基于土地利用与水文分区的空间代表性,在丹江口水库布设8个沉积物采样点(S1~S8),如图1所示。S6城镇毗邻区,全年平均水深小于5 m,水力扰动强烈,溶解氧充足;S1和S2平均水深介于20~30 m,属于过渡带;S3、S4、S5、S7和S8平均水深介于30~50 m,属于深水区,底层厌氧持续时间长,热分层效应显著^[18]。

采样于2024年枯水期(3月)和丰水期(9月)实施。使用重力柱状采泥器获取原状沉积柱,同步记录各点水深、水温及溶解氧垂向剖面。沉积柱经PVC管封装后,于4 ℃避光运输至实验室,联合ZrO-AT DGT与HR-Peeper装置(由中国南京智感有限公司提供)进行48 h平衡培养,以高分辨率解析沉积物-水界面有效态磷(DGT_P)和孔隙水中溶解性磷酸盐(SRP),具体实验过程可参考团队先前研究^[21]。

1.3 数据分析

基于深型水库热分层显著、底层长期缺氧及静水环境等特征^[1,22],耦合沉积物-水界面磷释放的物理驱动、迁移过程与环境胁迫状态(主要为溶解氧和温度分层环境因子)^[23-24],构建了沉积物磷释放的三维风险指数(TRI)模型,其核心表达式为

(1)TRI=AD×MP×ES

(2)AD=ln$\left(\frac{\mathrm{S}\mathrm{R}\mathrm{P}\times \mathit{H}}{\mathrm{D}\mathrm{R}{\mathrm{P}}_{\mathrm{w}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{r}}}+1\right)$

(3)MP=$\frac{\mathrm{D}\mathrm{G}\mathrm{T}\mathrm{P}}{\mathrm{S}\mathrm{R}\mathrm{P}}$

(4)ES=exp$\left(2.5-0.5\mathrm{D}{\mathrm{O}}_{\mathrm{b}\mathrm{o}\mathrm{t}\mathrm{t}\mathrm{o}\mathrm{m}}+\frac{\mathrm{\Delta }\mathit{T}}{\mathit{H}}\right)$

式中:H为水深;AD为活性驱动;MP为迁移潜力;ES为环境胁迫;DO_bottom为底层水体溶解氧含量;ΔT为表层与底层水温差;

AD表征沉积物向上覆水释放磷的根本驱动力,该因子主要取决于沉积物孔隙水溶解性活性磷(SRP)与上覆水溶解性活性磷(DRP_water)之间的浓度梯度,并考虑了水深H对水体分层稳定性的强化效应^[25],对数函数形式确保了结果的合理性^[26],常数“+1”避免了分母为0的情况。MP量化了溶解态磷从沉积物孔隙水向上覆水体扩散迁移的能力;DGT_P为有效态磷,当MP>1时,表明磷持续向上覆水迁移的风险较高。ES评估底层水体理化条件对沉积物磷释放/吸附行为的胁迫程度;参考相关文献[16,27],当底层水体溶解氧的浓度ρ_{DO_bottom}<2 mg/L时,磷还原释放剧烈,且较大温差ΔT和水深H也会导致ES值增大,因此采用指数函数放大低氧和强温跃层的胁迫效应;基于生态学理论,常数项2.5确保当底层溶解氧处于好氧基准值5 mg/L左右时,ES约等于1。风险分级参考Wu等^[28],TRI<4为低风险,4≤TRI<8为中风险,TRI≥8为高风险。

1.4 统计分析与数据绘制

数据制图和统计分析主要采用了ArcGIS、RStudio和Origin软件。其中,ArcGIS用于绘制采样图,并基于克里金空间插值可视化丹江口水库沉积物磷释放风险;RStudio用于构建随机森林机器学习方法,识别并量化磷释放关键影响因子及其交互作用;其余图像均采用Origin进行绘制。

