Search Results
You are looking at 1 - 6 of 6 items for :
- Author or Editor: Wei Li x
- Journal of Hydrometeorology x
- Refine by Access: All Content x
Abstract
The accurate prediction of surface soil moisture (SM) is crucial for understanding hydrological processes. Deep learning (DL) models such as the long short-term memory model (LSTM) provide a powerful method and have been widely used in SM prediction. However, few studies have notably high success rates due to lacking prior knowledge in forms such as causality. Here we present a new causality-structure-based LSTM model (CLSTM), which could learn time interdependency and causality information for hydrometeorological applications. We applied and compared LSTM and CLSTM methods for forecasting SM across 64 FLUXNET sites globally. The results showed that CLSTM dramatically increased the predictive performance compared with LSTM. The Nash–Sutcliffe efficiency (NSE) suggested that more than 67% of sites witnessed an improvement of SM simulation larger than 10%. It is highlighted that CLSTM had a much better generalization ability that can adapt to extreme soil conditions, such as SM response to drought and precipitation events. By incorporating causal relations, CLSTM increased predictive ability across different lead times compared to LSTM. We also highlighted the critical role of physical information in the form of causality structure to improve drought prediction. At the same time, CLSTM has the potential to improve predictions of other hydrometeorological variables.
Abstract
The accurate prediction of surface soil moisture (SM) is crucial for understanding hydrological processes. Deep learning (DL) models such as the long short-term memory model (LSTM) provide a powerful method and have been widely used in SM prediction. However, few studies have notably high success rates due to lacking prior knowledge in forms such as causality. Here we present a new causality-structure-based LSTM model (CLSTM), which could learn time interdependency and causality information for hydrometeorological applications. We applied and compared LSTM and CLSTM methods for forecasting SM across 64 FLUXNET sites globally. The results showed that CLSTM dramatically increased the predictive performance compared with LSTM. The Nash–Sutcliffe efficiency (NSE) suggested that more than 67% of sites witnessed an improvement of SM simulation larger than 10%. It is highlighted that CLSTM had a much better generalization ability that can adapt to extreme soil conditions, such as SM response to drought and precipitation events. By incorporating causal relations, CLSTM increased predictive ability across different lead times compared to LSTM. We also highlighted the critical role of physical information in the form of causality structure to improve drought prediction. At the same time, CLSTM has the potential to improve predictions of other hydrometeorological variables.
Abstract
Accurate spatiotemporal predictions of surface soil moisture (SM) are important for many critical applications. Machine learning models provide a powerful method for building an accurate and reliable predictive model of SM. However, the models used in recent studies have some limitations, including lack of spatial autocorrelation (SAC), vague representation of important features, and primarily focused on the one-step forecast. Thus, we proposed an attention-based convolutional long short-term memory model (AttConvLSTM) for multistep forecasting. The model includes three layers, spatial compression, axial attention, and encoder–decoder prediction, which are used for compressing spatial information, feature extraction, and multistep prediction, respectively. The model was trained using surface SM from the Soil Moisture Active Passive L4 product at 18-km spatial resolution over the United States. The results show that AttConvLSTM predicts 24 h ahead SM with mean R 2 and RMSE is equal to 0.82 and 0.02, respectively. Compared with LSTM, AttConvLSTM improves the model performance over 73.6% of regions, with an improvement of 8.4% and 17.4% in R2 and RMSE, respectively. The performance of the model is mainly influenced by temporal autocorrelation (TAC). Moreover, we also highlight the importance of SAC on model performance, especially over regions with high SAC and low TAC. Our model is also competent for SM predictions from several hours to several days, which could be a useful tool for predicting all meteorological variables and forecasting extremes.
