林地旱灾风险预警研究

Study on forest land drought risk early warning

【目的】为减少因林地旱灾带来的森林资源损失,运用粒子群优化算法(PSO-BP)建立林地旱灾风险预警模型对湖南省林地旱灾风险进行实时预警,为提前采取预防措施、合理配置应急资源提供技术支撑。【方法】以中分辨率成像光谱仪(MODIS)数据提取到的地表温度(LST)和归一化植被指数(NDVI)、热带降雨测量任务(TRMM)数据提取到的降水量(JYL)和连续无雨日(CDWR)、高程(GC)、坡度(PD)、坡向(PX)、植被类型空间分布(ZB)等8个因子为自变量,以LST与NDVI反演得到的温度植被干旱指数(TVDI)为因变量,基于PSO-BP算法建立林地旱灾风险预警模型,预测未来温度植被干旱指数(TVDI)来表征林地干旱程度,并以当日的实测数据进行验证。【结果】以湖南省2009年1月1日至2019年10月2日的遥感数据提取的风险因子构建PSO-BP林地旱灾风险预警模型,预测2019年10月2—4日的温度植被干旱指数,模型的相关系数R值均达到了0.9以上,预测结果准确度高,能够很好地预测未来的林地旱灾风险情况。从实测结果可知,2019年10月4日湖南发生了旱灾,发生的地点集中在永州市、郴州市以及株洲市的南部区域,湘西小部分区域也有些许旱情,其余区域都无旱;10月4日是旱情最为严重的时期,但在10月2日已经有了干旱的苗头,且由于连续无雨日过长,干旱情况将持续加重;到了10月3日,干旱情况已经非常严峻,湘南等地区出现大面积干旱;直到10月4日,干旱情况再次加重,该实测结果与预测结果吻合。【结论】构建的PSO-BP林地旱灾风险预警模型能够科学有效地对湖南省林地旱灾风险进行预警,为林地旱灾的应急处置提供科学支撑。

【Objective】In order to reduce the loss of forest resources caused by forest land drought,particle swarm optimization(PSOBP)algorithm was used to establish forest land drought risk early warning model,which could provide technical support for taking preventive measures in advance and reasonably allocating emergency resources.【Method】Surface temperature(LST)and normalized vegetation index(NDVI)extracted from MODIS data,rainfall(JYL)and continuous rainless days(CDWR)extracted from TRMM data,elevation(GC),slope(PD),slope aspect(PX),spatial distribution of vegetation types(ZB)Taking temperature vegetation drought index(TVDI)obtained from LST and NDVI inversion as the dependent variable,the forest drought risk warning model was established based on particle swarm optimization(PSO-BP)algorithm.The TVDI was predicted to characterize the forest drought degree,and the measured data of the day were used for verification.【Result】Based on the risk factors extracted from the daily remote sensing data of Hunan province from 2009 to October 2,2019,the PSO-BP forestland drought risk early warning model is constructed to predict the temperature vegetation drought index from October 2 to 4,2019.The correlation coefficient R value of the model is above 0.9,and the prediction result is accurate,which can well predict the future forestland drought risk.According to the measured results,the drought occurred in Hunan on October 4,2019,which concentrated in the south of Yongzhou,Chenzhou and Zhuzhou.There was a little drought in a small part of western Hunan,and there was no drought in other areas.October 4 was the most serious period of drought,but there were signs of drought on October 2,and the drought will continue to increase due to the long continuous rain free days On October 3,the drought situation was very severe,and a large area of drought appeared in southern Hunan and other areas;until October 4,the drought situation aggravated again,and the measured results were consistent with the predicted results.【Conclusion】The constructed PSO-BP forest land drought risk early warning model can scientifically and effectively early warn the forest land drought risk in Hunan province,and provide scientific support for the emergency disposal of forest land drought.