2 结果与讨论

2.1 TRI模型参数时空分布特征

丹江口水库沉积物磷释放的三维风险参数(AD、MP、ES)呈现显著的空间梯度与季节分异,如图2所示。枯水期,全域环境胁迫(ES)处于低值区间(0.03~0.05),主要得益于底层溶解氧充足(ρ_{DO_bottom}>10 mg/L)及水温分层微弱(ΔT为0~4 ℃),有效抑制沉积物磷解吸。空间上,库湾与河口区表现出潜在风险积累特征,S2因沉积物有效态磷富集(ρ_DGTP约为0.3 mg/L)致迁移潜力(MP)高达10.02,S3则因水深较深放大孔隙水ρ_SRP梯度(约为0.1 mg/L),活性驱动(AD)达5.87。而浅水开阔区(S6)受水体强混合影响,AD与MP均受抑制,ES维持较低水平。

丰水期,热分层与缺氧驱动参数格局重构。深水区ES呈指数级响应,尤以厌氧库湾S8为典型,其ρ_{DO_bottom}由枯水期11 mg/L降至3 mg/L,叠加ΔT约为4 ℃,ES值急升至3.61(约为枯水期的80倍),触发指数放大效应。入库河口区(S3)虽MP略降至0.98,但外源输入致ρ_SRP倍增(0.2 mg/L),AD升至4.31,耦合较高ES值,形成高风险组合。李燕妮等^[29]同样强调了河口区沉积物磷释放高潜力。与之对比,过渡带S1呈参数协同下降趋势,AD从枯水期4.39降至3.03,MP从2.05降至1.63,同时ES仅微增至0.24(温度弱分层效应导致),三者共同强化低风险屏障。浅水区(S6)因垂向温差微弱,及ρ_{DO_bottom}高于7 mg/L,ES稳定于0.24,即使MP因DGT_P升高达3.63(枯水期0.97),AD仅0.80,风险仍受控。

参数耦合机制的空间归因表明,深水库湾(如S4、S8)的高风险源于“地貌封闭性限制水体交换促进分层缺氧,诱发ES指数响应”与“外源输入推升AD”的协同作用;过渡带则因则因AD/MP同步下降与低ES,形成双重抑制机制。浅水开阔区(S6)的强混合作用则通过抑制ES形成“低胁迫屏障”,显著削弱高MP的潜在风险。值得注意的是,ES>1.0的阈值由ρ_{DO_bottom} <5 mg/L与ΔT>4 ℃共同触发,验证了环境胁迫因子的非线性敏感性,与Yan等^[30]研究结论一致。沉积物磷释放风险的空间分异由水深梯度(控制AD/ES响应强度)与地貌封闭性(主导MP分异)共同塑造;季节性剧变则主要依赖丰水期深水区DO_bottom与ΔT对ES的协同放大^[31-32]。未来需警惕过渡带在气候变暖下的潜在敏感性,若分层加剧导致DO_bottom持续下降,其ES抑制效应可能失效。

2.2 沉积物磷释放风险评估

基于TRI模型计算沉积物磷释放风险空间插值,如图3所示,结果表明释放风险呈现显著时空分异。枯水期所有点位TRI均低于<4,但空间差异明显。浅水开阔区S6风险最低(TRI=0.08),过渡带S1(TRI=0.26)与深水区S3、S5和S8(TRI分别为0.38、0.32和0.31)风险相近且较低,而入库河口区S2(TRI=1.55)与库湾区S4和S7(TRI分别为1.28和1.05)相对较高,反映了外源输入主导区域在枯水期的潜在风险积累。至丰水期,风险格局发生剧烈重构,深水封闭库湾S8(TRI=24.26)和入库河口深水区S4(TRI=12.40)跃升为超高风险双核心,与水库总磷时空污染特征相似^[17]。入库河口区S3(TRI=5.99)进入中风险等级,库湾区S7(TRI=3.05)、河口区S2(TRI=3.45)、深水区S5(TRI=2.51)、过渡带S1(TRI=1.19)及浅水区S6(TRI=0.70)均维持低风险,但S7与S2接近低风险阈值上限。Xu等^[33]研究同样强调了丹江口水库丰水期由于高流量、热分层和磷的成岩作用等因素,沉积物磷释放量更高。

