Page 94 - 《高原气象》2026年第3期
P. 94
高 原 气 象 45 卷
702
问题。一方面,模型对极端强降水的估计能力不 mos14010076.
足,对此可采用加权损失函数,对实况雨量超过 陈训来, 陈元昭, 赵春阳, 等, 2019. 梯度提升决策树在雷达定量降
水估测中的应用研究[J]. 气象科技进展, 9(3): 132-137.
1. 0 mm·min 的样本赋予更大的损失权重,以增强
-1
DOI: 10. 3969/j. issn. 2095-1973. 2019. 03. 018. Chen X L,
模型对极端降水特征的学习能力;另一方面,该模 Chen Y Z, Zhao C Y, et al, 2019. Application of gradient boost
型仅适用于成都平坝地区,且仅针对夏季强对流天 decision tree in radar quantitative precipitation estimation[J]. Ad‐
气,缺乏普适性。后续将针对上述问题开展更为深 vances in Meteorological Science and Technology, 9(3): 132-
入的研究。 137. DOI: 10. 3969/j. issn. 2095-1973. 2019. 03. 018.
傅德胜, 肖晨, 谭畅, 等, 2015. RBF 神经网络在雷达定量估测降水
参考文献(References): 中的应用研究[J]. 气象科学, 35(2): 199-203. DOI: 10. 3969/
2013jms. 0097. Fu D S, Xiao C, Tan C, et al, 2015. Application
Chandrasekar V, Bringi V N, 1988. Error structure of multiparameter of RBF neural network in radar quantitative precipitation estima‐
radar and surface measurements of rainfall part I: differential re‐ tion[J]. Journal of the Meteorological Sciences, 35(2): 199-
flectivity[J]. Journal of Atmospheric and Oceanic Technology, 5 203. DOI: 10. 3969/2013jms. 0097.
(6): 783-795. DOI: 10. 1175/1520-0426(1988)005<0783: 谷军霞, 师春香, 潘旸, 2018. 天气雷达定量估测降水研究进展[J].
esomra>2. 0. co; 2. 气象科技进展, 8(1): 71-78. DOI: 10. 3969/j. issn. 2095-1973.
Chandrasekar V, Bringi V N, Balakrishnan N, et al, 1990. Error 2018. 01. 008. Gu J X, Shi C X, Pan Y, 2018. Advances in re‐
structure of multiparameter radar and surface measurements of search of the weather radar quantitative precipitation estimation
rainfall. part III: specific differential phase[J]. Journal of Atmo‐ [J]. Advances in Meteorological Science and Technology, 8(1):
spheric and Oceanic Technology, 7(5): 621-629. DOI: 10. 71-78. DOI: 10. 3969/j. issn. 2095-1973. 2018. 01. 008.
1175/1520-0426(1990)007<0621: esomra>2. 0. co; 2. 关吉平, 高延波, 黄从雷, 等, 2024. 基于Attention-Unet网络的FY-
Chen H N, Chandrasekar V, Tan H M, et al, 2019. Rainfall estima‐ 4A 卫星降水估计[J]. 气象学报, 82(3): 398-410. DOI: 10.
tion from ground radar and TRMM precipitation radar using hy‐ 11676/qxxb2024. 20230037. Guan J P, Gao Y B, Huang C L, et
brid deep neural networks[J]. Geophysical Research Letters, 46 al, 2024. Estimating FY-4A satellite precipitation based on a deep
(17-18): 10669-10678. DOI: 10. 1029/2019GL084771. learning model Attention-Unet[J]. Acta Meteorologica Sinica, 82
Loshchilov I, Hutter F, 2016. SGDR: stochastic gradient descent (3): 398-410. DOI: 10. 11676/qxxb2024. 20230037.
with warm restarts[A]. International Conference on Learning 黄兴友, 张永轩, 李芳, 等, 2024. 利用雷达数据开展对流降水临近
Representations (ICLR).DOI: 10. 1109/ICML. 2017. 468. 预报的循环神经网络方法试验[J]. 大气科学, 48(6): 2329-
Oktay O, Schlemper J, Le L, et al, 2018. Attention U-Net: learning 2341. DOI: 10. 3878/j. issn. 1006-9895. 2405. 24009. Huang X
where to look for the pancreas[EB/OL]. arXiv: 1804. 03999. Y, Zhang Y X, Li F, et al, 2024. Experiments of convective pre‐
https: //arxiv. org/abs/1804. 03999 cipitation nowcasting based on a recurrent neural network method
Ronneberger O, Fischer P, Brox T, 2015. U-Net: convolutional net‐ and weather radar data[J]. Chinese Journal of Atmospheric Sci‐
works for biomedical image segmentation[M]//Medical Image ences, 48(6): 2329-2341. DOI: 10. 3878/j. issn. 1006-9895.
Computing and Computer-Assisted Intervention-MICCAI 2015. 2405. 24009.
Cham: Springer International Publishing: 234-241. DOI: 10. 柳臣中, 周筠珺, 谷娟, 等, 2016. 成都地区雨滴谱特征[J]. 应用气
1007/978-3-319-24574-4_28. 象 学 报 , 26(1): 112-121. DOI: 10. 11898/1001-7313.
Ryzhkov A V, Zrnić D S, 1995. Comparison of dual-polarization ra‐ 20150112. Liu C Z, Zhou Y J, Gu J, et al, 2016. Characteristics
dar estimators of rain[J]. Journal of Atmospheric and Oceanic of raindrop size distribution in Chengdu[J]. Journal of Applied
Technology, 12(2): 249-256. DOI: 10. 1175/1520-0426(1995) Meteorological Science, 26(1): 112-121. DOI: 10. 11898/
012<0249: codpre>2. 0. co; 2. 1001-7313. 20150112.
Shen C P, 2018. A transdisciplinary review of deep learning research 李佳慧, 马建立, 陈明轩, 等, 2025. 基于瓦片分区的雷达动态定量
and its relevance for water resources scientists[J]. Water Resources 降水估测方法研究[J]. 高原气象, 44(1): 122-133. DOI: 10.
Research, 54(11): 8558-8593. DOI: 10. 1029/2018WR022643. 7522/j. issn. 1000-0534. 2024. 00052. Li J H, Ma J L, Chen M
Woo S, Park J, Lee J Y, et al, 2018. CBAM: convolutional block at‐ X, et al, 2025. Research on dynamic quantitative precipitation es‐
tention module [M]//Computer Vision-ECCV 2018. Cham: timation method based onTile partitioning for radar[J]. Plateau
Springer International Publishing: 3-19. DOI: 10. 1007/978-3- Meteorology, 44(1): 122-133. DOI: 10. 7522/j. issn. 1000-
030-01234-2_1. 0534. 2024. 00052.
Zhang T, Wei W, Zheng L Y, et al, 2023. Characteristics and varia‐ 刘艳霞, 文军, 谢晓林, 2024. 青藏高原中东部和四川盆地的夏季
tions of raindrop size distribution in Chengdu of the western Sich‐ 雨滴谱对比分析研究[J]. 高原气象, 43(1): 28-41. DOI: 10.
uan Basin, China[J]. Atmosphere, 14(1): 76. DOI: 10. 3390/at‐ 7522/j. issn. 1000-0534. 2023. 00033. Liu Y X, Wen J, Xie X

