Characterizing Soil Moisture Memory by Soil Moisture Autocorrelation

Abstract

Abstract- An autocorrelation value only implies the strength of the soil moisture memory (SMM) without specifying its statistical significance. This study proposed a conversion of autocorrelation values into a time scale considering statistical significance at 95% confidence level. It computed a SMM time scale of 56.13 days for the Spoon river basin, Illinois, and was highly consistent with previous regional estimations (1.8-2.1 months). Additionally, the time scales showed a fair correlation with the 30-day-lagged autocorrelations ( R 2 =0.58). This time scale is easily understandable and provides more information than simple autocorrelations, suggesting significant autocorrelations deal with just a single number. Keywords-Soil Moisture Autocorrelation; Soil Moisture Memory; Soil Moisture Persistence; Soil Moisture Memory Timescale; Xinanjiang Model I. INTRODUCTION Soil moisture isa very important component in weather and climate prediction due to its persistence characteristics [1-8]. Soils can act as a temporary reservoir and hasa tendency to accumulate atmospheric forcing anomalies (deviation from the mean state). For example, an extended period of intense rainfall results ina positive anomaly in the soil moisture state. This anomaly then dissipates through evapotranspiration or runoff. Similarly, any negative anomaly created by a lengthy dry spell is dissipated through rainfall, snowmelt, or irrigation. However, this dissipation process and time scales differ from place to place due to variations in soil properties, scale of interest, and climate forcing.Literature suggests that the complete dissipation process can take hours or months [2, 9]. Therefore, the soil can remember an anomalous condition long after ithas occurred. This phenomenon of memorizing past anomalies is termed “soil moisture persistence” or “soil moisture memory” (hereinafter referred as SMM). The extentof this SMM has potential application in different research fields, primarilyin weather and climate or hydrological predictions. Knowledge of SMM has proven to bean additional toolfor improvingtraditional climate/hydrological forecast efficiency[3, 10]. Advantages of the SMM inclusive approach over traditional forecasts have been reported in several studies. [3, 11] documented the usefulness of SMM understandingto improve soil moisture initialization for weather and climate prediction processes. SMM knowledge can enhance prediction efficiency, mainly in reference to longer time scales and under high initial soil moisture anomaly conditions. Such knowledge mightalso improve the predictability of soil moisture and associated climate [5]. Moreover, soil moisture, through its influence on land energy balance (partitioning sensible and latent heat flux), provides additional feedback on temperature [12, 13] and precipitation [10, 14]. Furthermore, a persistent soil moisture anomaly may prolong the effects of drought [15, 16] and influence the magnitude, occurrence, and receding of floods [17] and streamflow dynamics. Low-frequency streamflow dynamicsare believed to be controlled by the catchment wetness, SMM [18]. Likewise, SMM is believed to be capable of propagating stream flows [19].Currently, streamflow-forecasting efficiency mostly depends on the prediction accuracy of atmospheric forcing or snow accumulation. However, soil moisture conditions evidently affect streamflow forecast accuracy. A dry soil below the snowpack receives and stores infiltrated water from snowmelt and later it evaporates rather than be coming runoff into streams. Conversely,wet soil beneaththe melting snowpack canstimulate streamflow [20]. Therefore, SMMpotentially affects streamflow forecast skills [20, 21]. II.