风险空间分异主要受环境过程与空间属性的耦合作用驱动。深水封闭库湾S8的超高风险源于其枝状地貌对水体交换的阻滞作用,丰水期强热分层诱发底层厌氧^[34],触发ES指数级响应;同时库湾滞留的外源颗粒磷与有机质降解推升SRP^[35],增强活性驱动AD,并维持陡峭的沉积物-水界面磷扩散梯度,形成“封闭地貌-分层缺氧-外源富集”的三元胁迫耦合。入库河口深水区S4的高风险则体现外源输入与深水环境的协同效应,支流携带的高负荷颗粒磷沉降扩容活性磷库^[29],显著提升SRP与AD;同时,水深促进的强热分层加剧底层缺氧^[8],驱动ES飙升;加之MP的扩散助力,共同导致风险暴发。与之对比,同属外源输入热点的S3,虽然AD与S4相当,但因MP和ES较低,属于中风险区域。库湾区S7则因MP骤降,显著削弱了ES胁迫效应,风险维持在低水平。

低风险区的稳定性源于特定空间属性形成的抑制机制。浅水开阔区S6强水力扰动抑制分层,维持底层高氧,其高MP被低AD和低ES共同压制,形成“低胁迫屏障”。过渡带S1和河口区S2因水深较浅缓冲分层效应,ES抑制风险升级。深水区S5虽ΔT高达8.2 ℃,但ρ_{DO_bottom}达7 mg/L,维持 ES=0.46,使AD与MP的组合风险可控。

综上,沉积物磷释放风险的空间分异本质是“地貌封闭性-水深梯度-外源输入”三维属性通过调控AD(活性驱动)、MP(迁移潜力)、ES(环境胁迫)参数形成的级联效应。深水封闭库湾与外源输入深水河口是两类主要的超高风险敏感区,其核心机制为空间属性诱导的“ES指数响应”与“AD/MP”提升的双重耦合。浅水强混合区则通过抑制ES构筑天然屏障。未来需重点监控库湾与河口深水区的溶解氧衰减与强分层事件,防范丰水期环境胁迫与活性迁移潜力的协同爆发。

2.3 关键因子识别及其交互作用机制

2.3.1 沉积物磷释放主导因子及因子对识别

为深入解析驱动丹江口水库沉积物磷释放风险的内在机制,基于随机森林,定量评估了TRI模型中各独立因子的单变量贡献度及其双因子交互效应,如图4所示。单因子分析揭示,DO_bottom是调控沉积物磷释放风险的最核心环境胁迫因子,其贡献率显著高于其他参数。这一发现与经典湖沼学理论高度吻合,即厌氧条件会破坏沉积物中铁/锰氧化物对磷的结合能力,并促进有机质矿化释放溶解态磷^[36-37],显著降低沉积物对磷的吸附/滞留阈值。其次为DGT_P,其作为直接量化沉积物-水界面有效扩散通量的指标,代表了沉积物向上覆水体供给磷的“源强”潜力。当DGT_P较高时,表明沉积物孔隙水附近存在充足的溶解态磷储备,一旦环境条件(如缺氧)触发解吸或扩散屏障减弱(如低DRP_water),磷向上覆水迁移的驱动力(MP)便显著增强,风险随之升高。DRP_water作为浓度梯度的另一端,其贡献度位列第三,较低的DRP_water值会增大SRP与其之间的化学势梯度,强化扩散驱动力(AD)。水温分层强度作为物理驱动因子,其贡献度高于孔隙水SRP和水深。ΔT通过强化水体热分层稳定性,阻碍上下层水体混合,导致底层水体复氧困难,从而间接加剧缺氧程度,并放大缺氧对磷释放的促进作用,与张雪瑞等^[38]对深型水库(密云水库)沉积物磷释放影响因子研究结论一致。值得注意的是,水深在本研究中的单因子贡献度相对最小,这主要源于丹江口水库作为典型深水水库,各采样点水深普遍较大且差异相对其他因子不显著(尤其在深水区),其效应已部分融入环境胁迫(ES)和活性驱动(AD)模型中。

