海表面温度预报在海洋相关领域具有重要的实用价值,随着遥感信息采集技术的不断发展和完善,区域内海表面温度数据采集的完整性得到了保障。现今大多数方法在预报海表面温度时,只考虑了海表面温度的时间相关性,并未利用其空间相关性,使...海表面温度预报在海洋相关领域具有重要的实用价值,随着遥感信息采集技术的不断发展和完善,区域内海表面温度数据采集的完整性得到了保障。现今大多数方法在预报海表面温度时,只考虑了海表面温度的时间相关性,并未利用其空间相关性,使得预报精度受到限制。针对该问题,本文将区域内每天的海表面温度数据作为一个矩阵输入模型,便于时间和空间信息的提取,并提出了CA-ConvLSTM模型来预报海表面温度。该模型首先利用卷积层对海表面温度矩阵进行局部特征提取,然后通过注意力模型为矩阵序列分配权重,将权重与矩阵序列对应相乘得到加权特征序列,最后,利用ConvLSTM进行预报,获得未来一天或五天内的海表面温度。通过实验确定模型的结构、输入尺寸和k值,再将CA-ConvLSTM与SVR、LSTM和ConvLSTM进行对比。实验结果表明:CA-ConvLSTM的均方根误差(Root Mean Square Error,RMSE)和预报精度(Prediction Accuracy,PACC)指标均要优于其他三种预报方法,验证了本文方法的有效性。展开更多
Process of sea surface diurnal warming has drawn a lot of attention in recent years, but that occurs in shelf seas was rarely addressed. In the present work, surface diurnal warming strength in the East China Sea was ...Process of sea surface diurnal warming has drawn a lot of attention in recent years, but that occurs in shelf seas was rarely addressed. In the present work, surface diurnal warming strength in the East China Sea was calculated by the sea surface temperature(SST) data derived from the MODIS sensors carried by the satellites Aqua and Terra. Due to transit time difference, both the number of valid data and the surface diurnal warming strength computed by the MODIS-Aqua data are relatively larger than Terra. Therefore, the 10-year MODIS-Aqua data from 2005 to 2014 were used to analyze the monthly variability of the surface diurnal warming. Generally, the surface diurnal warming in the East China sea is stronger in summer and autumn but weaker in winter and spring, while it shows different peaks in different regions. Large events with ΔT≥5 K have also been discussed. They were found mainly in coastal area, especially near the Changjiang(Yangtze) River estuary. And there exists a high-incidence period from April to July. Furthermore, the relationship between surface diurnal warming and wind speed was discussed. Larger diurnal warming mainly lies in areas with low wind speed. And its possibility decreases with the increase of wind speed. Events with ΔT ≥2.5 K rarely occur when wind speed is over 12 m/s. Study on surface diurnal warming in the East China Sea may help to understand the daily scale air-sea interaction in the shelf seas. A potential application might be in the marine weather forecasts by numerical models. Its impact on the coastal eco-system and the activities of marine organisms can also be pursued.展开更多
Reanalysis data from NCEP/NCAR are used to systematically study preceding signals of monthly precipitation anomalies in the early raining season of Guangdong province, from the viewpoints of 500-hPa geopotential heigh...Reanalysis data from NCEP/NCAR are used to systematically study preceding signals of monthly precipitation anomalies in the early raining season of Guangdong province, from the viewpoints of 500-hPa geopotential height field, outgoing longwave radiation (OLR) field, sea surface temperature (SST) and fourteen indexes of general circulation depicting atmosphere activity at high, middle and low latitutes. Being multiple tools of information, a number of conceptual models are formulated that are useful for prediction of the magnitude of monthly precipitation (drought, flood and normal conditionss).展开更多
文摘海表面温度预报在海洋相关领域具有重要的实用价值,随着遥感信息采集技术的不断发展和完善,区域内海表面温度数据采集的完整性得到了保障。现今大多数方法在预报海表面温度时,只考虑了海表面温度的时间相关性,并未利用其空间相关性,使得预报精度受到限制。针对该问题,本文将区域内每天的海表面温度数据作为一个矩阵输入模型,便于时间和空间信息的提取,并提出了CA-ConvLSTM模型来预报海表面温度。该模型首先利用卷积层对海表面温度矩阵进行局部特征提取,然后通过注意力模型为矩阵序列分配权重,将权重与矩阵序列对应相乘得到加权特征序列,最后,利用ConvLSTM进行预报,获得未来一天或五天内的海表面温度。通过实验确定模型的结构、输入尺寸和k值,再将CA-ConvLSTM与SVR、LSTM和ConvLSTM进行对比。实验结果表明:CA-ConvLSTM的均方根误差(Root Mean Square Error,RMSE)和预报精度(Prediction Accuracy,PACC)指标均要优于其他三种预报方法,验证了本文方法的有效性。
基金Supported by the Zhejiang Provincial Natural Science Foundation(No.LY17D060003)the Shandong Provincial Natural Science Foundation(No.ZR2015DQ006)+1 种基金the National Narutal Science Foundation of China(Nos.41306035,41206006)the National Key R&D Plan of China(No.2016YFC1401404)
文摘Process of sea surface diurnal warming has drawn a lot of attention in recent years, but that occurs in shelf seas was rarely addressed. In the present work, surface diurnal warming strength in the East China Sea was calculated by the sea surface temperature(SST) data derived from the MODIS sensors carried by the satellites Aqua and Terra. Due to transit time difference, both the number of valid data and the surface diurnal warming strength computed by the MODIS-Aqua data are relatively larger than Terra. Therefore, the 10-year MODIS-Aqua data from 2005 to 2014 were used to analyze the monthly variability of the surface diurnal warming. Generally, the surface diurnal warming in the East China sea is stronger in summer and autumn but weaker in winter and spring, while it shows different peaks in different regions. Large events with ΔT≥5 K have also been discussed. They were found mainly in coastal area, especially near the Changjiang(Yangtze) River estuary. And there exists a high-incidence period from April to July. Furthermore, the relationship between surface diurnal warming and wind speed was discussed. Larger diurnal warming mainly lies in areas with low wind speed. And its possibility decreases with the increase of wind speed. Events with ΔT ≥2.5 K rarely occur when wind speed is over 12 m/s. Study on surface diurnal warming in the East China Sea may help to understand the daily scale air-sea interaction in the shelf seas. A potential application might be in the marine weather forecasts by numerical models. Its impact on the coastal eco-system and the activities of marine organisms can also be pursued.
基金Short-term Climate Prediction Study for Guangdong Province a key project of Guangdong Science and Technology Committee in the national 9th five-year economic development plan Research on Long-term Tendency Prediction System for Floods/Drought and Typh
文摘Reanalysis data from NCEP/NCAR are used to systematically study preceding signals of monthly precipitation anomalies in the early raining season of Guangdong province, from the viewpoints of 500-hPa geopotential height field, outgoing longwave radiation (OLR) field, sea surface temperature (SST) and fourteen indexes of general circulation depicting atmosphere activity at high, middle and low latitutes. Being multiple tools of information, a number of conceptual models are formulated that are useful for prediction of the magnitude of monthly precipitation (drought, flood and normal conditionss).
