KMS XINJIANG INSTITUTE OF ECOLOGY AND GEOGRAPHY，CAS
| 基于贝叶斯网络的绿洲水资源综合管理研究 | |
| 薛杰 | |
| Subtype | 博士 |
| Thesis Advisor | 雷加强 |
| 2017-05-01 | |
| Degree Grantor | 中国科学院大学 |
| Place of Conferral | 新疆乌鲁木齐 |
| Degree Discipline | 理学博士 |
| Keyword | 绿洲 贝叶斯网络 不确定性分析 综合水资源管理 跨学科研究 Oasis Bayesian Networks Uncertainty Analysis Integrated Water Resource Management Interdisciplinary Research |
| Abstract | 绿洲作为干旱区独特的自然与人文景观，是动植物生存、人类定居的重要场所。其中，水资源一直是制约其存在、稳定以及发展的首要自然因素，并与特定的人文、经济以及水管理政策等社会因素共同决定着绿洲水资源利用的可持续性。忽略任一因素都不能通过水资源管理研究有效地解决绿洲生态环境与人类活动系统间的用水短缺与冲突，更不能为干旱区绿洲用水安全与可持续发展提供科学支撑。将水文、生物物理、生态环境等自然因素与人文、经济、用水政策等社会因素进行耦合的跨学科研究，已经成为国际水资源综合管理研究的前沿问题。在干旱绿洲区，如何以水为主线，开展绿洲“水资源-生态环境-社会经济”耦合系统的跨学科水资源综合管理研究，不仅是干旱区绿洲可持续发展面临的现实问题，也是水资源管理科学基础研究的热点问题。本论文选取西北干旱区塔里木盆地南缘的策勒绿洲为典型研究区，以“绿洲水资源来源分析—绿洲生态需水量估算—权衡与协调绿洲农业灌溉与生态需水间用水冲突—各需水部门利益相关者参与下的绿洲水资源综合管理”为研究思路，针对解决各个环节中出现的不确定性因素而开展相应的绿洲水资源综合管理研究。首先，重建策勒河流域的气象数据，通过建立水文模型模拟径流并分析其不确定性，探讨气候变化背景下的绿洲区地表水来源变化与不确定性特征；其次，拓展支持绿洲生态系统服务功能的生态需水量概念，利用最大值与兼容性原理估算其绿洲生态需水量；再次，提出权衡与协调系统，通过建立离散与混合贝叶斯网络模型，解决农业灌溉与生态需水间的用水冲突；最后，征求不同用水部门利益相关者与水管理专家的观点，构建绿洲生态系统服务功能综合于水资源管理的参与式贝叶斯网络模型，提出不同管理情景下的策勒绿洲水资源管理方案和建议。通过研究得到以下主要结论：（1）改进的 Delta 统计降尺度方法在多数据源下提高了气象数据的重建精度。利用多数据集合理论有效地减小了径流模拟的不确定性。根据气候相似性原理，利用改进的 Delta 方法和昆仑山北坡的 12 个参考气象站数据对策勒河流域高山区喀尔塔什站和策勒国家站的气温、降水以及蒸发数据的重建表明，相比重建的月降水量数据，重建的月平均气温和月蒸发量数据精度相对较高，这说明了连续时空场相对于离散时空场更容易重建。利用重建的气象数据作为 TL-FNN-BP (9, 1, 1) 水文模型输入数据获得较为理想的径流模拟结果（NSCE > 0.80）。尽管利用多气象数据集合理论能有效减小径流模拟的不确定性，但由于重建数据不确定性与模型结构不确定性二者的叠加，径流的模拟仍然具有一定的不确定性。 （2）量化的绿洲生态需水量占径流量的 40.29%-58.75% 之间。在各要素生态需水中，荒漠地下水恢复量占最大比例。根据绿洲生态系统特征与功能，拓展了支持干旱区绿洲生态系统服务功能生态需水量的概念，包括保障河流生态系统健康的最小流量、天然绿洲生态系统各要素需水量，以及荒漠地下水位恢复量。计算表明，策勒绿洲所需的生态需水量在最大、中等、最小三等级中分别估算为 0.752×108，0.619×108 和 0.516×108m3，占河流径流量的 58.75%，48.36% 和 40.29%。在各要素生态需水中，荒漠地下水恢复量占最大比例，在最大、中等、最小三等级中分别占绿洲生态需水量的 48.27%, 42.32% 和 37.03%。（3）提出的离散贝叶斯网络权衡系统耦合了绿洲水文和生态需水等自然因素与节水措施和灌溉方式等社会因素，有效地决策了最适绿洲生态需水量。根据离散贝叶斯网络决策系统分析，策勒绿洲最适生态需水量分别在丰水年、平水年和枯水年应为策勒河径流量的 50.24%，49.71% 和 48.73%。其中，在滴灌和节水工程管理下，分别在丰水年、平水年以及枯水年能够节省策勒河径流量的 1.93%，0.66% 和 0.43% 用于绿洲生态需水。（4）建立的混合贝叶斯网络协调系统利用动态离散化算法模拟了主要农作物经济补偿标准，更有效地协调了绿洲生态需水与农业灌溉间用水冲突。通过混合贝叶斯网络模拟结果分析，农作物经济补偿取决于可利用的供水和满足不同程度的生态需水量。在枯水年，保障“中等”程度的绿洲生态需水量对小麦、玉米、棉花以及大枣的经济补偿标准应该分别为 1995.90, 1675.42, 3239.76以及 3513.12 RMB/ha。同时，采用先进的节水工程设备与滴灌技术能分别节省冬小麦、夏玉米、棉花以及大枣 247，271，246 以及 237 RMB/ha 的经济补偿标准。（5）构建的参与式贝叶斯网络将生态系统服务功能综合于水资源管理系统中，成功地模拟了水资源管理情景下的结果与效应。通过参与式贝叶斯网络的构建与模拟，目前任何管理模式都不能很好地保持策勒绿洲的可持续水发展。建立水库是缓解春季灌溉不足最有效的措施，但建立水库导致下游绿洲区生态需水严重缺乏，这就需要科学地在策勒河上游按照下游绿洲区不同季节需水量排放一部分河水支持生态需水。强烈建议利益相关者之间通过互相合作共同处理用水之间的冲突，特别是建立水权机制，提高农业灌溉节水效率，将农业节水的一部分水用于绿洲生态需水。 |
| Other Abstract | As a unique natural and cultural landscape in arid and semiarid regions, the oasis is essential for human survival and settlement, as well as for natural ecosystem health. The water resources play a crucial role in ensuring the existence, stability, and development of the oasis, also restricting the sustainable development of ecological environment and social economy in the oasis areas. The interdisciplinary research of coupled “water resources-ecological environment-social economy” system has become primary work for achieving integrated water resource management in the oasis regions. Recently, in face of double pressure of the agricultural economic development and ecological environment protection, together with the global warming, the limited water resources have led to the competitions and conflicts among the water sectors remarkably. Therefore, the development and research of coupled support system are also the goal of sustainable water resources management in the oasis