Drought Remote Sensing Monitoring And Early Warning In Eastern Agricultural Areas Of Qinghai Province

Drought can cause serious losses with its slowly changing features. Drought monitoring and early warning is a fundamental in drought control. Because of the shortage of water resources in eastern agricultural area of Qinghai Province,drought has seriously affected the development of agriculture in this region. Developing a strong drought monitoring and early warning for the study area has a great deal of positive significance.This study takes meteorological data and MODIS data into account, and constructs drought warning model of the study area. Meanwhile, Drought in study area is monitored and early warned. The work also analyses suitabilities of drought model. The main conclusions of this article are as follows:(1) Comprehensive meteorological drought index(CI) and meteorological drought comprehensive monitoring index(MCI) of 12 meteorological stations in the study area are calculated based on rainfall and evaporation data in 1960~2005. Through the comparison and analysis of the monitoring results of MCI and month of CI drought index number and frequency of drought, some features of drought are able to be described. They are drought process, spatial distribution characteristics, seasonal variability and interannual variability, which then compare with the actual drought conditions in the region. The results indicate that MCI drought index is better than CI index for monitoring drought, for MCI index is developed from CI index.(2) Using the MODIS vegetation index and surface temperature data of the study area in 7~9 of 2002 ~ 2004, Ts- NDVI and Ts- EVI model are bulit respectively. According to the features of initial fitting space figure and eliminate different pixels, NDVI is selected from 0.2 to 1.0, and EVI selected from 0.15 to 0.8. The results illstrate that the spatial distributions of TEVI and TEVI are similar. Southeast drought is much more serious than other parts. Moderate drought is mainly distributed in the Huangshui River Valley, the Yellow River Valley and along the Datong area. From the aspect of time scale, severe drought oftenly happens in July, while there are a bit alleviation in August. In September degree of drought increases slightly. TVDI and TEVI present a tendency of positive correlation of monthly average ET0, and monthly average precipitation showed negative correlation trend, and the actual situation is in a good agreement. When precipitation has a trend of increase, alleviate the drought, drought index decreases. When the ET0 increased drought intensifies the tendency, drought index increased. In general, TEVI and precipitation and ET0 correlation coefficient is higher, and drought monitoring effect is more significant. The work also extracts TVDI and TEVI values from 12 sites in the study area, and then compares with results of SPI and M index in same periods. The results show that the consistency of TEVI was better than TVDI, but there were some sites that were more than M. Compared with the actual drought, TVDI and TEVI monitoring results than SPI. And M is closer to the actual drought TVDI and TEVI consistent with the actual drought rate were 40.0%, 46.7%, which is slightly better than the TVDI. On the other hand, accuracy of drought rate reflected by TEVI is not high. The reason may be use of remote sensing data of TVDI and TEVI, or the lack of considering meteorological factors.(3) The study constructs drought prediction model of MCI and TEVI to make drought monitoring and prediction, based on data from July to September in 2004. Combined with MCI and TEVI drought grade division standard, the levels of the drought early warning are determined. According to drought grade change rate, drought warning issued standards are developed. The results showed that drought gradually reduces from July to September. In August and September, the warning lights turn to be green light, blue light and yellow light. The proportion of blue light is largest, and the occupation of green light is a bit smaller than blue. Yellow light accounted for the smallest proportion. Yellow light regions are Huangshui River Valley, the Yellow River Valley, and a small part of yellow light scattered in the chase along Datong region. The temperature value is drought index monitoring results are basically in accord with the actual situation. Early warning index DWI and monthly average precipitation showed a negative correlation, and the monthly average ET0 was positively correlated with actrul trend. Compared with DWI value and M value of 12 meteorological stations in study area, grade consistent site probabilities are 33.3%, 25% and 16.7%. DWI values are greater than M, and the probabilities are of 58.3%, 75% and 75% respectively. Compared with the relative humidity index, results of DWI monitoring slightly heavier. DWI and M consistent with the actual probability of drought degree from July to September were 53.3% and 40.0%. Some sites, different with actual drought, basically have only one level higher than the actual drought. Using basic DWI in drought monitoring and early warning in the study area is accord with the actual distribution of drought. The monitoring precision is better than only using remote sensing data TEVI, for the precision increase from 46.7% to 53.3%. In summary, it is feasible to use DWI to make the drought monitoring and early warning of the monthly scale in study area.