Based on the measured water quality data of Huangyuan County, Huzhu Tu Autonomous County, and Minhe Hui Tu Autonomous County in Hehuang Valley of Qinghai province in the normal and wet seasons, the effects of land use and land cover patterns on regional seasonal water quality were analyzed using remote sensing technology and mathematical statistics. The results showed that:① the concentrations of total nitrogen and total phosphorus in the surface water of Hehuang Valley were high. Water pollution areas (Class Ⅳ and Ⅴ) were mainly concentrated in the lower reaches of the river and the junction of tributaries. ② The explanation rate of land use to water quality in the normal season was higher than that in the wet season. The optimal scale was the 200 m buffer scale in the normal season, and farmland and towns were the main influencing factors. The optimal scale in the wet season was the 5 km buffer scale, and the main influencing factor was the forest. ③ In the normal season, the proportion of farmland was positively correlated with the concentration of total nitrogen and permanganate index but negatively correlated with the concentration of total phosphorus. The proportion of town area was positively correlated with the water quality index. The proportion of grassland area in the wet season was positively correlated with the permanganate index. The proportion of forestland area was negatively correlated with water quality index in both periods. Farmland, grassland, and town areas were the "source" landscape of pollutants, but farmland also played a role in intercepting pollutants to a certain extent. Forest land was the "sink" landscape of pollutants. ④ The pattern of forestland in the 200 m buffer zone in the normal season had a high explanatory rate for water quality, and the largest patch index (LPI) and patch density (PD) were the main factors. The study showed that it is an important measure to purify the surface water quality of Hehuang Valley by rationally planning the proportion of residential land and cultivated land and improving the coverage rate and aggregation degree of forestland around the riparian zone.
Keywords: Hehuang Valley; land cover patterns; land use structure; random forest model; water quality evaluation.