A Hybrid Artificial Neural Network to Estimate Soil Moisture Using SWAT+ and SMAP Data
In this work, we developed a data-driven framework to predict near-surface (0–5 cm) soil moisture (SM) by mapping inputs from the Soil & Water Assessment Tool to SM time series from NASA’s Soil Moisture Active Passive (SMAP) satellite for the period 1 January 2016–31 December 2018. We developed...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-08-01
|
| Series: | Machine Learning and Knowledge Extraction |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2504-4990/2/3/16 |
| _version_ | 1797556295128055808 |
|---|---|
| author | Katherine H. Breen Scott C. James Joseph D. White Peter M. Allen Jeffery G. Arnold |
| author_facet | Katherine H. Breen Scott C. James Joseph D. White Peter M. Allen Jeffery G. Arnold |
| author_sort | Katherine H. Breen |
| collection | DOAJ |
| description | In this work, we developed a data-driven framework to predict near-surface (0–5 cm) soil moisture (SM) by mapping inputs from the Soil & Water Assessment Tool to SM time series from NASA’s Soil Moisture Active Passive (SMAP) satellite for the period 1 January 2016–31 December 2018. We developed a hybrid artificial neural network (ANN) combining long short-term memory and multilayer perceptron networks that were used to simultaneously incorporate dynamic weather and static spatial data into the training algorithm, respectively. We evaluated the generalizability of the hybrid ANN using training datasets comprising several watersheds with different environmental conditions, examined the effects of standard and physics-guided loss functions, and experimented with feature augmentation. Our model could estimate SM on par with the accuracy of SMAP. We demonstrated that the most critical learning of the physical processes governing SM variability was learned from meteorological time series, and that additional physical context supported model performance when test data were not fully encapsulated by the variability of the training data. Additionally, we found that when forecasting SM based on trends learned during the earlier training period, the models appreciated seasonal trends. |
| first_indexed | 2024-03-10T17:00:46Z |
| format | Article |
| id | doaj.art-5f36300575554da5bc76c6853515b3b9 |
| institution | Directory Open Access Journal |
| issn | 2504-4990 |
| language | English |
| last_indexed | 2024-03-10T17:00:46Z |
| publishDate | 2020-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Machine Learning and Knowledge Extraction |
| spelling | doaj.art-5f36300575554da5bc76c6853515b3b92023-11-20T10:58:47ZengMDPI AGMachine Learning and Knowledge Extraction2504-49902020-08-012328330610.3390/make2030016A Hybrid Artificial Neural Network to Estimate Soil Moisture Using SWAT+ and SMAP DataKatherine H. Breen0Scott C. James1Joseph D. White2Peter M. Allen3Jeffery G. Arnold4Department of Geosciences, Baylor University, Waco, TX 76798, USADepartment of Geosciences, Baylor University, Waco, TX 76798, USADepartment of Biology, Baylor University, Waco, TX 76798, USADepartment of Geosciences, Baylor University, Waco, TX 76798, USAUSDA-Agricultural Research Service, Temple, TX 76502, USAIn this work, we developed a data-driven framework to predict near-surface (0–5 cm) soil moisture (SM) by mapping inputs from the Soil & Water Assessment Tool to SM time series from NASA’s Soil Moisture Active Passive (SMAP) satellite for the period 1 January 2016–31 December 2018. We developed a hybrid artificial neural network (ANN) combining long short-term memory and multilayer perceptron networks that were used to simultaneously incorporate dynamic weather and static spatial data into the training algorithm, respectively. We evaluated the generalizability of the hybrid ANN using training datasets comprising several watersheds with different environmental conditions, examined the effects of standard and physics-guided loss functions, and experimented with feature augmentation. Our model could estimate SM on par with the accuracy of SMAP. We demonstrated that the most critical learning of the physical processes governing SM variability was learned from meteorological time series, and that additional physical context supported model performance when test data were not fully encapsulated by the variability of the training data. Additionally, we found that when forecasting SM based on trends learned during the earlier training period, the models appreciated seasonal trends.https://www.mdpi.com/2504-4990/2/3/16physics-guided machine learninghybrid architectureLSTMdata-driven science |
| spellingShingle | Katherine H. Breen Scott C. James Joseph D. White Peter M. Allen Jeffery G. Arnold A Hybrid Artificial Neural Network to Estimate Soil Moisture Using SWAT+ and SMAP Data Machine Learning and Knowledge Extraction physics-guided machine learning hybrid architecture LSTM data-driven science |
| title | A Hybrid Artificial Neural Network to Estimate Soil Moisture Using SWAT+ and SMAP Data |
| title_full | A Hybrid Artificial Neural Network to Estimate Soil Moisture Using SWAT+ and SMAP Data |
| title_fullStr | A Hybrid Artificial Neural Network to Estimate Soil Moisture Using SWAT+ and SMAP Data |
| title_full_unstemmed | A Hybrid Artificial Neural Network to Estimate Soil Moisture Using SWAT+ and SMAP Data |
| title_short | A Hybrid Artificial Neural Network to Estimate Soil Moisture Using SWAT+ and SMAP Data |
| title_sort | hybrid artificial neural network to estimate soil moisture using swat and smap data |
| topic | physics-guided machine learning hybrid architecture LSTM data-driven science |
| url | https://www.mdpi.com/2504-4990/2/3/16 |
| work_keys_str_mv | AT katherinehbreen ahybridartificialneuralnetworktoestimatesoilmoistureusingswatandsmapdata AT scottcjames ahybridartificialneuralnetworktoestimatesoilmoistureusingswatandsmapdata AT josephdwhite ahybridartificialneuralnetworktoestimatesoilmoistureusingswatandsmapdata AT petermallen ahybridartificialneuralnetworktoestimatesoilmoistureusingswatandsmapdata AT jefferygarnold ahybridartificialneuralnetworktoestimatesoilmoistureusingswatandsmapdata AT katherinehbreen hybridartificialneuralnetworktoestimatesoilmoistureusingswatandsmapdata AT scottcjames hybridartificialneuralnetworktoestimatesoilmoistureusingswatandsmapdata AT josephdwhite hybridartificialneuralnetworktoestimatesoilmoistureusingswatandsmapdata AT petermallen hybridartificialneuralnetworktoestimatesoilmoistureusingswatandsmapdata AT jefferygarnold hybridartificialneuralnetworktoestimatesoilmoistureusingswatandsmapdata |