Abstract
Accurate spatiotemporal predictions of surface soil moisture (SM) are important for many critical applications. Machine learning models provide a powerful method for building an accurate and reliable predictive model of SM. However, the models used in recent studies have some limitations, including lack of spatial autocorrelation (SAC), vague representation of important features, and primarily focused on the one-step forecast. Thus, we proposed an attention-based convolutional long short-term memory model (AttConvLSTM) for multistep forecasting. The model includes three layers, spatial compression, axial attention, and encoder–decoder prediction, which are used for compressing spatial information, feature extraction, and multistep prediction, respectively. The model was trained using surface SM from the Soil Moisture Active Passive L4 product at 18-km spatial resolution over the United States. The results show that AttConvLSTM predicts 24 h ahead SM with mean R 2 and RMSE is equal to 0.82 and 0.02, respectively. Compared with LSTM, AttConvLSTM improves the model performance over 73.6% of regions, with an improvement of 8.4% and 17.4% in R2 and RMSE, respectively. The performance of the model is mainly influenced by temporal autocorrelation (TAC). Moreover, we also highlight the importance of SAC on model performance, especially over regions with high SAC and low TAC. Our model is also competent for SM predictions from several hours to several days, which could be a useful tool for predicting all meteorological variables and forecasting extremes.
Abstract
Subseasonal to seasonal (S2S) weather forecasting has made significant advances and several products have been made available. However, to date few studies utilize these products to extend the hydrological forecast time range. This study evaluates S2S precipitation from eight model ensembles in the hydrological simulation of extreme events at the catchment scale. A superior bias correction method is used to correct the bias of S2S precipitation for hydrological forecasts, and the results are compared with direct bias correction of hydrological forecasts using raw precipitation forecasts as input. The study shows that the S2S models can skillfully forecast daily precipitation within a lead time of 11 days. The S2S precipitation data from the European Centre for Medium-Range Weather Forecasts (ECMWF), Korea Meteorological Administration (KMA), and United Kingdom’s Met Office (UKMO) models present lower mean error than that of other models and have higher correlation coefficients with observations. Precipitation data from the ECMWF, KMA, and UKMO models also perform better than that of other models in simulating multiple-day precipitation processes. The bias correction method effectively reduces the mean error of daily S2S precipitation for all models while also improving the correlation with observations. Moreover, this study found that the bias correction procedure can apply to either precipitation or streamflow simulations for improving the hydrological forecasts, even though the degree of improvement is dependent on the hydrological variables. Overall, S2S precipitation has a potential to be applied for hydrological forecasts, and a superior bias correction method can increase the forecasts’ reliability, although further studies are still needed to confirm its effect.
Abstract
Subseasonal to seasonal (S2S) weather forecasting has made significant advances and several products have been made available. However, to date few studies utilize these products to extend the hydrological forecast time range. This study evaluates S2S precipitation from eight model ensembles in the hydrological simulation of extreme events at the catchment scale. A superior bias correction method is used to correct the bias of S2S precipitation for hydrological forecasts, and the results are compared with direct bias correction of hydrological forecasts using raw precipitation forecasts as input. The study shows that the S2S models can skillfully forecast daily precipitation within a lead time of 11 days. The S2S precipitation data from the European Centre for Medium-Range Weather Forecasts (ECMWF), Korea Meteorological Administration (KMA), and United Kingdom’s Met Office (UKMO) models present lower mean error than that of other models and have higher correlation coefficients with observations. Precipitation data from the ECMWF, KMA, and UKMO models also perform better than that of other models in simulating multiple-day precipitation processes. The bias correction method effectively reduces the mean error of daily S2S precipitation for all models while also improving the correlation with observations. Moreover, this study found that the bias correction procedure can apply to either precipitation or streamflow simulations for improving the hydrological forecasts, even though the degree of improvement is dependent on the hydrological variables. Overall, S2S precipitation has a potential to be applied for hydrological forecasts, and a superior bias correction method can increase the forecasts’ reliability, although further studies are still needed to confirm its effect.
Abstract
The operational coupled land–atmosphere forecast model from the National Centers for Environmental Prediction (NCEP) is evaluated for the strength and characteristics of its coupling in the water cycle between land and atmosphere. Following the protocols of the Global Land–Atmosphere Coupling Experiment (GLACE) it is found that the Global Forecast System (GFS) atmospheric model coupled to the Noah land surface model exhibits extraordinarily weak land–atmosphere coupling, much as its predecessor, the GFS–Oregon State University (OSU) coupled system. The coupling strength is evaluated by the ability of subsurface soil wetness to affect locally the time series of precipitation. The surface fluxes in Noah are also found to be rather insensitive to subsurface soil wetness. Comparison to another atmospheric model coupled to Noah as well as a different land surface model show that Noah is responsible for some of the lack of sensitivity, primarily because its thick (10 cm) surface layer dominates the variability in surface latent heat fluxes. Noah is found to be as responsive as other land surface models to surface soil wetness and temperature variations, suggesting the design of the GLACE sensitivity experiment (based only on subsurface soil wetness) handicapped the Noah model. Additional experiments, in which the parameterization of evapotranspiration is altered, as well as experiments where surface soil wetness is also constrained, isolate the GFS atmospheric model as the principal source of the weak sensitivity of precipitation to land surface states.