然而,沉积物磷释放是一个高度复杂的非线性过程,单因子效应往往难以完全解释风险格局。双因子交互效应分析[图4(b)]揭示了更为深刻的驱动机制,结果显示,DO_bottom与DGT_P的交互作用对TRI变异的解释力最强。这意味着当底层水体处于低氧/厌氧状态时,沉积物中高水平的有效态磷会协同放大磷的释放风险。其驱动机理主要在于,缺氧环境解除了沉积物对磷的化学束缚(主要针对铁结合磷)^[39-40],而此时若沉积物界面附近存在高DGT_P,则被“解放”的磷可迅速形成显著的扩散梯度,驱动磷持续、大量地向上覆水体释放。这一交互作用是导致如S8、S4等库湾区在丰水期出现超高TRI值的核心机制。特别值得关注的是,DO_bottom与水温分层强度ΔT的交互作用(DO-ΔT)对风险的影响强度甚至超过了DO_bottom与DRP_water的交互作用,这突显了物理过程在诱发和加剧化学过程(缺氧驱动的磷释放)中的关键桥梁作用^[30,41]。强烈的热分层不仅直接造成底层水体物理性隔离,阻碍溶解氧补充,还会通过抑制垂向混合使底层缺氧状态持续恶化^[42]。DO降低与ΔT增大形成正反馈循环,经ES项正向放大,对环境胁迫产生了超线性增强作用,致使风险水平急剧升高(如S8在丰水期ES值激增80倍)。相比之下,DO_bottom与DRP_water交互作用相对较弱,可能因为DRP_water更多是磷释放的结果或梯度平衡的反映者,其动态变化在驱动机制链的因果关系序列中位于相对下游。

2.3.2 影响因子交互作用机制

为阐明关键因子对间的交互作用如何决定风险等级,进一步构建了交互作用响应面(图5),探究不同因子组合下的风险临界阈值与非线性格局。DO_bottom被证实为风险等级的最核心要素,而DGT_P与DO_bottom交互作用具有风险放大效应。当ρ_{DO_bottom}>5.5 mg/L时,无论ΔT与SRP处于何种水平,TRI值普遍稳定在低风险区间,与Lu等^[43]得出的6.5 mg/L的溶解氧浓度阈值较接近。这表明充足的好氧环境是抑制沉积物磷释放最有效的“屏障”。然而,当ρ_DGTP异常高(>0.35 mg/L)时,即使DO_bottom处于好氧范围,DO_bottom与DGT_P的强交互作用也能将风险推升至中风险。这反映了极高的沉积物有效磷“库存”即使在有氧条件下也可能构成潜在威胁。相反,ρ_{DO_bottom}下限阈值为3.5 mg/L,低于该阈值时,风险格局发生根本性转变,此时无论其他因子状态如何,TRI值均跃升至高风险等级。这证实缺氧作为触发大规模内源磷释放的充分条件^[44]。在ρ_{DO_bottom}介于3.5~5.5 mg/L(过渡或弱氧状态)的区域,风险等级通常处于中风险。然而,DGT_P在此范围具有风险放大作用,当ρ_DGTP超过0.35 mg/L时,两者交互作用会显著增强,导致TRI值突破中风险上限,进入高风险等级。这表明在氧浓度不再能提供充分保护的过渡状态下,沉积物中富集的有效态磷成为决定风险是否升级的关键驱动力。

DO_bottom与ΔT交互作用的热力图特征与库内高风险点的空间分布高度一致。这些点位均表现出低ρ_{DO_bottom}(<4 mg/L)与显著ΔT(>4 ℃)的耦合特征。强烈的热分层通过物理机制加剧和维持了底层缺氧^[22,32],两者协同作用通过ES项的指数放大,对环境胁迫产生倍增效应。相比之下,开阔浅水区即使存在一定的有效磷,但由于缺乏显著的分层且DO充足,DO_bottom与ΔT交互作用微弱,风险始终被有效抑制在低水平。这也从交互作用角度解释了为何深水封闭库湾是内源释放的高敏感区。

综上所述,DO_bottom是控制沉积物磷释放风险等级的绝对主导因子和基础阈值开关。然而,风险水平则强烈依赖于关键因子间的交互作用。DO_bottom与DGT_P交互作用,主要调控了在给定氧条件下沉积物“源强”潜力对释放风险的放大程度。而DO_bottom与ΔT交互作用,则体现了物理分层通过加剧并维持缺氧,以非线性方式放大环境胁迫,最终主导了深水区(如库湾、河口深槽)在丰水期的风险激增。这些非线性交互机制,为实施更精准的空间靶向预警和治理策略提供了坚实的科学依据。

