我国绿潮灾害时间序列特征的模拟与预测  

Simulation and prediction of the characteristics of green tide disaster time series in China

作　　者：刘旭[1,2,3] 姜珊[1] 王峥[1] 梁颖祺[1] 何恩业[1] LIU Xu;JIANG Shan;WANG Zheng;LIANG Yingqi;HE Enye(National Marine Environmental Forecasting Center,Beijing 100081 China;School of Economics&Management,Beijing Forestry University,Beijing 100083,China;Key Laboratory of Marine Hazards Forecasting,National Marine Environmental Forecasting Center,Ministry of Natural Resources,Beijing 100081,China)

机构地区：[1]国家海洋环境预报中心,北京100081 [2]北京林业大学经济管理学院,北京100083 [3]自然资源部海洋灾害预报技术重点实验室国家海洋环境预报中心,北京100081

出　　处：《海洋预报》2023年第2期56-66,共11页Marine Forecasts

基　　金：国家自然科学基金(41976221、41576029);国家重点研发计划“地球观测与导航”重点专项资助项目(2021YFB3900405)。

摘　　要：基于2010—2019年黄海绿潮卫星遥感影像资料,构建绿潮覆盖面积时间序列分析方法。将每年5月8日—8月7日成像条件较好的遥感监测数据预处理为周平均时间序列,将2010—2018年设定为模型训练集,2019年设定为模型验证集。基于Box-Jenkins法构建了差分整合自回归移动平均模型ARIMA(2,0,2)、加法季节性模型SARIMA(1,0,0)×(0,1,0)_(12)和乘法季节性模型SARIMA(1,0,0)×(0,1,1)_(12),3个模型都通过模型白噪声检验和参数显著性检验,具有较好的模拟效果和可预测性。SARIMA(1,0,0)×(0,1,1)_(12)的赤池信息准则值最小,2019年平均绝对误差为95.56 km^(2),均方根误差为156.74 km^(2),与ARIMA(2,0,2)相比,MAE提高12%,RMSE下降1.2%,SARIMA(1,0,0)×(0,1,0)12的预测精度最低,MAE和RMSE分别为115.12 km^(2)和192.16 km^(2)。Based on the satellite remote sensing images of green tide in the Yellow Sea from 2010 to 2019,an analysis method of the green tide coverage area time series is constructed in this paper.The remote sensing monitoring data with good imaging conditions from May 8th to August 7th each year is preprocessed into a weekly average time serie.The years from 2010 to 2018 are set as the model training set,and the year 2019 is set as the model validation set.Based on the Box-Jenkins method,the Autoregressive Integrated Moving Average ARIMA(2,0,2),additive seasonal model SARIMA(1,0,0)×(0,1,0)_(12)and multiplicative seasonal model SARIMA(1,0,0)×(0,1,1)12 are constructed,which all pass the model white noise test and parameter significance test with good simulation effect and predictability.Specially,the Akaike Information Criterion(AIC)value of SARIMA(1,0,0)×(0,1,1)_(12)is the smallest with the mean absolute error(MAE)and root mean squared error(EMSE)of 95.56 km^(2)and 156.74 km^(2),respectively in 2019,which improved Compared with ARMA(2,0,2),the MAE increases by 12%and the RMSE decreases by 1.2%.The prediction accuracy of SARIMA(1,0,0)×(0,1,0)_(12)is the lowest with the MAE and RMSE of 115.12 km^(2)and 192.16 km^(2),respectively.

关 键 词：绿潮 差分整合自回归移动平均模型 时间序列法 遥感监测 

分 类 号：X55[环境科学与工程—环境工程]