Abstract
The operational coupled land–atmosphere forecast model from the National Centers for Environmental Prediction (NCEP) is evaluated for the strength and characteristics of its coupling in the water cycle between land and atmosphere. Following the protocols of the Global Land–Atmosphere Coupling Experiment (GLACE) it is found that the Global Forecast System (GFS) atmospheric model coupled to the Noah land surface model exhibits extraordinarily weak land–atmosphere coupling, much as its predecessor, the GFS–Oregon State University (OSU) coupled system. The coupling strength is evaluated by the ability of subsurface soil wetness to affect locally the time series of precipitation. The surface fluxes in Noah are also found to be rather insensitive to subsurface soil wetness. Comparison to another atmospheric model coupled to Noah as well as a different land surface model show that Noah is responsible for some of the lack of sensitivity, primarily because its thick (10 cm) surface layer dominates the variability in surface latent heat fluxes. Noah is found to be as responsive as other land surface models to surface soil wetness and temperature variations, suggesting the design of the GLACE sensitivity experiment (based only on subsurface soil wetness) handicapped the Noah model. Additional experiments, in which the parameterization of evapotranspiration is altered, as well as experiments where surface soil wetness is also constrained, isolate the GFS atmospheric model as the principal source of the weak sensitivity of precipitation to land surface states.
Abstract
Soil moisture influences precipitation mainly through its impact on land–atmosphere interactions. Understanding and correctly modeling soil moisture–precipitation (SM–P) coupling is crucial for improving weather forecasting and subseasonal to seasonal climate predictions, especially when predicting the persistence and magnitude of drought. However, the sign and spatial structure of SM–P feedback are still being debated in the climate research community, mainly due to the difficulty in establishing causal relationships and the high degree of nonlinearity in land–atmosphere processes. To this end, we developed a causal inference model based on the Granger causality analysis and a nonlinear machine learning model. This model includes three steps: nonlinear anomaly decomposition, nonlinear Granger causality analysis, and evaluation of the quality of SM–P feedback, which eliminates the nonlinear response of interannual and seasonal variability and the memory effects of climatic factors and isolates the causal relationship of local SM–P feedback. We applied this model by using National Climate Assessment–Land Data Assimilation System (NCA-LDAS) datasets over the United States. The results highlight the importance of nonlinear atmosphere responses in land–atmosphere interactions. In addition, the strong feedback over the southwestern United States and the Great Plains both highlight the impacts of topographic factors rather than only the sensitivity of evapotranspiration to soil moisture. Furthermore, the SM–P index defined by our framework is used to benchmark Earth system models (ESMs), which provides a new metric for efficiently identifying potential model biases in modeling local land–atmosphere interactions and may help the development of ESMs in improving simulations of water cycle variability and extremes.
Abstract
Soil moisture influences precipitation mainly through its impact on land–atmosphere interactions. Understanding and correctly modeling soil moisture–precipitation (SM–P) coupling is crucial for improving weather forecasting and subseasonal to seasonal climate predictions, especially when predicting the persistence and magnitude of drought. However, the sign and spatial structure of SM–P feedback are still being debated in the climate research community, mainly due to the difficulty in establishing causal relationships and the high degree of nonlinearity in land–atmosphere processes. To this end, we developed a causal inference model based on the Granger causality analysis and a nonlinear machine learning model. This model includes three steps: nonlinear anomaly decomposition, nonlinear Granger causality analysis, and evaluation of the quality of SM–P feedback, which eliminates the nonlinear response of interannual and seasonal variability and the memory effects of climatic factors and isolates the causal relationship of local SM–P feedback. We applied this model by using National Climate Assessment–Land Data Assimilation System (NCA-LDAS) datasets over the United States. The results highlight the importance of nonlinear atmosphere responses in land–atmosphere interactions. In addition, the strong feedback over the southwestern United States and the Great Plains both highlight the impacts of topographic factors rather than only the sensitivity of evapotranspiration to soil moisture. Furthermore, the SM–P index defined by our framework is used to benchmark Earth system models (ESMs), which provides a new metric for efficiently identifying potential model biases in modeling local land–atmosphere interactions and may help the development of ESMs in improving simulations of water cycle variability and extremes.
