Mining activities are the leading cause of deforestation, land-use changes, and pollution. Land use/cover mapping in Vietnam every five years is not useful to monitor land covers in mining areas, especially in the Central Highland region. It is necessary to equip managers with a better tool to monitor and map land cover using high-resolution images. Therefore, the authors proposed using the U-Net convolutional network for land-cover classification based on multispectral Unmanned aerial vehicle (UAV) image in a mining area of Daknong province, Vietnam. An area of 0.5kmx0.8km was used for training and testing seven U-Net models using seven optimizer function types. The final U-Net model can interpret six land cover types: (1) open-case mining lands, (2) old permanent croplands, (3) young permanent croplands, (4) grasslands, (5) bare soils, (6) water bodies. As a result, two models using Nadam and Adadelta optimizer function can be used to classify six land cover types with accuracy higher than 83%, especially in open-case mining lands and polluted streams flowed out from the mining areas. The trained U-Net models can potentially update new land cover types in other mining areas towards monitoring land cover changes in real-time in the future.

No abstract

Investigating information on land cover changes is an indispensable task in studies related to the variation of the environment. Land cover changes can be monitored using multi-temporal satellite images at different scales. The commonly used method is the post-classification change detection which can figure out the replacement of a land cover by the others. However, the magnitude and dimension of the changes are not been always exploited. This study employs the mixture of categorical and radiometric change methods to investigate the relations between land cover classes and the change magnitude, the change direction of land covers. Applying the Change Vector Analysis (CVA) method and unsupervised classification for two Landsat images acquired at the same day of years in 2000 and in 2017 in Duy Tien district, the experimental results show that a low magnitude of change occurs in the largest area of direction I and direction IV regarding the increase of Normalized Difference Vegetation Index (NDVI), but the opposite trend of (Bare soil Index) BI in the rice field. Alternately, the high magnitude of change is seen in the build-up class which occupies the smallest area with 1700 ha. The characterized changes produced by the CVA method provide a picture of change dynamics of land cover over the period of 2000-2017 in the study area.