基于MODIS数据的东北地区气温反演及玉米冷害监测研究
发布时间：2018-11-10 07:41
【摘要】：低温冷害是我国农作物的主要气象灾害,玉米是低温冷害的主要作物之一。玉米作为东北三省主要的粮食作物,其产量严重影响着畜牧业的发展及人们的生活水平。全球气候变暖导致东北地区粮食的种植界限逐渐在向北移动,且随着种植面积和规模的逐渐扩大,倘若东北地区发生冷害,影响非常严重。应用遥感数据对于玉米冷害的监测目前并不多见。本文利用2005-2014年MODIS双星平台数据,选取MODIS高质量的LST数据建立关于气温的遥感估算模型,将MODIS传感器每日4次对地观测高质量LST数据进行融合,获取全天候气温遥感数据,经数据的时间插补获取连续的时间序列日数据。为得到缺失站点的数据需要对时间序列进行空间插补。最后参照气象行业标准中冷害的指标,基于空间插补后的数据对东北地区玉米低温冷害年份进行判别。本研究主要结论如下:(1)以气象台站观测日平均气温作为因变量,引入陆地表面温度(LST)、经度(LON)、纬度(LAT)、归一化植被指数(NDVI)、太阳天顶角(SAZ)、高程(ALT)以及日序数(N)作为模型的自变量,所选因子均达到0.01水平显著相关。选取四次观测的高质量LST数据作为数据源,利用多变量线性回归方法构建的气温遥感模型调整R2分别达到了 0.632、0.824、0.53及0.706,经2013-2014年的数据进行模型验证,估算模型大部分样本均落在(1:1)线附近。(2)考虑到数据的完整性,对双星平台每日4次过境数据进行融合,根据气温遥感模型的调整R2大小,以双星平台夜间数据优先,白天数据次之的规则进行融合,融合前后误差均满足正态分布;应用邻近时间气温的插补方法对缺失的气温进行时间序列的时间融合。融合插补后数据量增长1倍以上,插补后平均误差增加均不超过0.5℃。(3)采用生长季内≥10℃活动积温的距平指标(指标1)和5-9月平均气温之和的距平指标(指标2)对2005-2014年东北地区玉米低温冷害进行判别,两个判别指标对冷害年份及站点的判别大体上相同,但也略有差异。2005、2006、2009和2011年均发生冷害,2005年为冷害多发年份,冷害多集中在辽宁和吉林。指标2监测出的冷害年份多于指标1所监测的结果,在2006年和2011年指标2监测出的冷害站点数少于指标1。(4)用农业气象灾害数据和气象台站数据进行遥感估算冷害的指标验证,气象数据计算出的两个冷害指标对冷害年份和站点的判别一致性很高,但是在冷害等级划分上存在差异;用气象指标对遥感估算冷害进行验证,部分年份及站点表现出不一致现象;用农业气象灾害数据进行遥感估算冷害指标及台站冷害指标的验证,部分冷害年份和站点一致。
[Abstract]:Low temperature chilling injury is the main meteorological disaster of crops in China, and maize is one of the main crops of low temperature chilling injury. Corn is the main food crop in the three provinces of Northeast China, and its yield seriously affects the development of animal husbandry and people's living standard. Global warming causes the grain planting boundary to move northward gradually, and with the growing area and scale expanding gradually, if chilling damage occurs in Northeast China, the effect will be very serious. It is rare to use remote sensing data to monitor maize chilling damage. In this paper, the MODIS binary platform data from 2005 to 2014 are used to establish the remote sensing estimation model of air temperature with MODIS high quality LST data. The MODIS sensor fuses the high quality LST data of earth observation four times a day to obtain the all-weather temperature remote sensing data. Continuous time series daily data are obtained by time interpolation of data. In order to obtain missing site data, time series need to be spatially interpolated. Finally, referring to the index of chilling injury in meteorological industry standard, based on the data of space interpolation, the cold damage years of maize in Northeast China were judged. The main conclusions of this study are as follows: (1) with the daily mean temperature observed by meteorological stations as dependent variables, the land surface temperature (LST), longitude and latitude (LON), latitude (LAT), normalized vegetation index (NDVI),) and solar zenith angle (SAZ),) are introduced. The elevation (ALT) and the daily ordinal (N) are the independent variables of the model, and the selected factors are significantly related to each other at the level of 0. 01. Four high quality LST data were selected as data sources, and the adjusted R2 of temperature remote sensing model constructed by multivariate linear regression method reached 0.632n0.8240.53 and 0.706, respectively. The model was verified by the data from 2013-2014. Most of the samples of the estimation model fall near the (1:1) line. (2) considering the integrity of the data, the binary platform transiting data four times a day is fused, and R2 is adjusted according to the temperature remote sensing model. In the case of binary satellite platform, the night data is first used, then the daytime data is fused, and the errors before and after fusion are normal distribution. The time series of the missing temperature is fused by the interpolation method of the adjacent time temperature. After fusion interpolation, the amount of data has more than doubled. The mean error increased by less than 0.5 鈩，
本文编号：2321827
[Abstract]:Low temperature chilling injury is the main meteorological disaster of crops in China, and maize is one of the main crops of low temperature chilling injury. Corn is the main food crop in the three provinces of Northeast China, and its yield seriously affects the development of animal husbandry and people's living standard. Global warming causes the grain planting boundary to move northward gradually, and with the growing area and scale expanding gradually, if chilling damage occurs in Northeast China, the effect will be very serious. It is rare to use remote sensing data to monitor maize chilling damage. In this paper, the MODIS binary platform data from 2005 to 2014 are used to establish the remote sensing estimation model of air temperature with MODIS high quality LST data. The MODIS sensor fuses the high quality LST data of earth observation four times a day to obtain the all-weather temperature remote sensing data. Continuous time series daily data are obtained by time interpolation of data. In order to obtain missing site data, time series need to be spatially interpolated. Finally, referring to the index of chilling injury in meteorological industry standard, based on the data of space interpolation, the cold damage years of maize in Northeast China were judged. The main conclusions of this study are as follows: (1) with the daily mean temperature observed by meteorological stations as dependent variables, the land surface temperature (LST), longitude and latitude (LON), latitude (LAT), normalized vegetation index (NDVI),) and solar zenith angle (SAZ),) are introduced. The elevation (ALT) and the daily ordinal (N) are the independent variables of the model, and the selected factors are significantly related to each other at the level of 0. 01. Four high quality LST data were selected as data sources, and the adjusted R2 of temperature remote sensing model constructed by multivariate linear regression method reached 0.632n0.8240.53 and 0.706, respectively. The model was verified by the data from 2013-2014. Most of the samples of the estimation model fall near the (1:1) line. (2) considering the integrity of the data, the binary platform transiting data four times a day is fused, and R2 is adjusted according to the temperature remote sensing model. In the case of binary satellite platform, the night data is first used, then the daytime data is fused, and the errors before and after fusion are normal distribution. The time series of the missing temperature is fused by the interpolation method of the adjacent time temperature. After fusion interpolation, the amount of data has more than doubled. The mean error increased by less than 0.5 鈩，
本文编号：2321827
本文链接：https://www.wllwen.com/kejilunwen/nykj/2321827.html
