An official website of the United States government
Abstract. Climate change threatens our ability to grow food for an ever-increasing population. There is aneed for high-quality soil moisture predictions in under-monitored regionslike Africa. However, it is unclear if soil moisture processes are globallysimilar enough to allow our models trained on available in situ data tomaintain accuracy in unmonitored regions. We present a multitask longshort-term memory (LSTM) model that learns simultaneously from globalsatellite-based data and in situ soil moisture data. This model is evaluated inboth random spatial holdout mode and continental holdout mode (trained onsome continents, tested on a different one). The model compared favorably tocurrent land surface models, satellite products, and a candidate machinelearning model, reaching a global median correlation of 0.792 for the randomspatial holdout test. It behaved surprisingly well in Africa and Australia,showing high correlation even when we excluded their sites from the trainingset, but it performed relatively poorly in Alaska where rapid changes areoccurring. In all but one continent (Asia), the multitask model in theworst-case scenario test performed better than the soil moisture activepassive (SMAP) 9 km product. Factorial analysis has shown that the LSTM model'saccuracy varies with terrain aspect, resulting in lower performance for dryand south-facing slopes or wet and north-facing slopes. This knowledgehelps us apply the model while understanding its limitations. This model isbeing integrated into an operational agricultural assistance applicationwhich currently provides information to 13 million African farmers.
more »
« less
A Multiscale Deep Learning Model for Soil Moisture Integrating Satellite and In Situ Data
Abstract Deep learning (DL) models trained on hydrologic observations can perform extraordinarily well, but they can inherit deficiencies of the training data, such as limited coverage of in situ data or low resolution/accuracy of satellite data. Here we propose a novel multiscale DL scheme learning simultaneously from satellite and in situ data to predict 9 km daily soil moisture (5 cm depth). Based on spatial cross‐validation over sites in the conterminous United States, the multiscale scheme obtained a median correlation of 0.901 and root‐mean‐square error of 0.034 m3/m3. It outperformed the Soil Moisture Active Passive satellite mission's 9 km product, DL models trained on in situ data alone, and land surface models. Our 9 km product showed better accuracy than previous 1 km satellite downscaling products, highlighting limited impacts of improving resolution. Not only is our product useful for planning against floods, droughts, and pests, our scheme is generically applicable to geoscientific domains with data on multiple scales, breaking the confines of individual data sets.
more »
« less
- PAR ID:
- 10381069
- Publisher / Repository:
- DOI PREFIX: 10.1029
- Date Published:
- Journal Name:
- Geophysical Research Letters
- Volume:
- 49
- Issue:
- 7
- ISSN:
- 0094-8276
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract The Consistent Artificial Intelligence (AI)-based Soil Moisture (CASM) dataset is a global, consistent, and long-term, remote sensing soil moisture (SM) dataset created using machine learning. It is based on the NASA Soil Moisture Active Passive (SMAP) satellite mission SM data and is aimed at extrapolating SMAP-like quality SM back in time using previous satellite microwave platforms. CASM represents SM in the top soil layer, and it is defined on a global 25 km EASE-2 grid and for 2002–2020 with a 3-day temporal resolution. The seasonal cycle is removed for the neural network training to ensure its skill is targeted at predicting SM extremes. CASM comparison to 367 globalin-situSM monitoring sites shows a SMAP-like median correlation of 0.66. Additionally, the SM product uncertainty was assessed, and both aleatoric and epistemic uncertainties were estimated and included in the dataset. CASM dataset can be used to study a wide range of hydrological, carbon cycle, and energy processes since only a consistent long-term dataset allows assessing changes in water availability and water stress.more » « less
-
null (Ed.)Abstract. Soil moisture is key for understandingsoil–plant–atmosphere interactions. We provide a soil moisture patternrecognition framework to increase the spatial resolution and fill gaps ofthe ESA-CCI (European Space Agency Climate Change Initiative v4.5) soilmoisture dataset, which contains > 40 years of satellite soilmoisture global grids with a spatial resolution of ∼ 27 km. Weuse terrain parameters coupled with bioclimatic and soil type information topredict finer-grained (i.e., downscaled) satellite soil moisture. We assessthe impact of terrain parameters on the prediction accuracy bycross-validating downscaled soil moisture with and without the support ofbioclimatic and soil type information. The outcome is a dataset of gap-freeglobal mean annual soil moisture predictions and