areas. Since the oasis has the similar structure and function, the Qira oasis in the Northwest China is facing severe water competition between natural ecosystems and demands of human activities as the other oases are. In this dissertation, the Qira oasis is selected as the typical research area. Firstly, the temperature, precipitation, and evaporation time series of meteorological station are reconstructed by multiple sources of data in the Qira River basin, and then runoff simulation and its uncertainty is analyzed using hydrological model. Secondly, the concept of ecological water demands in the oasis areas is extended and determined at different levels based on the maximum and compatibility principle. Thirdly, considering the agricultural water having been occupying the water requirements of natural ecosystem, the discrete and hybrid Bayesian networks are developed to coordinate the water-use conflicts between agricultural irrigation and ecological water demands. Finally, to reflect senses of responsibility and ownerships of stakeholders among the different water sectors, the participatory Bayesian network model is developed to conduct the integration of ecosystem services into an integrated water resource management framework via public negotiation and feedback. According to scenario simulations under different management measures, the suggestions and schemes on relational water utilization are built for the sustainable development of water resources in Qira oasis area, thus it will achieve the interdisciplinary research of the integrated water resource management in the oasis areas. The main conclusions are listed as follows: (1) Modified Delta method improves the precision of the reconstruction of the meteorological data under multiple data sources. The approach of multiple datasets effectively reduces the uncertainty of runoff simulation. Surface runoff is major water supply for oasis in the arid regions. The runoff simulation becomes the basis of water resource allocation. The results show that the calibration of the monthly precipitation at the two target stations is relatively poor in comparison with that of monthly mean temperature and monthly evaporation. This reflects that the meteorological variables of continuous spatio-temporal field are much easier to be reconstructed in comparison with those of discrete space-time field. The monthly runoff simulated by the TL-FNN-BP model with structure (9, 1, 1) shows very well (NSCE > 0.80). However, while multiple datasets as the data of input variables can reduce the uncertainty of runoff simulation, the results also reveal that the efficiency in runoff simulation is limited because of the errors in the meteorological data and the deficiency in the model’s structure. (2) The ecological water demands account for 40.29%-58.75% of runoff in Qira oasis area. In the elements of ecological water demands, groundwater restoration in the desert accounts for higher proportion. The ecological water demands for supporting ecosystem services of oasis are crucial in maintaining the stability of oasis. Results indicate that the annual maximum, medium, and minimum ecological water demands are 0.752×108 , 0.619×108, and 0.516×108 m3, respectively, which account for 58.75%, 48.36%, and 40.29% of the natural river runoff. Moreover, the water requirement for groundwater restoration of the oasis–desert ecotone takes a large proportion, representing 48.27%, 42.32%, and 37.03% of the total ecological water demands at maximum, medium, and