Abstract: Based on statistical data on grain acreage, yields and agricultural disasters from 1961 to 2020, the variability characteristics of grain yields and disasters in China and five northern provinces were compared and analyzed. Disaster yield assessment models for Hebei, Shandong and Henan provinces were used to estimate the loss of grain production due to disasters and grain yields by inputting data on disasters from 2008 to 2020. The sensitivity and stability of the disaster-yield assessment model were examined. Based on the statistical modeling method, the model of grain crop disaster-yield evaluation in Shanxi and Shaanxi was constructed, and the universality of the model construction method was evaluated. The results showed that: (1) the grain planting area and total output of the five northern provinces accounted for 28% and 25% of the national total from 1961 to 2020, respectively. In the five northern provinces, the planting area of summer harvest grain and autumn harvest grain decreased significantly at rates of 3.39ha·a⁻¹ and 106.3ha·y⁻¹(P<0.01) respectively, while the total output increased significantly at rates of 137.3×10⁴t·y⁻¹ and 119.9×10⁴t·y⁻¹(P<0.01), respectively. From 2008 to 2020, the grain planting area and grain yield in the five northern provinces increased significantly at the rates of 209.42ha·y⁻¹ and 258.06×104t·y⁻¹(P<0.01), respectively. (2) From 1961 to 2020, the areas of covered, affected and destroyed disaster in the five northern provinces accounted for 28%, 28% and 23% of the national average, respectively, while the disaster situations in the five northern provinces and the whole country showed a significant trend of first increasing and then decreasing. After reaching historically high values in 2008, 2000 and 2000, the covered disaster, affected disaster and destroyed disaster area had declined year on year. The corresponding disaster situations in the five northern provinces showed a downward turning point in 1990, 1989 and 2004, respectively. Drought and flooding are the main causes of crop disasters in China, with 76 percent of the total area affected by drought and flooding. The disaster in the five northern provinces was mainly caused by drought. The areas affected by drought accounted for 66%, 61% and 58% of the disaster statistics, respectively. From 2008 to 2020, the area affected by drought in Shandong was the largest. The area of drought disaster in Hebei and Shanxi was relatively high. Hebei province has the highest area of flooding and hail. (3) When the data series is extended to 2020, the simulated value of grain yield is significantly correlated with the actual value(R²=0.95, P < 0.01), the simulation accuracy of the model was high. In the past 60 years, the grain loss rates of Hebei, Shandong, Henan, Shanxi and Shaanxi provinces were 8.99%, 18.02%, 9.79%, 12.84% and 20.04%, respectively. In the last 12 years, the grain loss rates of Hebei, Henan, Shanxi and Shaanxi provinces had recorded by 4.4%, 17.4%, 9.65%, 8.14% and 17.9%, respectively, influenced by the reduction of disaster zones and advantage in agricultural science and technology. They all went down. With the verification and construction of the model, the modeling statistical method performs well in evaluating the loss of grain yield due to meteorological disasters, had a promising performance in predicting grain yield, and is feasible for commercial applications. As the five northern provinces account for a high proportion of the country's grain output, it is important to prevent the risk of regional hydrometeorological disasters to ensure the country's food security in the new period.

Key words: Grain yield, Agrometeorological disasters, Regional disaster situation, Loss assessment due to disaster, Disaster-yield assessment model