遥感与水文模型融合的干旱绿洲流域尺度土壤盐渍化数值模拟

Numerical modelling of soil salinity by considering outer space of oasis and plain area as an integral shall enhance soil salinity prediction capability, meanwhile, the contribution of natural and human induced factors can also be analyzed in long time series. However, most of previous studies examined landscapes with salinities far in excess of the range where nearly all agriculture occurs, thus has little commercial value for food production or land rehabilitation. A physically based distributed hydrological model MIKE SHE was setup based on multi-source remote/proximal sensing data and parameter inversion method in the delta oasis of Weigan-Kucha river. We would like to focus on exploring the potential of using multi-factor data mining and multi-variable geostatistical model in estimating soil electrical conductivity and assessing the spatial and temporal changes in soil salinity using AD module and empirical statistic methods respectively. Secondly, the optimal model was adopted to predict water and salinity dynamics over the past 50 years in watershed scale. Wavelet analysis was incorporated to reveal the pattern of soil salinity and its control factors in different scale. Lastly, with reference to previous land management policy and climate change hypothesis, different levels of natural and human affected environmental conditions were predefined to predict the extent and evolution of saline areas over the next 50 years so as to provide some basic information to decision makers.

绿洲外围与平原结合的流域土壤盐渍化数值模拟有助于提高农业产区土壤盐渍化预警能力，分析长时间序列范围内自然和人为因素对土壤盐渍化变化的贡献率。然而，以往研究大都集中于绿洲外围，对农业生产和土地恢复的价值有限。本项目以渭干河-库车河流域为研究区，采用遥感陆面参数反演方法结合星-空-地多源数据构建分布式水文模型。重点探讨多因子数据挖掘模型和多变量地统计模型在土壤电导率参数化方面的可行性和溶质运移模块与经验统计分析法分别估算土壤盐渍化变化的潜力。其次，采用最优模型方法实现流域尺度过去50年土壤水和盐渍化数值模拟。结合小波尺度分析方法，揭示干旱区土壤盐渍化空间分布规律和关键控制因素。最后，参考气候变化假设和过往耕作管理方式，设置不同自然和人为影响梯度，模拟未来50年土壤盐渍化演变趋势，为关相农业生产管理决策提供基础信息。