minimum levels, respectively. (3) Discrete Bayesian network proposed by coupling natural factors (such as the oasis hydrology and ecological water demands) and social factors (such as water saving measures and irrigation methods) effectively determines optimal ecological water demands. The tradeoff for water resources between agricultural and natural oasis ecosystems has become the basis for water resource management in oasis areas. The results indicate that BNs can model ecological water demand decision-making associated with agricultural economic loss effectively, as a powerful tool to coordinate water-use conflicts. The ecological water demand is determined as 50.24%, 49.71% and 48.73% of the natural river flow in wet, normal and dry years, respectively. Without further agricultural economic loss, 1.93%, 0.66% and 0.43% of more river discharge can be allocated to eco-environmental water demands under the combined strategy in wet, normal and dry years, respectively. (4) Hybrid Bayesian network established simulates the compensation standards of main economic crops using dynamic discretization algorithm, more effectively coordinating the conflicts between ecological water requirements and irrigation water. A trade-off framework is proposed to coordinate the water-use conflict between agriculture and the eco-environment, based on economic compensation for irrigation stakeholders. The results demonstrate that the water trade-off based on economic compensation depends on the available water supply and environmental flows at different levels. In dry years, the economic compensation for wheat, maize, cotton, and Chinese date is 1995.90, 1675.42, 3239.76, and 3513.12 RMB/ha, respectively. Agricultural irrigation water extracted for grain crops should be preferentially guaranteed to ensure food security, in spite of higher economic compensation in other cash crops’ irrigation for water coordination. Updating water-saving engineering and adopting drip irrigation technology in agricultural facilities after satisfying ecological water demands would greatly relieve agricultural water shortage and save 247, 271, 246, 237 RMB/ha for four irrigation stakeholders. (5) Development of participatory Bayesian network integrates ecosystem services into water resources management system. It successfully simulated the results and the effects of water resources management scenarios. Interdisciplinary water management is the goal of integrated water resource management for oasis. The Bayesian network model developed provides effective integration of ecosystem services into a quantitative IWMR framework via public negotiation and feedback. The intervention scenarios from the model conclude that any water management measure remains unable to sustain the ecosystem health in water-related ecosystem services. Greater cooperation among the stakeholders is highly necessary for dealing with such water conflicts. In particular, a proportion of the agricultural water saved through improving water-use efficiency should be transferred to natural ecosystems via water trade. |
| Subject Area | 自然地理学 |
| Language | 中文 |
| Document Type | 学位论文 |
| Identifier | http://ir.xjlas.org/handle/365004/14793 |
| Collection | 研究系统_荒漠环境研究室 研究系统_空间对地观测与系统模拟研究室 |
| Affiliation | 中国科学院新疆生态与地理研究所 |
| Recommended Citation GB/T 7714 | 薛杰. 基于贝叶斯网络的绿洲水资源综合管理研究[D]. 新疆乌鲁木齐. 中国科学院大学,2017. |
| Files in This Item: | There are no files associated with this item. | |||||
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.
Edit Comment