associated predictionvariances for 28 years (1991–2018) across 15 km grids. We use independent in siturecords from the International Soil Moisture Network (ISMN, 987 stations)and in situ precipitation records (171 additional stations) only for evaluating thenew dataset. Cross-validated correlation between observed and predicted soilmoisture values varies from r= 0.69 to r= 0.87 with root mean squarederrors (RMSEs, m3 m−3) around 0.03 and 0.04. Our soil moisturepredictions improve (a) the correlation with the ISMN (when compared withthe original ESA-CCI dataset) from r= 0.30 (RMSE = 0.09, unbiased RMSE (ubRMSE) = 0.37) tor= 0.66 (RMSE = 0.05, ubRMSE = 0.18) and (b) the correlation with local precipitation records across boreal (from r= < 0.3 up to r= 0.49) ortropical areas (from r= < 0.3 to r= 0.46) which are currentlypoorly represented in the ISMN. Temporal trends show a decline of globalannual soil moisture using (a) data from the ISMN (-1.5[-1.8,-1.24] %),(b) associated locations from the original ESA-CCI dataset (-0.87[-1.54,-0.17] %), (c) associated locations from predictions based on terrainparameters (-0.85[-1.01,-0.49] %), and (d) associated locations frompredictions including bioclimatic and soil type information (-0.68[-0.91,-0.45] %). We provide a new soil moisture dataset that has no gaps andhigher granularity together with validation methods and a modeling approachthat can be applied worldwide (Guevara et al., 2020,https://doi.org/10.4211/hs.9f981ae4e68b4f529cdd7a5c9013e27e).more » « less
-
The Soil Moisture Active Passive (SMAP) mission measures important soil moisture data globally. SMAP's products might not always perform better than land surface models (LSM) when evaluated against in situ measurements. However, we hypothesize that SMAP presents added value for long-term soil moisture estimation in a data fusion setting as evaluated by in situ data. Here, with the help of a time series deep learning (DL) method, we created a seamlessly extended SMAP data set to test this hypothesis and, importantly, gauge whether such benefits extend to years beyond SMAP's limited lifespan. We first show that the DL model, called long short-term memory (LSTM), can extrapolate SMAP for several years and the results are similar to the training period. We obtained prolongation results with low-performance degradation where SMAP itself matches well with in situ data. Interannual trends of root-zone soil moisture are surprisingly well captured by LSTM. In some cases, LSTM's performance is limited by SMAP, whose main issue appears to be its shallow sensing depth. Despite this limitation, a simple average between LSTM and an LSM Noah frequently outperforms Noah alone. Moreover, Noah combined with LSTM is more skillful than when it is combined with another LSM. Over sparsely instrumented sites, the Noah-LSTM combination shows a stronger edge. Our results verified the value of LSTM-extended SMAP data. Moreover, DL is completely data driven and does not require structural assumptions. As such, it has its unique potential for long-term projections and may be applied synergistically with other model-data integration techniques.more » « less
-
Abstract The changing thermal state of permafrost is an important indicator of climate change in northern high latitude ecosystems. The seasonally thawed soil active layer thickness (ALT) overlying permafrost may be deepening as a consequence of enhanced polar warming and widespread permafrost thaw in northern permafrost regions (NPRs). The associated increase in ALT may have cascading effects on ecological and hydrological processes that impact climate feedback. However, past NPR studies have only provided a limited understanding of the spatially continuous patterns and trends of ALT due to a lack of long-term high spatial resolution ALT data across the NPR. Using a suite of observational biophysical variables and machine learning (ML) techniques trained with availablein situALT network measurements (n= 2966 site-years), we produced annual estimates of ALT at 1 km resolution over the NPR from 2003 to 2020. Our ML-derived ALT dataset showed high accuracy (R2= 0.97) and low bias when compared within situALT observations. We found the ALT distribution to be most strongly affected by local soil properties, followed by topographic elevation and land surface temperatures. Pair-wise site-level evaluation between our data-driven ALT with Circumpolar Active Layer Monitoring data indicated that about 80% of sites had a deepening ALT trend from 2003 to 2020. Based on our long-term gridded ALT data, about 65% of the NPR showed a deepening ALT trend, while the entire NPR showed a mean deepening trend of 0.11 ± 0.35 cm yr−1[25%–75% quantile: (−0.035, 0.204) cm yr−1]. The estimated ALT trends were also sensitive to fire disturbance. Our new gridded ALT product provides an observationally constrained, updated understanding of the progression of thawing and the thermal state of permafrost in the NPR, as well as the underlying environmental drivers of these trends.more » « less
