Spatial and temporal variation characteristics of cultivated land in the upper Yellow River from 2002 to 2018 based on time series MODIS

The upper reaches of the Yellow River contribute 56.77% of the water resources in the whole basin, while agricultural water in the upper reaches accounts for more than 40% of the whole basin. Using MODIS data with medium-resolution remote sensing and crop phenology data, the cultivated land area in the upper reaches of the Yellow River from 2002 to 2018 was extracted, which provided basic data for studying the impact of cultivated land change in the upper reaches on water resource consumption in the basin. Using Harmonic Analysis of Time Series (HANTS) to smooth the cut and spliced MOD13Q1 data, the NDVI time series curve was easier to identify. Combined with the decision tree classification method and crop growth period and other phenological information, the decision tree rules were compiled, and then the smooth MOD13Q1 data were classified by using the classification rules. The distribution and change of cultivated land in the study area from 2002 to 2018 are obtained. Then, the confusion matrix verification of the extraction results was carried out by using the real ground sample data obtained from field investigation and the county-level statistical data in the study area. The extraction accuracy of cultivated land was above 75%, and R² reached 0.85. The main results showed that the cultivated land in the study area were increased from 2002 to 2018, with a total increase of 88.21×10⁴ hm². The most rapidly increased was Ningmeng Irrigation District, the total area of cultivated land in Ningxia increased by 64%, reaching 76.61×10⁴ hm² in 2018; and Inner Mongolia increased reaching 44.74×10⁴ hm², accounting for 44% of the total area; and Gansu section increased reaching 18.89×10⁴ hm²; Qinghai section showed an obvious trend of decrease in cultivated land, a total of 5.36×10⁴ hm². On the whole, the increase rate of cultivated land area is 5.18×10⁴ hm²·a⁻¹, in which the Qinghai section decreases at a rate of 0.2×10⁴ hm²·a⁻¹, the Gansu section increases at a rate of 1.05×10⁴ hm²·a⁻¹, the Ningxia section increases at a rate of 1.87×10⁴ hm²·a⁻¹, and the Inner Mongolia section increases at a rate of 2.46×10⁴ hm²·a⁻¹. Mann-Kendall analysis method was used to analyze the trend of NDVI change in the study area, and it was found that NDVI showed an increasing trend. In addition, compared with the change of precipitation, it also showed an increasing trend. Then using stepwise regression analysis to analyze the selected main indicators, the residents’ disposable income was the main factor affecting the change of cultivated land. Finally, using the collected data for qualitative analysis, it was concluded that water resources policy and engineering facilities construction are the factors affecting the change of cultivated land. The main conclusions of this paper were as follows. The cultivated land in the basin from Longyangxia to Hekou Town in the upper reaches of the Yellow River shows an increasing trend. The medium-resolution MODIS data can be used to extract the cultivated land in the upper reaches of the Yellow River. Water resources policy and local socio-economic environment change were the main driving factors for the change of cultivated land area in the study area.