Positioning Error Of Legacy Soil Sample And Uncertainty Analysis Of Positioning Error Within Digital Soil Mapping

Soil is an important foundation of agricultural production and terrestrial ecosystems. Land users, stakeholders of enterprises and government departments all needs quantified soil information as basis of land resource management. At the same time, with issues such as precision agriculture, ecological environment protection and sustainable development and global change becoming increasing serous, it is urgent that high precision, low cost, detailed and real-time soil resource information should be provided. Scientists had already collected plenty of soil data in the past time.Legacy soil datum have very high application value. Rising from the late1990s, Digital Soil Mapping which makes full use of mathematical methods and soil information system to predict spatial distribution of soil properties rapidly, provides solutions to obtain real-time and accurate soil information.A large amount of valuable soil foundation data were accumulated in the process of soil science study. For example, soil census data, lacked accurate positioning devices such as GPS when collecting from the field, however, the space coordinate information of samples is essential during mapping and the surface process simulation. In order to solve the problem of spatial location and positioning error of legacy soil samples, Anhui Province were choosed as study area in this research and208typical profiles were collected from whole province, GPS coordinate information and environment description information of those profiles are recorded. Now assuming that those profiles are without GPS coordinate information, then positioning is just implemented by environment description information such as village name, soil, parent material, topography and land use, and then calculate the offset distance between the predict and the true GPS coordinate.After positioning, we establish different radius buffer of GPS coordinate-centric, and analyze positioning probability falling in each buffer. Using RANDBETWEEN (bottom, top) to realize GPS random jump function in different radius buffer. Finally, the surface soil organic carbon was used as an example to digital soil mapping. The accuracy of digital soil mapping results was also discussed.The results are as follows:(1) Descriptive information positioning has two kinds of positioning methods. The first one use village name to position, Average positioning offset distance of the185soil samples is1516m, offset distance of sample samples within0-3000m is89.19%, and the average offset distance varies with the reference precision. the township average offset4431.55m is biggest, the administrative village average offset1104.38m takes second place, the natural village average offset615.17m is least. The second positioning method needs other information help to locate. Firstly, the soil samples are positioned with the village name and soil types. Secondly, they are positioned with the village name、 soil types and soil parent material. Thirdly, they are positioned with the village name、soil types、soil parent material and land use. Finally, village name, soil, parent material, land use and the altitude are all used to positioning. The average offset distances are1436m,1401m,1375m,1274m respectively and they decrease gradually, the medians decrease gradually, the standard deviations decrease gradually. Combined with the five positioning information can receive the best results, About90percent offset distance of soil samples offset is less than3000m. Considering from the two positioning results of positioning methods, the biggest impact factor effecting positioning is village name, and other informations are secondary.(2) Analysis of different regions in Anhui Province, the offset analysis results in accordance with three geographical areas dividing by the main drainage system and the altitude showed that:The average offset distances of three geographic areas average have little difference. Soil samples at a high altitude south of Anhui have a large offset distance.(3) The soil type map, parent material, land use map and residential layers superposition satisfy each typical soil sample of the surface patches, we establish different radius buffer of GPS coordinate-centric.They are1000m,2000m,3000m,4000m and5000m buffers. Statistical average probability of soil sample positioning falling in0-1000m,1000-2000m,2000-3000m,3000-4000m,4000-5000m were19.14%,23.29%,21.79%,19.69%,16.09%, respectively. We select a fixed point of each soil sample in different buffer for prediction mapping, and then prediction mapping results are compared with the GPS coordinate mapping results, we found that the offset positioning less than2000m has little effect on the mapping results.(4) Establising different radius(3000m,5000m8000m,10000m) circles of GPS coordinate-centric, soil samples jump randomly in the circles. Following with jumping analysis, ordinary Kriging is used to map prediction. The prediction mapping results show that offset distance increases, the prediction accuracy is worse.