3 结论

(1)构建了耦合“活性驱动(AD)-迁移潜力(MP)-环境胁迫(ES)”的沉积物磷释放三维风险指数模型,实现了对深型水库沉积物磷释放风险的系统评估与空间解析。证实深水封闭库湾与外源输入型河口为超高风险敏感单元,而浅水强混合区因抑制ES形成天然屏障。

(2)通过机器学习方法明确了关键因子的阈值效应与非线性交互机制。DO_bottom是风险等级的绝对主导因子并存在阈值效应,ρ_{DO_bottom}<3.5 mg/L直接触发高风险,ρ_{DO_bottom}>5.5 mg/L时普遍低风险。DO_bottom与ΔT交互通过“热分层-缺氧正反馈-ES指数放大”(S4的ES激增80倍)主导深水区风险剧增。然而,DO_bottom与DGT_P的交互则揭示了“缺氧解磷束缚-高库存释放”链式反应机制,即使在好氧条件下,高ρ_DGTP (>0.35 mg/L) 仍可引发中风险,突破传统缺氧单一驱动认知。

(3)基于敏感单元识别与驱动机制解析,需重点监控深水封闭库湾(S8类)与入库河口深水区(S4类)的丰水期溶解氧衰减及强分层事件,阻断“外源输入推升AD”与破坏“缺氧-分层正反馈”并重,是管理沉积物磷释放风险的核心过程。

参考文献

原文顺序 | 出版日期 | 本文引用

[1]	Sun C, Wang S, Wang H, et al. Internal nitrogen and phosphorus loading in aseasonally stratified reservoir: implications for eutrophication management of deep-water ecosystems[J]. Journal of Environmental Management, 2022, 319: 115681.

[2]	马煜萱, 包峻松, 洪坤钰, 等. 春灌期阅海水体的氮磷分布特征及富营养化趋势[J]. 科学技术与工程, 2025, 25(2): 850-861.

[3]	Ma Yuxuan, Bao Junsong, Hong Kunyu, et al. Distribution characteristics of nitrogen and phosphorus and the eutrophication trend of Yuehai Lake during the spring irrigation period[J]. Science Technology and Engineering, 2025, 25(2): 850-861.

[4]	Zhuo T, He L, Chai B, et al. Micro-pressure promotes endogenous phosphorus release in a deep reservoir by favouring microbial phosphate mineralisation and solubilisation coupled with sulphate reduction[J]. Water Research, 2023, 245: 120647.

[5]	Mermillod-Blondin F, Gautreau E, Pinasseau L, et al. Interactions between sediment characteristics and oxygen conditions at the sediment-water interface of reservoirs: influences on nutrient dynamics and eutrophication[J]. Hydrobiologia, 2024, 851(14): 3433-3452.

[6]	Cai O, Xiong Y, Yang H, et al. Phosphorus transformation under the influence of aluminum, organic carbon, and dissolved oxygen at the water-sediment interface: a simulative study[J]. Frontiers of Environmental Science & Engineering, 2020, 14(3). DOI: 10.1007/s11783-020-1227-z.

[7]	Chen M, Ding S, Wu Y, et al. Phosphorus mobilization in lake sediments: experimental evidence of strong control by iron and negligible influences of manganese redox reactions[J]. Environmental Pollution, 2019, 246: 472-481.

[8]	张易凡, 吴文强, 刘晓波, 等. 长江上游水库沉积物中磷赋存特征及其迁移转化机制的研究进展[J]. 环境生态学, 2025, 7(5): 41-51.

[9]	Zhang Yifan, Wu Wenqiang, Liu Xiaobo, et al. Research progress on the characteristics of phosphorus occurrence and its migration and transformation mechanism in reservoir sediments of the upper reaches of the Yangtze River[J]. Environmental Ecology, 2025, 7(5): 41-51.

[10]	周荧, 孙景昱, 刘建刚, 等. 温度和采样深度对于静态培养法测定太湖沉积物磷释放的影响[J]. 环境工程学报, 2025, 19(5): 1074-1081.

