A Causal Inference Model Based on Random Forests to Identify the Effect of Soil Moisture on Precipitation

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.

Development of observation-based global multilayer soil moisture products for 1970 to 2016

“…The high performance variability of merged products across space (Fig. S5) is consistent with previous studies (Beck et al, 2021;Karthikeyan et al, 2017;Li et al, 2020b;Wang et al, 2021a;Yuan and Quiring, 2017). The high RM-SEs of the merged products in the shallower soil layers across the water bodies and evergreen needleleaf forests (Fig.…”

“…The ranges of performance metrics of the new datasets against in situ data (Figs. 2 and S5) were broadly within the estimates reported by previous SM evaluations, although making a strict comparison is difficult because of the widely different spatiotemporal coverages and resolutions (Beck et al, 2021;Karthikeyan et al, 2017;Li et al, 2020b;Wang et al, 2021a;Yuan and Quiring, 2017). These results demonstrated that the merging procedures (unweighted averaging, OLC, EC) used were effective in creating relatively accurate long-term multilayer SM data at the global scale.…”

“…High-quality global soil moisture (SM) datasets benefit many applications, such as understanding drought changes and ecosystem dynamics (Green et al, 2019;Kumar et al, 2019), studying land-atmosphere feedbacks (Li et al, 2020a), benchmarking model capabilities (Loew et al, 2013), and initializing weather and climate forecast systems (Sospedra-Alfonso and Merryfield, 2018). The majority of SM products fall into five categories: in situ measurements, satellite observations, offline land surface model (LSM) simulations, reanalysis, and Earth system model (ESM) simulations.…”

Characterization of the Coherence Between Soil Moisture and Precipitation at Regional Scales

“…Li, Shangguan, et al. (2020) had shown that the topographic factors highlighted the SM–P interactions in the southwestern United States. By comparison, at all frequencies, GCCs were significantly correlated with meteorological factors in Nebraska, while soil texture was more important in determining GCCs in Utah.…”

“…To address the above questions, two regions with different climatic conditions (e.g., from semiarid and subhumid climates in Nebraska to arid and semiarid climates in Utah; T. Wang et al., 2017a) were selected from the continental United States, which have been hotspots for studying SM–P interactions (L. Chen & Dirmeyer, 2017; Ford, Rapp, et al., 2015; Koster et al., 2004; Li, Shangguan, et al., 2020). Observed long‐term SM and P data were obtained from regional monitoring networks in both regions, namely the Nebraska Mesonet (NM) for Nebraska and part of the Soil Climate Analysis Network (SCAN) for Utah.…”