...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Current Science: A Fortnightly Journal of Research >Improvement in nearest neighbourweather forecast model performancewhile considering the previous day'sforecast for drawing forecast for thefollowing day
【24h】

Improvement in nearest neighbourweather forecast model performancewhile considering the previous day'sforecast for drawing forecast for thefollowing day

机译：考虑到前一天的预报以绘制下一天的预报时，改进最近邻天气预报模型的性能

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Nearest neighbour model for prediction of weather interms of snow dayo snow day for consecutive threedays in advance (lead time up to 72 h) was tested intwo different modes of prediction for two differentstations; Dhundi in Himachal Pradesh and Stage-II inJammu and Kashmir (J&K), in the Pir Panjal rangeof NW Himalaya, with two different types of data. Thedata of station Stage-II are incomplete with less dataof 12 winters (winter 1991-92 to winter 2003-04, miss-ing data of 1994-95) and those of station Dhundi arecomplete with more data of 15 winters (winter 1989-90 to winter 2003-04). The model performance was testedwith incomplete and complete data respectively, intwo different modes. First, in mode I prediction ofweather is made based on the probability of snowfallcalculated from nearest daysearest situations. Sec-ondly, in mode II the prediction was made consideringthe previous day's probability of snowfall also, alongwith the probability of snowfall calculated from near-est days/situations, i.e. while forecasting for day-2 (leadtime 48 h), probability of snowfall for day-1 (lead time24 h) is also taken into account.The model performance is found to be better formode II compared to mode I for all three days exceptfor day-1 forecast with incomplete data. The modelperformance is better for Stage-II compared to Dhundiin both the modes. Significant difference in the modelperformance for day-1 and day-2 forecasts is foundbetween those with incomplete data compared to thosewith complete data. The model results are briefly dis-cussed here.
机译：在两个不同站点的两种不同预测模式下，测试了用于连续三天（提前期长达72小时）的连续下雪天天气预报的最近邻居模型。位于喜马拉雅山西北部Pir Panjal范围内的喜马al尔邦的Dhundi和贾木和克什米尔（J＆K）的第二阶段，有两种不同类型的数据。 Stage II站的数据不完整，缺少12个冬季（1991-92年冬季至2003-04冬季，1994-95年冬季缺失数据），而Dhundi站的数据不完整，具有15个冬季（1989-90年冬季）更多数据到2003-04年冬季）。使用两种不同的模式分别使用不完整和完整的数据测试了模型性能。首先，在模式I中，天气预测是根据根据最近的天/最近情况计算的降雪概率进行的。其次，在模式II中，还考虑了前一天的降雪概率，以及根据最近的天/地计算出的降雪概率，即在预测第2天（提前期48小时）时的降雪概率。还考虑了第1天（交货时间24小时）。在模式3中，与模式1相比，在所有三天中，模型1的模型性能都优于模型1，但第一天的预测数据不完整。与Dhundiin两种模式相比，第二阶段的模型性能都更好。发现具有不完整数据的模型与具有完整数据的模型之间在第一天和第二天预测中模型性能的显着差异。在此简要讨论模型结果。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号