Font Size: a A A

The Soil Spectral Feedback Surface Is Applied To The Study Of Soil Mapping Methods In Flat Areas

Posted on:2016-12-12Degree:DoctorType:Dissertation
Country:ChinaCandidate:S X GuoFull Text:PDF
GTID:1360330482457953Subject:Photogrammetry and Remote Sensing
Abstract/Summary:PDF Full Text Request
Soil is one of the fundamental nature resources supporting daily human life and production of the entire society. Soil types and soil properties are the major topics in soil science field. Furthermore, the spatial variety of soil type and soil properties are the fundamental inputs to many models, such as ecological modeling, hydrological modeling, precision farming, forest monitoring etc. The digital soil mapping is the branch of soil science and combined with GIS technology to produce soil maps which represent the spatial pattern of soil types and soil properties. The soil landscape model as the most important model in digital soil mapping indicate that soil are highly related to the local environments. Jenny (1941) had first concluded that the formation process of soil is extremely related to five environmental factors which are Parent material, Time, Topography, Climate and Organisms. The hypothesis of soil landscape model is the same soil will be born in the similar environments, and the different environments will product the different soils. Under this concept, it is possible to predict the spatial variation of soil by indicate the spatial patterns from environmental factors.Among of these five environmental factors, topography play a crucial role to determine the water movement, which is a key process of soil settlement and soil formation. Many studies in digital soil mapping field were using topography factor as the most important information to indicate the spatial variation of soil. The previous studies have confirmed that the pattern of topography can indicate the spatial pattern of soil in high relief area, such as mountain area. However, over low relief areas, such as plains and most farmland, commonly used environmental covariates, such as elevation, slope gradient, vegetation, often fail to effectively indicate the spatial variations of soils. More important those low relief areas are farmland which is a crucial field of interaction of human and nature. To predict the soil pattern in those areas will have more meaning for both economic and society.With the rapid research and development of remote sensing technology, in previous researches, to address this issue, soil feedback dynamic pattern was introduced as a new environmental covariate to effectively indicate the spatial variations of soils over low relief area. This pattern expresses the soil reflectance changes during the drying process and can be used to indicate the information of soil types and soil properties at that location. Early study also shows the variation of patterns showing at different locations is indicative of the variation of soil types at those locations.However, these four scientific questions need to addressed before applying soil feedback dynamic pattern for soil mapping into a large flat area. First, how the local environmental parameters influence soil feedback dynamic pattern? And which parameter is most related to the change of soil surface reflectance during soil drying process? Second, the data gaps caused by cloud cover may lead to many incomplete soil feedback patterns. In addition, when data gaps occur on different days for two patterns with different locations, it is hard to directly compare these two incomplete patterns. How to fill those data gaps to rebuild the complete feedback pattern for each location is an essential question to make patterns comparable for differentiating the soils from different locations. Third, the soil feedback dynamic patterns representing the soil drying process are highly influenced by local evaporation conditions. How to make the soil feedback dynamic patterns obtained in different evaporation conditions comparable remains a challenge. Fourth, when mapping soil types by using soil feedback pattern as the only environmental factor, the multiple relationship between soil types and soil feedback patterns are limited to predict soil spatial pattern with traditional soil-landscape model. How to maximum guessing the probability of soil types base on soil feedback dynamic pattern is a question.In this dissertation, I propose the theory, idea and experiment for each scientific question. The major contribution of this study is as follows.1) A new relationship has been derived mathematically to illustrate the relationship between soil reflectance in water in the sensitive band (MODIS band 7) and the square root of cumulated reference evapotranspiration (CET00.5). A case study in flat farmland in northeastern Illinois shows a good relationship between MODIS band 7 and the square root of cumulated reference evapotranspiration (CET00.5) in most of the bare soil farm land (with average R2=0.55, p<0.001; and average NRMSE 10.40%).2) This dissertation presents a method for solving the problem of missing data caused by cloud cover in the construction of dynamic soil feedback patterns. The main argument is that the cumulative reference evapotranspiration can be used as auxiliary data to assist gap filling in the feedback pattern.3) This dissertation also presents an improved soil dynamic feedback pattern space to solve the comparable problems. The improved pattern space is an evaporation related space, which concerns the local evaporation conditions as a component. In this new pattern space, soils with the similar properties tend to present the same pattern regardless of local evaporation conditions. This makes patterns in different evaporation conditions to be able to compare with each other.4) For the multiple indication problem when apply soil feedback dynamic pattern in to soil type prediction in a flat area, this dissertation introduced a methodology based on Bayesian theory to get the probability of soil existence for a particular location.The theory, idea and methodology presented in this thesis started by soil science. The problems this thesis addressed are curtail issues in digital soil mapping field. But more important, these idea and theory open a new perspective for analyzing spectrum pattern in periodic series in remote sensing field. In this point of view, the soil is only a case study of remote sensing periodic series analysis.
Keywords/Search Tags:Digital soil mapping, Soil feedback dynamic pattern, Soil reflectance, Soil evaporation, Soil spectrum pattern space, Remote sensing periodic series analysis
PDF Full Text Request
Related items