[11]	Zhou Ying, Sun Jingyu, Liu Jiangang, et al. Effects of temperature and sampling depth on phosphorus release from Lake Taihu sedi-mentsdetermined by static incubation[J]. Chinese Journal of Environmental Engineering, 2025, 19(5): 1074-1081.

[12]	Sano K, Ueda J, Hafuka A, et al. Development of a novel in-sediment passive sampler for profiling orthophosphate and internal phosphorus release nearthe sediment-water interface in a eutrophic lake[J]. Water Research, 2025, 282: 123634.

[13]	Hu M, Yan R, Ni R, et al. Coastal degradation regulates the availability and diffusion kinetics of phosphorus at the sediment-water interface: mechanisms and environmental implications[J]. Water Research, 2024, 250: 121086.

[14]	Wu R, Shi X, Kang X, et al. Characteristics of nitrogen and phosphorus migration at sediment-water interface in seasonal frozen lakes and the mechanism of microbial driven cycling: a case study of Lake Daihai[J]. Journal of Contaminant Hydrology, 2025, 273: 104598.

[15]	Young T, Hargiss C L M, Daigh A, et al. A high degree of phosphorus sediment saturation in a shallow reservoir system[J]. Discover Water, 2024, 4(1). DOI: 10.1007/s43832-024-00067-z.

[16]	Li Y, Liu Y, Wang H, et al. In situ remediation mechanism of internal nitrogen and phosphorus regeneration and release in shallow eutrophic lakes by combining multiple remediation techniques[J]. Water Research, 2023, 229: 119394.

[17]	张一泉, 王卫星, 次珍, 等. 南四湖菹草生长周期对内源磷释放的影响[J]. 环境科学学报, 2024, 44(8): 370-382.

[18]	Zhang Yiquan, Wang Weixing, Ci Zhen, et al. The effect of growth cycle of Potamogeton crispus on internal phosphorus release in Nansihu Lake[J]. Acta Scientiae Circumstantiae, 2024, 44(8): 370-382.

[19]	Wang F, Wang J, Cao T, et al. Seasonal hypoxia enhances sediment iron-bound phosphorus release in a subtropical river reservoir[J]. The Science of the total environment, 2024, 936: 173261.

[20]	Labrie R, Hupfer M, Lau M P. Anaerobic duration predicts biogeochemical consequences of oxygen depletion in lakes[J]. Limnology and Oceanography Letters, 2023, 8(4): 666-674.

[21]	尹炜, 王超, 王立, 等. 丹江口水库总磷时空分布特征及其影响因素[J]. 人民长江, 2023, 54(1): 1-7, 10.

[22]	Yin Wei, Wang Chao, Wang Li, et al. Spatial and temporal distribution characteristics and influencing factors of total phosphorus in Dajiangkou Reservoir[J]. Yangtze River, 2023, 54(1): 1-7, 10.

[23]	邓鑫龙, 陈端, 石浩洋, 等. 基于原型观测数据的丹江口水库水温时空分布特征[J]. 长江科学院院报, 2026(3): 71-78, 87.

[24]	Deng Xinlong, Chen Duan, Shi Haoyang, et al. Spatiotemporal distribution characteristics of water temperature in Danjiangkou Reservoir based on the prototype observation data[J]. Journal of Yangtze River Scientific Research Institute, 2026(3): 71-78, 87.

[25]	刘付立, 黄玄敏, 潘璐, 等. 丹江口水库洪水期水质变化特征及影响因素分析[J]. 人民长江, 2024, 55(S2): 33-37.

[26]	Liu Fuli, Huang Xuanmin, Pan Lu, et al. Analysis of water quality variations and influencing factors during the flood season in the Danjiangkou Reservoir[J]. Yangtze River, 2024, 55(S2): 33-37.

[27]	刘伟, 周斌, 王丕波, 等. 沉积物再悬浮氮磷释放的机制与影响因素[J]. 科学技术与工程, 2020, 20(4): 1311-1318.

[28]	Liu Wei, Zhou Bin, Wang Pibo, et al. Mechanism and influencing factors of nitrogen and phosphorus release via sediment re-suspension[J]. Science Technology and Engineering, 2020, 20(4): 1311-1318.

