由于我们对砷的估算模型在时间上是动态的，为了解释高砷风险概率发生变化这

Since our arsenic estimation model is dynamic in time, in order to explain the change in the probability of high arsenic risk, the researchers also focused on the driving effect of the main control predictor variables that change over time, and produced The marginal effect diagram of the dynamic predictor variables of the model is shown (Figure 5). Mainly including human activities (hydraulic gradient, interannual groundwater variation, groundwater accumulation variation) (Figure 5-eg), climate (AET, precipitation, temperature) (Figure 5-ik) and the relationship between high arsenic risk probability and human activities, climate The marginal effect of the three dynamic variables on the probability of high arsenic risk (Figure 5-h, 5-l). As can be seen from Figure 5, as human activities and climate change, high As probability will produce a certain degree of response. Changes in groundwater levels in the alluvial plains of the lower reaches of the Yellow River are mainly affected by human mining activities. The impact of hydraulic gradient (Figure 5-e), interannual groundwater variation (Figure 5-f), and groundwater accumulation variation (Figure 5-g) on ​​the probability of high arsenic is very obvious. The change in arsenic probability has a greater impact, with the probability reaching 13%. After the hydraulic gradient is greater than 0.07, as the hydraulic gradient increases, the probability of high arsenic occurrence decreases significantly. The interannual changes in water levels reflect the changes in groundwater levels between adjacent years. Compared with the previous year, as the groundwater levels rise, the interannual changes in water levels gradually increase from negative values, and the probability of high arsenic gradually decreases. The cumulative change in water level refers to the total decrease in groundwater level in the past 60 years. It can be seen that when the cumulative change in groundwater level reaches 0m, the probability of high arsenic reaches a peak. As the cumulative decrease in groundwater level increases, the probability of high arsenic increases. The probability shows a trend of decreasing first and then increasing. When the cumulative variation in water level reaches 5m, the probability of high arsenic is reduced to the minimum. In the multi-factor marginal distribution (Figure 5-h), the PDP probability ranges from 0.35 to 0.60. When the hydraulic gradient is less than 0.02, the annual drop in water level is greater than 5m, and when the cumulative change is less than 5m, the probability of high arsenic reaches a maximum of 0.60. These dynamic variables have different effects on the risk probability of arsenic in groundwater. As the hydraulic gradient decreases, the interannual decline in groundwater levels increases. When the accumulated changes in water levels over the years remain stable, it provides favorable conditions for the release of arsenic in the aquifer. . Evapotranspiration (Figure 5-i), precipitation (Figure 5-j), and temperature (Figure 5-k) all have varying degrees of effects on the changes in the probability of high arsenic. Among them, precipitation and temperature have the greatest impact on the changes in the probability of high arsenic. The extreme difference reaches 8%. As evapotranspiration increases, the overall probability of high arsenic shows an increasing trend. After the evapotranspiration exceeds 440mm, the probability of high arsenic decreases to a certain extent. When the precipitation increases to 610mm, the probability of high arsenic reaches the maximum. Subsequently, as precipitation increased, the probability of high arsenic showed an overall downward trend. When the temperature is less than 15.35°C, there is a negative correlation between the probability of high arsenic and temperature. As the temperature increases, the relationship between the probability of high arsenic and temperature changes to a positive correlation. It can be seen from the marginal effect of the three climate factors (Figure 5-l) that the PDP probability range is between 0.30-0.65, the evapotranspiration is 360-440mm, the temperature is less than 15℃, and the precipitation is around 600mm, high arsenic The probability reaches a maximum of 0.65. It can be seen that in the alluvial plain area of ​​the lower reaches of the Yellow River, as evapotranspiration increases, the temperature decreases and the precipitation changes within a certain range can provide favorable conditions for the release of As from the aquifer.

