Study On Vertical Correlation And Forecasting Of Soil Moisture

Soil water is of critical importance to soil evolution and land productive forces,furthermore, can restrict formation and development of vegetation, thus is considered tobe a kind of important influence factor to ecosystem. The research about vertical spatialdistribution of soil moisture is the foundation of temporal and spatial correlation of soilmoisture. Studying and investigating temporal and spatial correlation of soil moisture, isnot only useful for us to further understand the rule of temporal and spatial distribution ofsoil moisture, but also can provide some effective ways to forecast and simulate soilmoisture change of more closer actual situation. This experiment was carried on 4 kindsof different treatments, respectively composed to naked land, forestland, grassland andfarmland on the experiment station in xianning Hubei. We measured daily soil watercontent by using PR1 at different depths of 10cm,20cm, 30cm, 40cm and 60cm in sixmouths :November, December, March, April, July, August ,and recorded weatherinformation and growth course of vegetation above ground simultaneously. Based on theabove content, the aim of this paper was to know spatial correlation of soil moisture andits form reasons according to combine land use and fertilization management .withrespect to research model, prediction and simulation for soil water change in deeperlayers was investigated make use of 4 kinds of different methods: PolynomialDistributed Lag (PDL) and Auto-regressive Distributed Lag (ADL) method of timeseries ,conventional method and data assimilation of physics model. The main findingsare as follows.1. The vertical distribution characteristic of soil moisture is that soil moisture contentincreased as soil depth increased, soil water content in depth layers changed along withthe surface layer water content changed, and its fluctuate range is gradually diminishwhen soil depth increased. The spatial distribution of soil water is closely related tometeorology, land use and fertilization management. All of these influence factors havean impact on soil moisture pattern but some will be more important than others in aparticular setting. For example, meteorology is the most important influence factor for thespatial distribution and the dynamic change of soil moisture content, vegetation factor issecondary, it is well know that vegetation affect soil water content of different soil layerschange depend on its root distribution. Fertilization measures impact on the spatialdistribution and dynamic change of soil moisture content in soil depths of 10cm and 20cm,but has no obvious effect on deeper layers. 2. Soil moisture under vertical direction had strong temporal and spatial correlation.Soil moisture under vertical direction at different vegetation coverage conditions anddifferent fertilizations had strong temporal and spatial correlation. Temporal and spatialcorrelation of Soil moisture under vertical direction was affected by some factor such asmeteorology, land use and fertilization measures. All siol moisture in different layers hastemporal and spatial correlation when soil moisture fluctuates of surface layer becausestability, but influence ability of each factor is not the same. For example, meteorologyhas most strong impact on temporal and spatial correlation of soil water content. Land useand fertilization measures have no effect on temporal and spatial correlation of soil watercontent when soil moisture fluctuate of surface layer because stability, but if the fluctuateis acute, their influence was evident. Compared with meteorology, vegetation has morestrong impact on temporal and spatial correlation of soil water content at different soillayers. However, Fertilization measures can decrease the trend of the correlation.3. On the basis of simulating and forecasting the soil water content with the methodsof time series, we found the precision of the forecasting Autoregressive Distributed Lagmodels was better than PDL. The results showed that the predicted value of ADL withsimplified models fitted the measured value fairly good, and could more actually reflectthe situation of soil moisture distribution.4. Comparing to conventional simulation, the prediction value of Data assimilationwas more approximately equal to test value, and the error of prediction was less. Withdata assimilation, model estimates more closely matched the measured dynamicfluctuations of soil moisture in response to rainfall events. The results showed that thepredicted value of Data assimilation was better than conventional simulation.5. The simulated soil water content in depth layers with data assimilation wascompared with ADL. With data assimilation, model estimates more closely matched themeasured dynamic fluctuations of soil moisture in response to rainfall events.But therewere many parameters in data assimilation, and it brings difficulty to the directlyapplication of data assimilation in short-term. The simulated soil water content in moredepth layers with ADL matched the measured dynamic fluctuations of soil moisture theoverall reform. The modeling and forecasting precision of ADL was higher than that ofother methods.