[关键词]
[摘要]
降水量是流域水资源管理、洪涝灾害预报以及农业用水计划等研究中的关键参数, 对防洪预报、水资源规划等 具有重要意义。根据铁岭地区1960 年- 2006 年逐月降水资料, 在分析其气候倾向率的基础上利用M ann2Kendall 非参数检验法进行显著性检验, 探讨该地区降水量的变化趋势, 并首次将广义回归神经网络( GRNN) 模型应用在该 地区的降水预测中, 利用误差率和预测对比图对模型的预测效果进行分析。结果表明: 近47 年来, 铁岭地区降水量 有缓慢减少趋势, 但该趋势未达到显著水平( P> 01 1) ; 从预测效果来看, 各月的预测值与真实值相差不大, 其中预 测效果最好的是10 月( 误差为31 39%) , 效果最不理想是2 月( 误差为191 45%) 。
[Key word]
[Abstract]
Pr ecipitation is the key parameter for water resources manag ement, flood disaster fo recast, and agricultural water usage. Its ac2 curate prediction is of g reat sig nificance for the flood forecast and water resources regulation. In this paper, the precipitation trend in Tieling distr ict was analyzed based on the monthly pr ecipitation data fr om 1960 to 2006. The Mann2Kendall nonparametric method was used to test the significance level on the basis of analysis of climate tendency r ate. The GRNN ( generalized reg ressio n neural netw ork) model w as used in precipitatio n for ecast, and the prediction r esults were analyzed using the error deviat ion and predictio n compariso n fig2 ures. The results showed that prediction in Tieling decreases slow ly in the past 47 years but the tr end hasn. t r eached the sig nifi2 cance lev el ( P > 01 1) . The pr edicted precipit ation w as sim ilar to the actual va lue fo r each mo nth. October has the best predictio n effect w hile Febr ua ry has the least w ith the er ro r o f 31 39% and 191 45%, respectiv ely .
[中图分类号]
[基金项目]