PDF(2235 KB)
PDF(2235 KB)
PDF(2235 KB)
利用机器学习模型预测中国沿海海平面变化
Prediction of sea level changes along the coast of China using machine learning models
该文利用线性回归函数,根据卫星测高及中国沿海6个验潮站数据估算出1993—2020年中国沿海绝对海平面上升速率为4.17±1.32 mm/a,相对海平面上升速率为4.47±0.90 mm/a。将1958—2020年的大气数据、海洋数据及气候模态指数作为预报因子,建立了长短期记忆神经网络模型(LSTM模型)、循环神经网络模型(RNN模型)、门控循环单元神经网络模型(GRU模型)和支持向量机回归模型(SVR模型)等多种神经网络模型对中国沿海6个验潮站周边的相对海平面变化趋势进行预测。模型评估结果表明,同时引入大气变量、海洋变量及气候模态指数变量的LSTM模型取得的预测值与观测值的平均相关系数和均方根误差分别为0.866和19.279 mm,在4种模型中表现最佳,可以作为一种新型的预测相对海平面变化的方法。
Based on the data of satellite altimetry and six tide gauge stations along the coast of China, linear regression function was used to estimate the absolute sea level rise rate in the coastal areas of China from 1993 to 2020, which was 4.17±1.32 mm/a, and the relative sea level rise rate was 4.47±0.90 mm/a. Taking the atmospheric data, ocean data and climate modal index from 1958 to 2020 as prediction factors, a variety of neural network models such as long short-term memory neural network model (LSTM model), recurrent neural network model (RNN model), gated recurrent unit neural network model (GRU model) and support vector machine regression model (SVR model) were established to predict the trend of relative sea level changes around the six tide gauge stations along the coast of China. The model evaluation results show that the average correlation coefficient and root mean square error of the observed value and the predicted value obtained by the LSTM model that simultaneously introduces atmospheric and ocean variables and climate modal index variables are 0.866 and 19.279 mm, respectively, which performs the best among the four models, and therefore the LSTM model can be used as a new method for predicting relative sea level changes.
海平面上升 / 相对海平面 / 绝对海平面 / 垂直地壳运动 / 卫星测高 / 潮位 / LSTM神经网络模型 / 时间序列预测
sea level rise / relative sea level / absolute sea level / vertical land movement / satellite altimetry / tide level / LSTM neural network model / time series forecasting
| [1] |
|
| [2] |
We propose a simple relationship linking global sea-level variations on time scales of decades to centuries to global mean temperature. This relationship is tested on synthetic data from a global climate model for the past millennium and the next century. When applied to observed data of sea level and temperature for 1880-2000, and taking into account known anthropogenic hydrologic contributions to sea level, the correlation is >0.99, explaining 98% of the variance. For future global temperature scenarios of the Intergovernmental Panel on Climate Change's Fourth Assessment Report, the relationship projects a sea-level rise ranging from 75 to 190 cm for the period 1990-2100.
|
| [3] |
|
| [4] |
Using a 25-y time series of precision satellite altimeter data from TOPEX/Poseidon, Jason-1, Jason-2, and Jason-3, we estimate the climate-change-driven acceleration of global mean sea level over the last 25 y to be 0.084 ± 0.025 mm/y Coupled with the average climate-change-driven rate of sea level rise over these same 25 y of 2.9 mm/y, simple extrapolation of the quadratic implies global mean sea level could rise 65 ± 12 cm by 2100 compared with 2005, roughly in agreement with the Intergovernmental Panel on Climate Change (IPCC) 5th Assessment Report (AR5) model projections.Copyright © 2018 the Author(s). Published by PNAS.
|
| [5] |
自然资源部. 2022年中国海平面公报[EB/OL].(2023-04-05)[2023-08-25]. http://gi.mnr.gov.cn/202304/t20230412_2781114.html.
|
| [6] |
We review the characteristics of sea level variability at the coast focussing on how it differs from the variability in the nearby deep ocean. Sea level variability occurs on all timescales, with processes at higher frequencies tending to have a larger magnitude at the coast due to resonance and other dynamics. In the case of some processes, such as the tides, the presence of the coast and the shallow waters of the shelves results in the processes being considerably more complex than offshore. However, 'coastal variability' should not always be considered as 'short spatial scale variability' but can be the result of signals transmitted along the coast from 1000s km away. Fortunately, thanks to tide gauges being necessarily located at the coast, many aspects of coastal sea level variability can be claimed to be better understood than those in the deep ocean. Nevertheless, certain aspects of coastal variability remain under-researched, including how changes in some processes (e.g., wave setup, river runoff) may have contributed to the historical mean sea level records obtained from tide gauges which are now used routinely in large-scale climate research.