Due to the dynamic nature of our arsenic estimation model over time, in order to explain the phenomenon of changes in the probability of high arsenic risk, researchers also focused on the driving effect of the main predictive variables that will change over time, and created a marginal effect diagram of the model's dynamic predictive variables (Figure 5). This mainly includes human activities (hydraulic gradient, interannual variation of groundwater, cumulative variation amplitude of groundwater) (Figure 5-e-g), the relationship between climate (AET, precipitation, temperature) (Figure 5-i-k) and the probability of high arsenic risk, as well as the marginal effects of human activities and climate on the probability of high arsenic risk under the joint action of three dynamic variables (Figure 5-h, 5-l). From Figure 5, it can be seen that with human activities and climate changes, high As probabilities will have a certain degree of response. The changes in groundwater level in the alluvial plain area of the lower Yellow River are mainly influenced by human mining activities. The influence characteristics of hydraulic gradient (Figure 5-e), interannual variation of groundwater (Figure 5-f), and cumulative variation amplitude of groundwater (Figure 5-g) on the probability of high arsenic are very obvious. Among them, hydraulic gradient and interannual variation of groundwater have a significant impact on the probability of high arsenic, with a probability range of 13%. After the hydraulic gradient is greater than 0.07, the probability of high arsenic occurrence decreases significantly as the hydraulic gradient increases. The interannual variation of water level reflects the variation of groundwater level between adjacent years. Compared with the previous year, as the groundwater level rises, the interannual variation of water level gradually increases from negative value, and the probability of high arsenic gradually decreases. The cumulative variation of water level refers to the total extent of groundwater level decline in the past 60 years. It can be seen that as the cumulative variation of groundwater level reaches 0m, the probability of high arsenic reaches its peak. As the cumulative decrease of groundwater level increases, the probability of high arsenic shows a trend of first decreasing and then increasing. When the cumulative variation of water level reaches 5m, the probability of high arsenic decreases to the lowest. In the multi factor marginal distribution (Figure 5-h), the probability range of PDP is between 0.35 and 0.60. When the hydraulic gradient is less than 0.02, the interannual decrease of water level is greater than 5m, and the cumulative variation is less than 5m, the probability of high arsenic reaches the maximum of 0.60. These dynamic variables have different impacts on the risk probability of arsenic in groundwater. As the hydraulic gradient decreases, the interannual decrease in groundwater level increases. When the cumulative variation of groundwater level remains stable for many years, favorable conditions are provided for the release of arsenic in aquifers. Evapotranspiration (Figure 5-i), precipitation (Figure 5-j), and temperature (Figure 5-k) all have varying degrees of influence on the probability of high arsenic, with precipitation and temperature having the greatest impact on the probability of high arsenic, with a probability range of up to 8%. With the increase of evapotranspiration, the overall probability of high arsenic shows an increasing trend. After the evapotranspiration exceeds 440mm, the probability of high arsenic content decreases to a certain extent. When the precipitation increases to 610mm, the probability of high arsenic reaches its maximum. Subsequently, with the increase of precipitation, the overall probability of high arsenic shows a decreasing trend. When the temperature is less than 15.35 ℃, the probability of high arsenic shows a negative correlation with temperature, and as the temperature increases, the probability of high arsenic changes to a positive correlation with temperature. From the marginal effects of the joint action of the three climatic factors (Figure 5-l), it can be seen that

Because our arsenic estimation model is dynamic in time, in order to explain the phenomenon that the risk probability of high arsenic changes, the researchers also focus on the driving effect of the main control predictive variables that change with time, and make the marginal effect diagram of the dynamic predictive variables of the model (Figure 5). It mainly includes the relationship between human activities (hydraulic gradient, annual variation of groundwater and amplitude variation of groundwater accumulation) (Figure 5-e-g), climate (AET, precipitation, temperature) (Figure 5-i-k) and high arsenic risk probability, and the marginal effect on high arsenic risk probability under the joint action of human activities and climate (Figures 5-h, 5-l). As can be seen from Figure 5, with the change of human activities and climate, high AS probability will produce a certain degree of response. The variation of groundwater level in alluvial plain of the lower Yellow River is mainly influenced by human exploitation activities. Hydraulic gradient (Figure 5-e), interannual variation of groundwater (Figure 5-f) and cumulative variation of groundwater (Figure 5-g) have obvious influence characteristics on the high arsenic probability, among which hydraulic gradient and interannual variation of groundwater have great influence on the high arsenic probability, and the probability range reaches 13%. After the hydraulic gradient is greater than 0.07, with the increase of hydraulic gradient, the probability of high arsenic occurrence decreases greatly. The interannual change of water level reflects the change of groundwater level between adjacent years. Compared with the previous year, with the rise of groundwater level, the interannual change of water level gradually increases from negative value, and the probability of high arsenic gradually decreases. Cumulative amplitude of water level refers to the total amplitude of groundwater level decline since the last 60 years. It can be seen that the probability of high arsenic reaches its peak when the cumulative amplitude of groundwater level is 0m, and the probability of high arsenic first decreases and then increases with the increase of cumulative amplitude of groundwater level decline. When the cumulative amplitude of water level reaches 5m, the probability of high arsenic is minimized. In the multifactor marginal distribution (Figure 5-h), the probability of PDP ranges from 0.35 to 0.60. When the hydraulic gradient is less than 0.02, the annual drop of water level is more than 5m, and when the cumulative amplitude is less than 5m, the probability of high arsenic reaches the maximum of 0.60. These dynamic variables have different effects on the risk probability of arsenic in groundwater. With the decrease of hydraulic gradient, the annual decline of groundwater level increases, and when the cumulative fluctuation of groundwater level remains stable for many years, it provides favorable conditions for arsenic release in aquifers. Evapotranspiration (Figure 5-i), precipitation (Figure 5-j) and temperature (Figure 5-k) all have different effects on the change of high arsenic probability, among which precipitation and temperature have the greatest influence on the change of high arsenic probability, and the probability range reaches 8%. With the increase of evapotranspiration, the probability of high arsenic shows an overall increasing trend. When the evapotranspiration is more than 440mm, the probability of high arsenic decreases to some extent. When the precipitation increases to 610mm, the probability of high arsenic reaches the maximum. Subsequently, with the increase of precipitation, the probability of high arsenic showed a downward trend as a whole. When the temperature is less than 15.35℃, there is a negative correlation between high arsenic probability and temperature, and with the increase of temperature, there is a positive correlation between high arsenic probability and temperature. It can be seen from the marginal effect of the combined action of the three climatic factors (Figure 5-l) that the probability range of PDP is 0.30-0.65, when the evapotranspiration is 360-440mm, the temperature is less than 15℃ and the precipitation is around 600mm, the probability of high arsenic reaches the maximum of 0.65. It can be seen that in the alluvial plain area of the lower Yellow River, with the increase of evapotranspiration, the decrease of temperature and the change of precipitation in a certain range can provide favorable conditions for the release of As from the aquifer.