[29]	张潇月, 刘杰, 夏佩琦, 等. 深型水库沉积物-水界面磷释放时空特征与驱动机制: 以丹江口水库为例[J]. 环境科学, 2025. DOI: 10.13227/j.hjkx.202504090.

[30]	Zhang Xiaoyue, Liu Jie, Xia Peiqi, et al. Spatiotemporal characteristics and driving mechanisms of phosphorus release from the sediment-water interface in deep-water reservoirs: a case study of Danjiangkou Reservoir[J]. Environmental Science, 2025. DOI: 10.13227/j.hjkx.202504090.

[31]	Lü Y, Zhang M, Yin H. Phosphorus release from the sediment of a drinking water reservoir under the influence of seasonal hypoxia[J]. Science of the Total Environment, 2024, 917: 170490.

[32]	沈逸伦, 李财, 孙全, 等. 典型城市入湖河流沉积物磷的时空分布规律与释放风险研究[J]. 环境科学学报, 2025, 45(4): 231-239.

[33]	Shen Yilun, Li Cai, Sun Quan, et al. Spatiotemporal distribution and release risk of phosphorus in sediments of typical urban into-lake rivers[J]. Acta Scientiae Circumstantiae, 2025, 45(4): 231-239.

[34]	Qin B, Zhou J, Elser J J, et al. Water depth underpins the relative roles and fatesof nitrogen and phosphorus in lakes[J]. Environmental Science & Technology, 2020, 54(6): 3191-3198.

[35]	Zhang H Boegmanl, Scavia D, et al. Spatial distributions of external and internal phosphorus loads in Lake Erie and their impacts on phytoplankton and water quality[J]. Journal of Great Lakes Research, 2016, 42(6): 1212-1227.

[36]	Wang J, Chen J, Guo J, et al. Combined Fe/P and Fe/S ratios as a practicable index for estimating the release potential of internal-P in freshwater sediment[J]. Environmental Science and Pollution Research, 2018, 25(11): 10740-10751.

[37]	朱伟, 侯豪, 刘环, 等. 太湖底泥磷释放量及释放规律的研究综述[J]. 湖泊科学, 2025, 37(1): 14-35.

[38]	Zhu Wei, Hou Hao, Liu Huan, et al. A review of sediment phosphorus release amount and rules in Lake Taihu[J]. Journal of Lake Sciences, 2025, 37(1): 14-35.

[39]	Wu Z, Wang S, Ji N. Phosphorus (P) release risk in lake sediment evaluated by DIFS model and sediment properties: a new sediment P release risk index (SPRRI)[J]. Environmental Pollution, 2019, 255: 113279.

[40]	李燕妮, 朱梦圆, 权秋梅, 等. 三峡水库支流沉积物磷赋存特征及其对水质的影响[J]. 环境科学, 2025(12). DOI: 10.13227/j.hjkx.202412013.

[41]	Li Yanni, Zhu Mengyuan, Quan Qiumei, et al. Distribution and water quality impact of sediment phosphorus fractions in tributaries of the Three Gorges Reservoir[J]. Environmental Science, 2025(12). DOI: 10.13227/j.hjkx.202412013.

[42]	Yan C, Xia R, Chen Y, et al. Endogenous phosphorus release from plateau lakes responds significantly to temperature variability over the last 50 years[J]. Journal of Environmental Management, 2024, 371: 123259.

[43]	朱兴林, 司建华, 王军德, 等. 疏勒河流域生态环境质量评价及驱动机制[J]. 科学技术与工程, 2025, 25(25): 10964-10973.

[44]	Zhu Xinglin, Si Jianhua, Wang Junde, et al. Evaluation of ecological environment quality and driving mechanism in Shule River Basin[J]. Science Technology and Engineering, 2025, 25(25): 10964-10973.

[45]	王兴菊, 于文晴, 赵华青, 等. 大冶水库夏秋季热分层对沉积物氮磷释放的影响[J]. 环境科学与技术, 2023, 46(10): 42-52.

[46]	Wang Xingju, Yu Wenqing, Zhao Huaqing, et al. Effect of thermal stratification on nitrogen and phosphorus release from sediments in Daye Reservoir[J]. Environmental Science & Technology, 2023, 46(10): 42-52.

