多层递阶回归方法在地下水资源评价中的应用——以吉林中部平原区德惠市为例

Application of Multi-Level Recursive Regression for Groundwater Resources Assessment in Dehui,Jilin

地下水资源是具有时空动态变化的系统,以往用于地下水位预测和资源计算的数理统计方法均为固定参数模型,不免影响了计算结果的可靠性。而多层递阶回归方法是针对时空动态系统而言的,它摒弃了数理统计方法中的固定参数预报模型,模型中的系统参数是随时空变化的,是建立在将预报对象(如地下水位)作为随机动态时变系统的基础之上的。它首先进行动态系统的时空参数预报,然后在此基础之上进行系统状态的预报,从而提高了预测结果的可靠性。该方法对计算含水层的水文地质参数和边界条件无严格要求,只要求地下水水位、开采量与降水量等补给量的多年统计资料,因此非常适用于水文地质条件复杂又有地下水长系列动态观测资料的地下水开发区。本文以吉林中部平原区德惠市为例,阐明了该方法的运用过程。

The groundwater system can be assumed to have spatio-temporal variability, with changes in the factors and parameters of groundwater. Previously used mathematical methods for prediction of groundwater levels and estimates of groundwater have fixed parameters, which reduce the accuracy of results. Using a multi-level recursive regression is one way of dealing with spatial and temporal variability, since the model parameters are variable. The dependability of forecast results are enhanced because the state of groundwater is predicted based on time-sensitive parameters forecast in the model. The parameters reflect the impact of input factors （e. g. precipitation, human use） on the groundwater system, and are expressed by an exact numerical value. Initially, we only have data on groundwater level, human use and precipitation, and are lacking accurate parameters for aquifers and boundary conditions. Thus, stochastic methods are appropriate for an area with complex hydrogeological conditions and abundant groundwater level data. First, we use the data to establish the multi-level recursive regression model with parameters that vary over time. Then, the model is used to analyze changes in the groundwater system and the impact of different yields by testing different trial yields in the model. Then, we estimate a potential withdrawal. This quantity is then put into the model again, and the factual withdrawal will be worked out through the restrictive terms. This paper illustrates this method, using as an example the city of Dehui in the plains area of central Jilin. The multiple layers of aquifers in the region are thin and complicated, and the area is lacking precise data on hydrogeological conditions. On the other hand, the main groundwater recharge of the region is precipitation, and the withdrawals are mainly due to human activity. The input and output of groundwater are distributed regionally, so it is optimal to apply multi-level recursive regression method for the region. The results show that the allowable withdrawal of Quaternary pore groundwater in Dehui is about 1.3 ×10^8m^3/year. From 1991, the actual withdrawal was more than 1.3 ×10^8m^3/year, so the water table has dropped below a safe level. Thus, the supply of groundwater is being depleted, and actions should be taken to limit withdrawals.