Abstract: Net Primary Productivity (NPP) is the net carbon sequestration of terrestrial vegetation, which is of great importance for the study of global carbon estimation. The rocky mountainous areas of North China is a typical arid and semi-arid climatic zone, and the accurate estimation of the NPP and its variation characteristics in this area is of great significance for the construction of China's ecological forestry engineering. Based on the climate data from 1980 to 2020 from Henan Xiaolangdi Forest Ecosystem National Observation and Research Station, this paper used 5 climate productivity models, including the Miami model, Thornthwaite Memorial model, Chikugo model, Zhu Zhihui model and Zhou Guangsheng model, to estimate and analyze the changing trend of NPP. The random forest algorithm was used to explore the influencing factors of NPP, and the regional MODIS NPP data was used as the standard to evaluate the suitable climate productivity model for estimating NPP in this area. The results showed that: (1) the annual average temperature, annual average maximum temperature, annual average minimum temperature, and annual precipitation in the rocky mountainous areas of North China showed an upward trend, and the rates were 0.05℃y⁻¹,0.04℃y⁻¹,0.05℃y⁻¹ and 1.58mmy⁻¹, respectively. The annual average solar radiation and annual average relative humidity showed a downward trend, with rates of 0.46MJm⁻²y⁻¹ and 0.17 percent pointsy⁻¹, respectively. (2) The NPP calculated by 5 models showed an upward trend, but the NPP values were different, ranging from 739.35gCm⁻²y⁻¹ to 958.48gCm⁻²y⁻¹, with an average of 862.19gCm⁻²y⁻¹. Among them, the Miami model had the maximum estimated value (958.48gCm⁻²y⁻¹), and the Zhou Guangsheng model had the minimum estimated value (739.35gCm⁻²y⁻¹). (3) The random forest algorithm showed that precipitation was the predominant factor affecting NPP in the region. The applicability analysis indicated that the estimated value of the Zhou Guangsheng model was the closest to MODIS NPP, with the relative error, RMSE, and MAE of 1.45%, 451.05gCm⁻²y⁻¹, and 446.03gCm⁻²y⁻¹, respectively, and the correlation coefficient was the highest (0.49). This study showed that the Zhou Guangsheng model was more suitable for NPP estimation in this area, and priority should be given to the Zhou Guangsheng model when using the climate productivity model to estimate NPP in the rocky mountainous areas of North China.

Key words: Net primary productivity, Climate productivity model, Rocky mountainous areas of North China, MODIS NPP, Climatic factor