Near-Real-Time Forecast of Satellite-Based Soil Moisture Using Long Short-Term Memory with an Adaptive Data Integration Kernel

Kuai Fang, Chaopeng Shen

Journal of Hydrometeorology · 2020 · 179 citations · 38 references

DOIFull text

Open access

Concepts

TL;DR

Near‑real‑time soil‑moisture forecasts from SMAP can aid hazard monitoring and agricultural planning. The study aims to deliver high‑fidelity NRT soil‑moisture forecasts by augmenting LSTM with a novel data‑integration kernel that assimilates the latest SMAP observations, and to estimate the random component of SMAP retrieval error. The adaptive DI kernel incorporates irregular SMAP observations into an LSTM framework, and the DI‑LSTM is benchmarked against a projection LSTM without data integration. On the CONUS, the DI‑LSTM achieved unprecedented accuracy, outperforming existing NRT forecasts—especially at longer lead times—by eliminating autocorrelated forcing errors and leveraging additional inputs and training data.

Abstract

Abstract Nowcasts, or near-real-time (NRT) forecasts, of soil moisture based on the Soil Moisture Active and Passive (SMAP) mission could provide substantial value for a range of applications including hazards monitoring and agricultural planning. To provide such a NRT forecast with high fidelity, we enhanced a time series deep learning architecture, long short-term memory (LSTM), with a novel data integration (DI) kernel to assimilate the most recent SMAP observations as soon as they become available. The kernel is adaptive in that it can accommodate irregular observational schedules. Testing over the CONUS, this NRT forecast product showcases predictions with unprecedented accuracy when evaluated against subsequent SMAP retrievals. It showed smaller error than NRT forecasts reported in the literature, especially at longer forecast latency. The comparative advantage was due to LSTM’s structural improvements, as well as its ability to utilize more input variables and more training data. The DI-LSTM was compared to the original LSTM model that runs without data integration, referred to as the projection model here. We found that the DI procedure removed the autocorrelated effects of forcing errors and errors due to processes not represented in the inputs, for example, irrigation and floodplain/lake inundation, as well as mismatches due to unseen forcing conditions. The effects of this purely data-driven DI kernel are discussed for the first time in the geosciences. Furthermore, this work presents an upper-bound estimate for the random component of the SMAP retrieval error.

References

38