基于Conv1D-LSTM混合模型的长时间序列日最高温预测研究

针对传统方法难以处理高维度数据捕捉气温数据中的非线性模式和复杂动态特征的问题,本文提出一种基于卷积神经网络(Conv1D)与长短期记忆网络(LSTM)相结合的混合模型,用于长时间序列高温预测研究。数据集包含北京市2014年至2023年间的气象数据,包括天气、日最低温、日最高温、风向等特征。通过特征工程处理,将天气和风向特征编码,并对温度特征归一化。构建的Conv1D-LSTM混合模型创新性地融合Conv1D以捕获时间序列中的局部特征,融合LSTM以学习长期依赖关系。与传统模型相比,该混合模型的均方根误差(RMSE)和平均绝对误差(MAE)分别降低约17.3%和20.5%,同时R2分数提高约1.06%,表明该模型具有更高的预测精度和泛化能力。

苹果果面日最高温与主要气象因子的关系(英文)

一天中 ,苹果果实表面最高温度与气温、日照、相对湿度和风速有着密切的关系。通过生长季自动监测果实表面温度的变化 ,并与设置在监测树旁边气象记录仪数据对比 ,揭示出树冠西南部果实表面温度与气温、日照、相对湿度和风速呈高度相关。结果还表明 :日照和气温是导致果实表面高温 (>45℃ )的两个主要因子。在大多数情况下 ,这两个因子相互作用决定果实高温是否发生 ,但气温的作用更为重要。一般在晴天 ,树冠西南部果实表面高温通常出现在 1 3 :3 0～ 1 5 :5 0。果实达到临界日烧高温 (>45℃ )的综合气象条件是 (1 0 :0 0～ 1 6 :0 0平均值 ) :日照 >5 80 W/m2 ;气温 >3 1 .8℃ ;风速 <1 .0 m/s;相对湿度 <3 0 %。气温、日照、风速和相对湿度与果实表面温度都呈高度相关 ,其多元回归方程为 :果实表面温度 (℃ ) =1 9.7+0 .842×气温 (℃ ) +0 .0 0 989×日照 (W/m2 ) - 2 .0 8×风速 (m/s) - 0 .1 48×相对湿度 (% ) ,r2 =0 .6

Interpretation of 850 hPa Temperature in Temperature Forecast of Jiamusi Area

Based on the initial field temperature data of ECMWF 850 hPa from Jan- uary 2012 to December 2015, linear interpolation method of ECMWF was employed to calculate the 850 hPa temperature values at 8：00 and 20：00 of 7 stations （Jiamusi, Tangyuan, Huachuan, Huanan, Fujin, Tongjiang, Fuyuan）. Combined with the observed daily minimum and maximum air temperatures at the same time of the 7 stations, the correlations of the 850 hPa temperature values at 8：00 and 20：00 with the daily maximum or minimum air temperature of the ground meteorological obser- vation stations were established, and the ground observation data in accordance with the relevant analysis and correlation test principle of the prediction equation for factor were primarily selected. Regression method was used to establish forecast e- quation dividing into counties, month by month. The results showed that the 850 hPa temperature values at 8：00 and 20：00 were significantly correlated with the daily maximum or minimum air temperature, and the established temperature fore- cast equation was of certain guiding significance for the forecast of daily minimum and maximum temperature, which could help to improve the forecast accuracy.

“火霜”及其防御——答山阴县读者王金天问

山阴县马营庄乡云水庄新村王金天来信询问“火霜”问题,我们约请在雁北、朔州工作多年的老科技干部霍举、刘德宝等帮助答复,现摘登如下。谁还有好的经验,好的办法,欢迎来信告诉我们。

Change in Extreme Climate Events over China Based on CMIP5

The changes in a selection of extreme climate indices(maximum of daily maximum temperature(TXx),minimum of daily minimum temperature(TNn),annual total precipitation when the daily precipitation exceeds the 95th percentile of wet-day precipitation(very wet days,R95p),and the maximum number of consecutive days with less than 1 mm of precipitation(consecutive dry days,CDD))were projected using multi-model results from phase 5 of the Coupled Model Intercomparison Project in the early,middle,and latter parts of the 21st century under different Representative Concentration Pathway(RCP)emissions scenarios.The results suggest that TXx and TNn will increase in the future and,moreover,the increases of TNn under all RCPs are larger than those of TXx.R95p is projected to increase and CDD to decrease significantly.The changes in TXx,TNn,R95p,and CDD in eight sub-regions of China are different in the three periods of the 21st century,and the ranges of change for the four indices under the higher emissions scenario are projected to be larger than those under the lower emissions scenario.The multi-model simulations show remarkable consistency in their projection of the extreme temperature indices,but poor consistency with respect to the extreme precipitation indices.More substantial inconsistency is found in those regions where high and low temperatures are likely to happen for TXx and TNn,respectively.For extreme precipitation events(R95p),greater uncertainty appears in most of the southern regions,while for drought events(CDD)it appears in the basins of Xinjiang.The uncertainty in the future changes of the extreme climate indices increases with the increasing severity of the emissions scenario.

Observed trends in diurnal temperature range over Nigeria

The long-term trend of diurnal temperature range(DTR)over Nigeria was examined using daily station-based datasets for the period 1971–2013.The results show that the regionally averaged DTR has decreased significantly(-0.34°C per decade)over the Nigerian Sahel(north of 10°N),but there has been a slight increasing trend(0.01°C per decade)over the Nigerian Guinea Coast.The annual decreasing trend of DTR in the Nigerian Sahel is mainly attributable to the significant increasing trend in daily minimum temperature(Tmin,0.51°C per decade),which far outstrips the rate of increase in the daily maximum(Tmax,0.17°C per decade).In contrast,the comparable trends in Tmin(0.19°C per decade)and Tmax(0.20°C per decade)may explain the non-significant trend of the DTR averaged over the Guinea Coast region.It is observed that the DTR has decreased more in boreal summer(June–July–August)than in boreal winter(December–January–February)for the regions.Furthermore,it is found that the significant DTR declining trend over the Nigerian Sahel is closely associated with an increasing trend of annual and summer precipitation in the region,but the increasing DTR trend in the Nigerian Guinea Coast region can be attributed to the decreasing trend of cloud cover over the region.

Temporal and spatial variations in extreme temperatures in the Qilian Mountains-Hexi Corridor over the period 1960-2013

Based on daily maximum and minimum temperatures at 18 meteorological stations in the Qilian Mountains and Hexi Corridor between 1960 and 2013,temporal and spatial variations in extreme temperatures were analysed using linear trends,tenpoint moving averages and the Mann-Kendall test.The results are as follows:The trends in the majority of the extreme temperature indices were statistically significant,and the changes in the extreme temperatures were more obvious than the changes in the extreme values.The trends were different for each season,and the changes in rates and intensities in summer and autumn were larger than those in spring and winter.Unlike the cold indices,the magnitudes and trends of the changes in the warm indices were larger and more significant in the Hexi Corridor than in the Qilian Mountains.Abrupt changes were detected in the majority of the extreme temperature indices,and the extreme cold indices usually occurred earlier than the changes in the extreme warm indices.The abrupt changes in the extreme temperatures in winter were the earliest among the four seasons,indicating that these temperature changes were the most sensitive to global climate change.The timing of the abrupt changes in certain indices was consistent throughout the study area,but the changes in the cold indices in the Hexi Corridor occurred approximately four years before those in the Qilian Mountains.Similarly,the changes in the warm indices in the western Hexi Corridor preceded those of the other regions.