关中平原干旱遥感监测指数对比和应用研究
Comparison and application of remote sensing monitoring indexes of drought in Guanzhong Plain

采用2000-2016年5月份MODIS数据，构建NIR-Red特征空间，对比分析基于该特征空间的垂直干旱指数（Perpendicular Drought Index，PDI）、改进型垂直干旱指数（Modified Perpendicular Drought Index，MPDI）、土壤湿度监测指数（Soil Moisture Monitoring Index，SMMI）及改进型土壤湿度监测指数（Modified Soil Moisture Monitoring Index，MSMMI）这四种干旱监测指数的有效性，并与实测土壤湿度进行相关性分析；最后采用精度最高的SMMI分析关中平原的旱情时空分布特征和规律。结果表明：（1）PDI、MPDI、SMMI及MSMMI均与10cm深土壤湿度存在负相关关系，可决系数R2分别为0.60、0.40、0.64、0.40，表明PDI、MPDI、SMMI及MSMMI均可作为旱情监测指标，且SMMI略优于其它三种监测指数；（2）关中平原东部、中部、西部部分地区旱情严重，西南部地区旱情较轻，且旱情呈年际波动显著的特征；（3）SMMI与月平均气温呈正相关关系区域占75.66%，与月降水量呈负相关关系区域占74.34%，其中通过90%显著性检验区域分别占总面积的27.36%、17.26%，说明降雨和温度不是导致旱情变化的主要影响因子。
By using MODIS data from May 2000 to May 2016, constructing the NIR-Red space, and collecting the Perpendicular Drought Index (PDI), Modified Perpendicular Drought Index (MPDI), Soil Moisture Monitoring Index (SMMI) and Modified Soil Moisture Monitoring Index (MSMMI),we studied the validity of these four indexes and the correlation of them with the measured soil moisture data. Then, the spatial and temporal distribution characteristics of drought in Guanzhong plain was assessed based on SMMI that had the highest precision.The results showed that: (1) There was a negative correlation between PDI, MPDI, SMMI and MSMMI with measured soil moisture in the top 10 cm layer with R2, coefficient of correcation of 0.60,0.40,0.64, and 0.40, respectively. It was suggested that all four indexes could be used as drought indicators while SMMI was slightly better than the others. (2) The drought conditions in the eastern, central, and western regions of Guanzhong plain were more severe while soil water content was relatively higher in the southwestern area. There was a significant inter-annual fluctuation in drought of Guanzhong plain. (3)There was a positive correlation between SMMI and average monthly temperature in 75.66% of the total area but a negative correction of SMMI with monthly precipitation in 74.34% of the area. The significance test showed that 27.36% and 17.26% of that are at P<0.1 level, respectively, indicating that precipitation and temperature were not the major factors causing the changes in drought.