|
| [7] |
|
| [8] |
|
| [9] |
|
| [10] |
|
| [11] |
|
| [12] |
|
| [13] |
|
| [14] |
|
| [15] |
|
| [16] |
|
| [17] |
|
| [18] |
|
| [19] |
|
| [20] |
|
| [21] |
|
| [22] |
|
| [23] |
|
| [24] |
|
| [25] |
|
| [26] |
|
| [27] |
|
| [28] |
Satellite radar altimetry provides a unique sea level data set that extends over more than 25 years back in time and that has an almost global coverage. However, when approaching the coasts, the extraction of correct sea level estimates is challenging due to corrupted waveforms and to errors in most of the corrections and in some auxiliary information used in the data processing. The development of methods dedicated to the improvement of altimeter data in the coastal zone dates back to the 1990s, but the major progress happened during the last decade thanks to progress in radar technology [e.g., synthetic aperture radar (SAR) mode and Ka-band frequency], improved waveform retracking algorithms, the availability of new/improved corrections (e.g., wet troposphere and tidal models) and processing workflows oriented to the coastal zone. Today, a set of techniques exists for the processing of coastal altimetry data, generally called "coastal altimetry." They have been used to generate coastal altimetry products. Altimetry is now recognized as part of the integrated observing system devoted to coastal sea level monitoring. In this article, we review the recent technical advances in processing and the new technological capabilities of satellite radar altimetry in the coastal zone. We also illustrate the fast-growing use of coastal altimetry data sets in coastal sea level research and applications, as high-frequency (tides and storm surge) and long-term sea level change studies.
|
| [29] |
|
| [30] |
|
| [31] |
|
| [32] |
|
| [33] |
|
| [34] |
郭金运, 王建波, 胡志博, 等. 由TOPEX/Poseidon和Jason-1/2探测的1993—2012中国海海平面时空变化[J]. 地球物理学报, 2015, 58(9):3103-3120.
海平面变化是社会经济发展和科学研究的重要内容.利用1993年1月至2012年12月共20年的TOPEX/Poseidon、Jason-1和Jason-2卫星测高数据,研究中国海海平面的时空变化.首先通过三颗卫星伴飞阶段数据得到三颗卫星之间的逐点海面高系统偏差,进行逐点海面高改正,建立了20年的中国海海面高异常时间序列.分析了中国海海面高异常空间分布,给出了1月到12月月均平均海平面异常的空间变化规律.分析了中国海海面高异常的时变规律,分别给出了年、季度和月的海面上升速率.利用小波分析研究了中国海海面高异常周期变化规律,分别给出了渤海、黄海、东海和南海的海面高变化周期.讨论了ENSO对海面高异常的影响.
|
| [35] |
李建成, 王正涛, 胡建国. 联合多种卫星测高数据分析全球和中国海海平面变化[J]. 武汉测绘科技大学学报, 2000, 25(4):343-347.
|
| [36] |
|
| [37] |
詹金刚, 王勇, 柳林涛. 中国近海海平面季节尺度变化的时频分析[J]. 地球物理学报, 2003, 46(1):36-41.
|
| [38] |
|
| [39] |
|
| [40] |
|
| [41] |
|
| [42] |
于振龙, 许东峰, 姚志雄, 等. 基于多变量LSTM神经网络模型的PDO指数预测研究[J]. 海洋学报, 2022, 44(6):58-67.
|
| [43] |
|
| [44] |
|
| [45] |
|
| [46] |
|
| [47] |
We analyze the sea level rise along the Bohai Sea, the Yellow Sea, the East China Sea, and the South China Sea (the "China Seas") coastline using 25 tide gauge records beginning with Macau in 1925, but with most starting during the 1950s and 60s. The main problem in estimating sea level rise for the period is the lack of vertical land movement (VLM) data for the tide gauge stations. We estimated VLM using satellite altimetry covering the 18 stations with records spanning 1993-2016. The results show that many tide gauge stations, typically in cities, have undergone significant subsidence due to groundwater extraction. After removing the VLM from tide gauge records, the 1993-2016 sea level rise rate is 3.2 +/- 1.1 mm/yr, and 2.9 +/- 0.8 mm/yr over the longer 1980-2016 period. We estimate the steric sea level contribution to be up to 0.9 +/- 0.3 mm/yr, and contributions from ice mass loss from glaciers and ice sheets of up to 1.1 +/- 0.1 mm/yr over the last 60 years. Contributions from VLM range between - 4.5 +/- 1.0 mm/yr and 1.4 +/- 1.3 mm/yr across the stations. Projections of coastal sea level probability distributions under future climate scenarios show that the steric factor is the main contributor under both the RCP 4.5 and High-end RCP 8.5 scenarios except in the upper tails under High-end RCP 8.5 when the Antarctic ice sheet makes the greatest contribution. By 2100 we expect median coastal sea level rises at the stations of 48-61 cm under RCP 4.5, and 84-99 cm under High-end RCP 8.5 scenario.
|
| [48] |
We propose a robust learning algorithm and apply it to recurrent neural networks. This algorithm is based on filtering outliers from the data and then estimating parameters from the filtered data. The filtering removes outliers from both the target function and the inputs of the neural network. The filtering is soft in that some outliers are neither completely rejected nor accepted. To show the need for robust recurrent networks, we compare the predictive ability of least squares estimated recurrent networks on synthetic data and on the Puget Power Electric Demand time series. These investigations result in a class of recurrent neural networks, NARMA(p,q), which show advantages over feedforward neural networks for time series with a moving average component. Conventional least squares methods of fitting NARMA(p,q) neural network models are shown to suffer a lack of robustness towards outliers. This sensitivity to outliers is demonstrated on both the synthetic and real data sets. Filtering the Puget Power Electric Demand time series is shown to automatically remove the outliers due to holidays. Neural networks trained on filtered data are then shown to give better predictions than neural networks trained on unfiltered time series.
|
| [49] |
|
| [50] |
|
/
| 〈 |
|
〉 |
