1 School of Transportation and Logistics Engineering, Xinjiang Agricultural University, Urumqi 830052, China
2 Technology Research and Development Center, Xinjiang Transportation Planning, Survey and Design Institute Co., Ltd., Urumqi 830006, China
3 Xinjiang Key Laboratory for Safety and Health of Transportation Infrastructure in Alpine and High-altitude Mountainous Areas, Urumqi 830006, China
Show less
文章历史+
Received
Published
Revised
2025-06-11
2026-04-28
2025-11-26
Online Date
2026-05-29
PDF (6745K)
摘要
针对传统降雨预测模型无法适应新疆干旱区季节性降雨差异及小样本条件下预测精度衰减显著的问题,提出一种融合动态阈值优化与多特征LightGBM的预测方法。通过构建包含大气可降水量(precipitable water vapor,PWV)基准值、PWV增量、PWV增率等9维特征体系,采用动态阈值分月建模机制,有效解决了季节性降雨差异与训练样本不足的双重挑战。结果表明:动态阈值模型在4月临界成功指数达到85.5%,较全年最优固定阈值提升7.5%;模型在7月强对流降雨与9月层云降雨中命中率均保持在92%以上,且误报率在22%以下,验证了动态阈值机制与多特征融合策略对季节性降雨的适应性,提升了小样本条件下的降雨预测精度,为提升新疆干旱区公路地质灾害防治能力提供技术支撑。
Abstract
Traditional rainfall prediction models have limitations in adapting to seasonal precipitation variations and suffer from significant accuracy degradation under small-sample conditions in Xinjiang's arid regions. To address this, a prediction method integrating dynamic threshold optimization and multi-feature LightGBM was proposed. A 9-dimensional feature system was constructed, including precipitable water vapor (PWV) baseline values, PWV increments, PWV growth rates, and other auxiliary atmospheric variables. A monthly dynamic threshold modeling mechanism was adopted to address the dual challenges of seasonal rainfall differences and insufficient training samples. The results show that the dynamic threshold model achieves a critical success index of 85.5% in April, which is 7.5 percentage points higher than the annual optimal fixed threshold. The model maintains a detection rate above 92% for both convective rainfall in July and stratiform rainfall in September, with a false alarm rate below 22%. The effectiveness of the dynamic threshold mechanism and multi-feature fusion strategy is confirmed in adapting to seasonal rainfall patterns and improving prediction accuracy under limited sample conditions, thereby providing technical support for the prevention of highway geological disasters in the arid regions of Xinjiang.
CHEN Zhi-qiang, LIU Qi, WANG Yang, HUANG Ji-chen, LIU Jie.
Dynamic Threshold Optimization with Multi-feature-based LightGBM for Precipitation Prediction from GNSS PWV: K105 Section of Route S101, Xinjiang[J].
Science Technology and Engineering, 2026, 26(12): 4940-4947 DOI:10.12404/j.issn.1671-1815.2504356
ManandharS, DevS, LeeY H, et al. Systematic study of weather variables for rainfall detection[C]// Proceedings of 2018 IEEE International Geoscience and Remote Sensing Symposium. Valencia: IEEE, 2018: 3027-3030.
ZhangKefei, LiHaobo, WangXiaoming, et al. Recent progresses and future prospectives of ground-based GNSS water vapor sounding[J]. Acta Geodaetica et Cartographica Sinica, 2022, 51(7): 1172-1191.
ZhengZhiqing, ZhangKefei, ShiJiaqi, et al. Using GNSS-PWV combined with multiple meteorological elements to analyze the process of “21·7” severe rainstorm in Henan Province[J]. Journal of Geodesy and Geodynamics, 2023, 43(8): 809-815.
LiuYang, ZhaoQingzhi, YaoWanqiang. Rainfall forecast based on multihidden layers neural network and GNSS-PWV[J]. Bulletin of Surveying and Mapping, 2019(S1): 36-40.
ZhaoQingzhi, LiuYang, YaoWanqiang. Establishment of short-term rainfall forecast model by least square support vector machine[J]. Journal of Geodesy and Geodynamics, 2021, 41(2): 152-156.
WangShitai, ZhangDinghong, YinMin, et al. Short-term and imminent rainfall forecast based on LSTM model[J]. Radio Engineering, 2021, 51(11): 1278-1283.
[12]
CaiM, LiJ Y, LiuL H, et al. Weighted mean temperature hybrid models in China based on artificial neural network methods[J]. Remote Sensing, 2022, 14(15): 22-23.
[13]
BenevidesP, CatalaoJ, MirandaA M P. On the inclusion of GPS precipitable water vapour in the nowcasting of rainfall[J]. Natural Hazards and Earth System Sciences, 2015, 15(12): 2605-2616.
[14]
ManandharS, LeeH Y, MengS Y. GPS-PWV based improved long-term rainfall prediction algorithm for tropical regions[J]. Remote Sensing, 2019, 11(22): 1-17.
ShanLulu, YaoYibin, ZhaoQingzhi, et al. A short-term rainfall forecasting method based on GNSS PWV data[J]. Journal of Geomatics2019, 44(1): 22-26.
[17]
SaastamoinenJ. Atmospheric correction for the troposphere and stratosphere in radio ranging satellites[J]. Geophysical Monograph Series, 1972, 15: 247-251.
[18]
ZengJ, GengJ H, LiG C, et al. Improving cycle slip detection in ambiguity-fixed precise point positioning for kinematic LEO orbit determination[J]. GPS Solutions, 2024, 28(3): 135-149.
WangHaoyu, HanLing, LiLiangzhi. Daily precipitation forecasting using a global weather model and regional pre-training optimization: a case study in Shaanxi Province[J]. Science Technology and Engineering, 2025, 25(20): 8379-8391.
ChenWeinan, HuZhijian, YueJingpeng, et al. Short-term loadprediction based on combined model of long short-term memory network and Light gradient boosting machine[J]. Automation of Electric Power Systems, 2021, 45(4): 91-97.
YangLi, ZhouZixi, WangTingting, et al. Single well production forecasting method based on the CNN-GRU-LightGBM model[J]. Science Technology and Engineering, 2024, 24(18): 7606-7614.
[27]
DonaldsonR I, DyerR M, KrausM J. Objective evaluator of techniques for predicting severe weather events[J]. Bulletin of the American Meteorological Society, 1975, 56(7): 29-34.
[28]
YaoY, ShanL, ZhaoQ. Establishing a method of short-term rainfall forecasting based on GNSS derived PWV and its application[J]. Scientific Reports, 2017, 7(1): 12465.
LuHaizhao, ZhangLieping, WangShoufeng, et al. Indoor fingerprint location algorithm based on improved LightGBM[J]. Science Technology and Engineering, 2024, 24(15): 6306-6312.
JiangHuimin, LiuChunyun, JiaJian, et al. The temporal and spatial characteristics of convective precipitation in Xinjiang among the summer and causes analysis[J]. Plateau Meteorology, 2019, 38(2): 340-348.