Prolongation of SMAP to Spatiotemporally Seamless Coverage of Continental U.S. Using a Deep Learning Neural Network

The Soil Moisture Active Passive (SMAP) mission has delivered valuable sensing of surface soil moisture since 2015. However, it has a short time span and irregular revisit schedules. Utilizing a state-of-the-art time series deep learning neural network, Long Short-Term Memory (LSTM), we created a system that predicts SMAP level-3 moisture product with atmospheric forcings, model-simulated moisture, and static physiographic attributes as inputs. The system removes most of the bias with model simulations and improves predicted moisture climatology, achieving small test root-mean-square errors ( 0.87 for over 75% of Continental United States, including the forested southeast. As the first application of LSTM in hydrology, we show the proposed network avoids overfitting and is robust for both temporal and spatial extrapolation tests. LSTM generalizes well across regions with distinct climates and environmental settings. With high fidelity to SMAP, LSTM shows great potential for hindcasting, data assimilation, and weather forecasting.

[1]  Chaopeng Shen,et al.  Full‐flow‐regime storage‐streamflow correlation patterns provide insights into hydrologic functioning over the continental US , 2017 .

[2]  W. J. Steenburgh,et al.  Regional Soil Moisture Biases and Their Influence on WRF Model Temperature Forecasts over the Intermountain West , 2016 .

[3]  Heather McNairn,et al.  Soil Moisture Active Passive (SMAP) Project Calibration and Validation for the L2/3_SM_P Beta-Release Data Products Version 2 , 2015 .

[4]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[5]  Geoffrey E. Hinton,et al.  Deep Learning , 2015, Nature.

[6]  Steven M Quiring,et al.  Confronting weather and climate models with observational data from soil moisture networks over the United States. , 2016, Journal of hydrometeorology.

[7]  Steven M. Quiring,et al.  Comparison of NLDAS-2 Simulated and NASMD Observed Daily Soil Moisture. Part I: Comparison and Analysis , 2015 .

[8]  D. Lawrence,et al.  Improving the representation of hydrologic processes in Earth System Models , 2015 .

[9]  Praveen Kumar,et al.  A catchment‐based approach to modeling land surface processes in a general circulation model: 1. Model structure , 2000 .

[10]  Paul Voosen,et al.  The AI detectives. , 2017, Science.

[11]  Zoubin Ghahramani,et al.  A Theoretically Grounded Application of Dropout in Recurrent Neural Networks , 2015, NIPS.

[12]  Jiancheng Shi,et al.  The Soil Moisture Active Passive (SMAP) Mission , 2010, Proceedings of the IEEE.

[13]  D. Entekhabi,et al.  Soil Moisture Active Passive (SMAP) Project Calibration and Validation for the L2/3_SM_P , 2015 .

[14]  R. Tibshirani.  Regression Shrinkage and Selection via the Lasso , 1996 .

[15]  S. Quiring,et al.  Evaluation of soil moisture in CMIP5 simulations over the contiguous United States using in situ and satellite observations , 2016 .

[16]  N. Batjes,et al.  A homogenized soil data file for global environmental research: A subset of FAO, ISRIC and NRCS profiles (Version 1.0) , 1995 .

[17]  Jürgen Schmidhuber,et al.  LSTM: A Search Space Odyssey , 2015, IEEE Transactions on Neural Networks and Learning Systems.

[18]  Kelly K. Caylor,et al.  Validation of SMAP surface soil moisture products with core validation sites , 2017, Remote Sensing of Environment.

[19]  Jesse E. Bell,et al.  An Evaluation of the North American Regional Reanalysis Simulated Soil Moisture Conditions during the 2011–13 Drought Period , 2017 .

[20]  Zachary Chase Lipton.  A Critical Review of Recurrent Neural Networks for Sequence Learning , 2015, ArXiv.

[21]  Bart Nijssen,et al.  Global Retrospective Estimation of Soil Moisture Using the Variable Infiltration Capacity Land Surface Model, 1980–93 , 2001 .

[22]  Sonia I. Seneviratne,et al.  Analysis of soil moisture memory from observations in Europe , 2012 .

[23]  M. Borga,et al.  Flash flood warning based on rainfall thresholds and soil moisture conditions: An assessment for gauged and ungauged basins , 2008 .

[24]  Wojciech Zaremba,et al.  Recurrent Neural Network Regularization , 2014, ArXiv.

[25]  Chaopeng Shen,et al.  Improving Budyko curve‐based estimates of long‐term water partitioning using hydrologic signatures from GRACE , 2016 .

[26]  R. Koster,et al.  A Data-Driven Approach for Daily Real-Time Estimates and Forecasts of Near-Surface Soil Moisture. , 2017, Journal of hydrometeorology.

[27]  D. Lawrence,et al.  Regions of Strong Coupling Between Soil Moisture and Precipitation , 2004, Science.

[28]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[29]  Randal D. Koster,et al.  The components of a 'SVAT' scheme and their effects on a GCM's hydrological cycle , 1994 .

[30]  R. Srinivasan,et al.  Development and evaluation of Soil Moisture Deficit Index (SMDI) and Evapotranspiration Deficit Index (ETDI) for agricultural drought monitoring , 2005 .

[31]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[32]  Jingjing Xie,et al.  Daily reservoir inflow forecasting using multiscale deep feature learning with hybrid models , 2016 .

[33]  J. D. Tarpley,et al.  Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model , 2003 .