Abstract-To verify the effect of artificial flow on the improvement of local DO(dissolved oxygen) imbalance distribution, Lake Jing in Guangxi University was taken as the experimental area, five groups of experiments were designed, pump with different flow rate and different nozzle were used to push flow in experimental area, RBF Neural Network which is widely used in geographic information prediction was used to fitting the spatial distribution of dissolved oxygen and the dispersion coefficient was taken as the equilibrium index of DO horizontal distribution to analysis the effect of flow. The result shows that100 m 3 /min, 130 m 3 /min, 180 m 3 /min pump with circular nozzle and 130L/min pump with duckbill nozzle respectively need 1.5 hour, 1.0 hour, 0.5 hour, 1.5 hour to decrease the DO dispersion coefficient from 0.3 to 0.05, witch prove push flow has obvious effect on improving the DO balance and the effect is related with flow rate and nozzle, a larger flow pump and circular nozzle is preferred.

No abstract

Abstract. In order to explore a new way of optimization for soil sampling layout, in this paper, spatial distribution of heavy metal concentrations which is based on RBF neural network fitting was studied by genetic algorithm, corresponding network structure and algorithm flow were constructed, in addition the network and algorithm parameters settings were analyzed and a sampling experiment using this method was conducted in an abandoned mine located in a county of Guangxi. The results of optimizing the layout point prove that the number of the sampling points can be reduced by about a half under the premise of meeting the fitting accuracy, this will greatly reduce the cost of sampling and analysis and the redundancy of the data, and is expected to promote the relevant application in soil composition analysis and other related fields.