Abstract
Global warming and anthropogenic activities have imposed noticeable impacts on rainfall pattern changes at both spatial and temporal scales in recent decades. Systematic diagnosis of rainfall pattern changes is urgently needed at spatiotemporal scales for a deeper understanding of how climate change produces variations in rainfall patterns. The objective of this study was to identify rainfall pattern changes systematically under climate change at a subcontinental scale along a rainfall gradient ranging from 1800 to 200 mm yr−1 by analyzing centennial rainfall data covering 230 sites from 1910 to 2017 in the Northern Territory of Australia. Rainfall pattern changes were characterized by considering aspects of trends and periodicity of annual rainfall, abrupt changes, rainfall distribution, and extreme rainfall events. Our results illustrated that rainfall patterns in northern Australia have changed significantly compared with the early period of the twentieth century. Specifically, 1) a significant increasing trend in annual precipitation associated with greater variation in recent decades was observed over the entire study area, 2) temporal variations represented a mean rainfall periodicity of 27 years over wet to dry regions, 3) an abrupt change of annual rainfall amount occurred consistently in both humid and arid regions during the 1966–75 period, and 4) partitioned long-term time series of rainfall demonstrated a wetter rainfall distribution trend across coastal to inland areas that was associated with more frequent extreme rainfall events in recent decades. The findings of this study could facilitate further studies on the mechanisms of climate change that influence rainfall pattern changes.
Significance Statement
Characterizing long-term rainfall pattern changes under different rainfall conditions is important to understand the impacts of climate change. We conducted diagnosis of centennial rainfall pattern changes across wet to dry regions in northern Australia and found that rainfall patterns have noticeably changed in recent decades. The entire region has a consistent increasing trend of annual rainfall with higher variation. Meanwhile, the main shifting period of rainfall pattern was during 1966–75. Although annual rainfall seems to become wetter with an increasing trend, more frequent extreme rainfall events should also be noticed for assessing the impacts of climate changes. The findings support further study to understand long-term rainfall pattern changes under climate change.
Abstract
Global warming and anthropogenic activities have imposed noticeable impacts on rainfall pattern changes at both spatial and temporal scales in recent decades. Systematic diagnosis of rainfall pattern changes is urgently needed at spatiotemporal scales for a deeper understanding of how climate change produces variations in rainfall patterns. The objective of this study was to identify rainfall pattern changes systematically under climate change at a subcontinental scale along a rainfall gradient ranging from 1800 to 200 mm yr−1 by analyzing centennial rainfall data covering 230 sites from 1910 to 2017 in the Northern Territory of Australia. Rainfall pattern changes were characterized by considering aspects of trends and periodicity of annual rainfall, abrupt changes, rainfall distribution, and extreme rainfall events. Our results illustrated that rainfall patterns in northern Australia have changed significantly compared with the early period of the twentieth century. Specifically, 1) a significant increasing trend in annual precipitation associated with greater variation in recent decades was observed over the entire study area, 2) temporal variations represented a mean rainfall periodicity of 27 years over wet to dry regions, 3) an abrupt change of annual rainfall amount occurred consistently in both humid and arid regions during the 1966–75 period, and 4) partitioned long-term time series of rainfall demonstrated a wetter rainfall distribution trend across coastal to inland areas that was associated with more frequent extreme rainfall events in recent decades. The findings of this study could facilitate further studies on the mechanisms of climate change that influence rainfall pattern changes.
Significance Statement
Characterizing long-term rainfall pattern changes under different rainfall conditions is important to understand the impacts of climate change. We conducted diagnosis of centennial rainfall pattern changes across wet to dry regions in northern Australia and found that rainfall patterns have noticeably changed in recent decades. The entire region has a consistent increasing trend of annual rainfall with higher variation. Meanwhile, the main shifting period of rainfall pattern was during 1966–75. Although annual rainfall seems to become wetter with an increasing trend, more frequent extreme rainfall events should also be noticed for assessing the impacts of climate changes. The findings support further study to understand long-term rainfall pattern changes under climate change.
