Abstract—

An attempt has been made to construct models that approximate the time-range statistical properties of the useful inflow into Lake Baikal and its relationships with potential predictors that permit predictive assessments to be obtained. The problem solution consists of two parts. In the first part, the constructed physical and statistical models record the known patterns of runoff from the catchment. These patterns are subject to significant regression relationships between the predicted value of the monthly inflow and its hydrometeorological predictors. These models can be used to forecast the average useful inflow for the next month. The predictive correction for interannual variability to the seasonal inflow wave in these models is based on variability in the predictor values that are known at the forecasting time. The second part of the research is focused on possible trends and cycles in the interannual variability in useful inflow in the most water-abundant third quarter in order to predict it with a lead time of 1 year or more. Various cycle models with parameters estimated using the training sampling set do not provide a stable solution in the control sampling set. The best result has been obtained by a cyclical model with fixed periods, which records the possible impact on the inflow variability provided by gravitational influence of the Moon and major planets on the Earth and the Sun. Nevertheless, this model also should not be considered statistically significant due to the small proportion of the explainable variance and insufficiency of the data on useful inflow into Baikal.

中文翻译：

---

贝加尔湖有用流入量的变化模式和模型

摘要-

已经尝试构建模型来近似贝加尔湖有用流入量的时间范围统计特性及其与允许获得预测评估的潜在预测因子的关系。问题解决方案由两部分组成。在第一部分，构建的物理和统计模型记录了流域径流的已知模式。这些模式受每月流入量预测值与其水文气象预测值之间的显着回归关系的影响。这些模型可用于预测下个月的平均有用流入量。在这些模型中，对季节性流入波的年际变化的预测校正基于预测时已知的预测变量值的变化。研究的第二部分侧重于水资源最丰富的第三季度有用流入量的年际变化的可能趋势和周期，以便在 1 年或更长时间的提前期进行预测。使用训练采样集估计参数的各种循环模型在控制采样集中没有提供稳定的解决方案。最好的结果是通过固定周期的循环模型获得的，该模型记录了月球和地球和太阳上主要行星的引力影响对流入变化的可能影响。尽管如此，由于可解释的方差比例很小，而且贝加尔湖的有用流入数据不充分，因此该模型也不应被视为具有统计意义。