传统旱涝急转评估指数的局限和改进

针对传统旱涝急转指数存在错漏判及无法统一旱涝急转等级划分的局限, 在明晰旱涝急转定义基础上, 提出了一种改进的标准化旱涝急转指数。通过理论模拟和案例实践, 结果表明: ①传统的旱涝急转指数变化范围较大, 且随着权重参数增大, 错判率减小但漏判率增大, 因此无法避免错漏判问题；错判情形发生在正常状态转为较为严重的涝或较为严重的涝转为正常状态, 漏判发生在相邻的旱和涝等级均较轻的情形。②改进的标准化旱涝急转指数与旱涝指数的变化范围差别较小, 旱涝急转评估可与旱涝评估(如基于气象或水文干旱指数)的分级及阈值设置保持一致, 且通过先识别相邻事件是否分别为旱、涝而避免了旱涝急转的错判, 构建的新权重项避免了旱涝急转的漏判。③作为粤港澳大湾区供水水源的东江流域旱转涝主要发生在前汛期开始的4—5月, 发生比例为33.8%；涝转旱主要发生在后汛期结束后的10月, 发生比例为18.5%；旱转涝和涝转旱并存的年份有12 a。

Limitations and improvement of the traditional assessment index for drought-wetness abrupt alternation

Using the traditional drought-flood abrupt alternation index (DFAI), it was difficult to classify abrupt alternations which leads to the misjudgment and omission for certain alternations. This study proposed an improved standardized drought-wetness abrupt alteration index (SDWAI), by clarifying the definition of drought-wetness abrupt alteration. Simulation and case study showed that the range of DFAI was large. The larger the weight parameter of DFAI was, the smaller the misjudgment rate of abrupt alternation event was but the larger the omission rate was, implying that misjudgment and omission of abrupt alternation events were unavoidable. The misjudgment generally occurred when a normal event turned into a severer flood or a severer flood turned into a normal for the adjacent flows. The omission normally occurred when adjacent drought and flood were light. Due to small differences between the ranges of SDWAI and drought-wetness index, the classified levels and their thresholds of drought-wetness abrupt alternations were consistent with those of drought-wetness assessment. The misjudgment of an abrupt event was avoided by recognizing in advance that the adjacent events were drought and wetness events, respectively. The omission of an abrupt event was avoided by the reconstructed weight parameter. In the Dongjiang River basin, which is the source of water supply to the Guangdong-Hong Kong-Macao Greater Bay Area, 33.8% the abrupt alternations from drought to flood mainly occurred in April and May at the beginning of the former flood season, and 18.5% of that from flood to drought mainly occurred in October at the end of the latter flood season. In addition, there were twelve years for the coexistence of the abrupt alternations from drought to flood and from flood to drought.