| With the constant development of national economy and continuous improvement of people's living standard, the vegetable industry has entered a new period of rapid development since our country decided to implement the "Vegetable Basket Project". The rapid advance of urbanization makes it the growing demand for vegetable. Therefore, obtaining land use distribution information of greenhouse vegetable area rapidly and accurately, particularly the distribution of protected vegetable land, and researching on the effect of soil quality under different land use will provide the scientific basis for the rational use of land resources, improvement of soil fertility and pollution-free agricultural production in greenhouse vegetable area. It is also important to achieve the balance between demand and supply of vegetable and ensure a healthy and orderly development and scientific and rational management of vegetable industry.This paper took Shouguang city of Shandong province as the study case. On the basis of various data collected, support vector machine (SVM) classification technique was used to deal with the spectrum and texture information of TM image to obtain accurate distribution of greenhouse vegetable land, open-air vegetable land and farmland, etc. Based on the RS classification and SPSS.v.16.0 software, this paper systematically analyzed the mean, standard deviation, coefficient of variation and correlation coefficients of soil nutrients and used LSD of one-way ANOVA to research soil nutrient change under different land use types. At last, measures and suggestions were put forward for soil fertility improvement and rational use of cultivated land in the study area.The main results were as follows:(1) The new bands were constructed based on TM image using MNF, K-T and NDVI. The new bands including former four bands of MNF, former three bands of K-T and NDVI band were used to classification. The results showed that the fusion of image had an obvious improvement on the separation in the training samples, compared with the original image.(2) In order to extracting texture information, the TM image was firstly analyzed by principle component transform, and secondly, the PCA1 and PCA2 image were analyzed with Gray Level Co-occurrence Matrix using eight texture features to obtain the new band with sixteen texture characteristics. Then discuss the texture windows. The result showed that the 5×5 window was better.(3) Using former four bands of MNF, NDVI band, former three bands of K-T, and sixteen texture data layers of the windows size of 5×5 as a data source, SVM was used to classify by selecting the RBF kernel function. The result showed that SVM made better use of multi-source data and had higher precision and adaptability. Compared with SVM only based on spectrum information, the result showed that texture information might improve classification accuracy.(4) The results showed that there was an obvious order in the distribution of soil nutrient content under different land use types: the distribution of organic matter was greenhouse vegetable land>open-air vegetable land>garden>farmland>saline-alkaline land; the distribution of total-N, available-N, K, P and Zn was greenhouse vegetable land>garden>open-air vegetable land>saline-alkaline land>farmland; the distribution of pH value and exchangeable-Mg was saline-alkaline land>farmland>open-air vegetable land>garden>greenhouse vegetable land; the distribution of exchangeable-Ca under different land use types was saline-alkaline land>farmland>open-air vegetable land>greenhouse vegetable land>garden; the distribution of available-B was garden>greenhouse vegetable land>saline-alkaline land>open-air vegetable land>farmland; the distribution of available-Fe was greenhouse vegetable land>garden>open-air vegetable land>farmland>saline-alkaline land; the distribution of available-Cu and available-Mn was garden>greenhouse vegetable land>open-air vegetable land>saline-alkaline land>farmland. The variance of nutrient content between open-air vegetable land, greenhouse vegetable land and garden was comparatively low, and the nutrient content between farmland and saline-alkaline land was similar. Except pH, the variance coefficients of the other 12 soil nutrients under different land use types were all comparatively high, of which the highest variance coefficient was available-P. Except that exchangeable-Mg and B had little correlation with other nutrients, the relevance among the others were almost or highly significant positive correlated. The pH value was significant or highly significantly negative correlation with the other soil nutrients, while the relevance of organic matter with other nutrients showed significant or highly significantly positive correlation.(5) Generally speaking, the condition of soil nutrients was good in Shouguang. The percentage of cultivated area needing to improve soil nutrients is 43.55% of the total cultivated area, mainly located in the north and the mid-east region. Overall, soil nutrients needing to improve were organic matter, available-P and available-B. The cultivated area lacking of organic matter, available-P and available-B was 3190.17 hectares which accounted for 3.70% of the total cultivated area. The cultivated area lacking of organic matter and available-P was 11405.98 hectares. The cultivated area lacking of organic matter and available-B was 2190.03 hectares. The cultivated area only lacking of organic matter, available-P and available-B were respectively 7230.72 hectares, 5498.08 hectares, and 8056.41 hectares. Seeing from the area ratio of various nutrients types, it needed to increase the use of phosphate and boron. Besides, it also should focus on the application of organic fertilizer, and gradually improved soil organic matter content.
|
PDF Full Text Request