[47]	Xu Z, Yu C, Sun H, et al. The response of sediment phosphorus retention and release to reservoir operations: numerical simulation and surrogate model development[J]. Journal of Cleaner Production, 2020, 271: 122688.

[48]	Zhang Y, Gao X, Sun B, et al. Oxygen evolution and its drivers in a stratified reservoir: a supply-side perspective for informing hypoxia alleviation strategies[J]. Water Research, 2024, 257: 121694.

[49]	胥梦琦, 杨雪, 徐剑波, 等. 沉积物团聚体稳定性和总磷释放特征[J]. 科学技术与工程, 2022, 22(33): 14959-14966.

[50]	Xu Mengqi, Yang Xue, Xu Jianbo, et al. Stability of sediment aggregates and the release characteristics of total phosphorus[J]. Science Technology and Engineering, 2022, 22(33): 14959-14966.

[51]	宋丽媛, 高帅帅, 范献方, 等. 太湖梅梁湾水华生消对沉积物磷迁移转化的影响[J]. 环境科学学报, 2024, 44(11): 184-192.

[52]	Song Liyuan, Gao Shuaishuai, Fan Xianfang, et al. Effect of outbreak and extinction of algal bloom on the migration and transformation of phosphorus in sediment of Meiliang Bay, Lake Taihu[J]. Acta Scientiae Circumstantiae, 2024, 44(11): 184-192.

[53]	Zhang S, Zeng Y, Liu R, et al. Impact of varying dissolved organic carbon load on sediment phosphorus release and its periodic mechanisms[J]. Environmental Research, 2024, 259: 119558.

[54]	张雪瑞, 郭一玮, 王晓燕, 等. 密云水库沉积物-水界面磷交换特征及其对环境因子的响应[J]. 湖泊科学, 2025, 37(3): 835-849.

[55]	Zhang Xuerui, Guo Yiwei, Wang Xiaoyan, et al. Characteristics of phosphorus exchange at sediment-water interface and its response to environmental factors water interface and its response to environmental factors in Miyun Reservoir[J]. Journal of Lake Sciences, 2025, 37(3): 835-849.

[56]	宁一泓, 王海燕, 侯文宁, 等. 三峡库区笋溪河沉积物污染及其氮磷营养盐释放特征[J]. 应用与环境生物学报, 2025, 31(2): 208-220.

[57]	Ning Yihong, Wang Haiyan, Hou Wenning, et al. Characteristics of sediment pollution and nitrogen and phosphorus release in the sediment overlying water interface in the Sunxi River in the Three Gorges Reservoir Area[J]. Chinese Journal of Applied and Environmental Biology, 2025, 31(2): 208-220.

[58]	陈婷, 梁启斌, 王艳霞, 等. 沉积物-上覆水界面有效铁浓度对内源磷再移动的影响[J]. 环境科学研究, 2023, 36(10): 1937-1945.

[59]	Chen Ting, Liang Qibin, Wang Yanxia, et al. Effect of DGT-Labile Fe on endogenous phosphorus remobilization at the sediment-water interface[J]. Research of Environmental Sciences, 2023, 36(10): 1937-1945.

[60]	Cai Y, Wang H, Zhang T, et al. Seasonal variation regulate the endogenous phosphorus release in sediments of Shijiuhu Lake via water-level fluctuation[J]. Environmental Research, 2023, 238: 117247.

[61]	王斌, 黄廷林, 陈凡, 等. 亚热带水库水质特征及沉积物内源污染研究[J]. 中国环境科学, 2021, 41(10): 4829-4836.

[62]	Wang Bin, Huang Yanlin, Chen Fan, et al. Water quality characteristics and sediments endogenous pollution of subtropical stratified reservoir[J]. China Environmental Science, 2021, 41(10): 4829-4836.

[63]	Lu J, Ma T, Liu T, et al. Study on sediment phosphorus release mechanism of the environmental Impact in Dahekou Reservoir[J]. Environmental Science & Technology, 2017, 40(7): 72-78.

[64]	Shou C Y, Yue F J, Zhou B, et al. Chronic increasing nitrogen and endogenous phosphorus release from sediment threaten to the water quality in a semi-humid region reservoir[J]. Science of the Total Environment, 2024, 931: 172924.