Monitoring drought condition based on HJ-1A/1B data in natural rubber plantation

Abstract: Crop drought monitoring is one of the main tasks of agricultural monitoring, and the monitoring and quantitative assessment of natural rubber drought has vital practical significance to disaster control, resource safety and the ecology and environment protection of rubber plantations. Yangjiang Farm (109o38′-109o49′E, 19o16′-19o25′N), one of the largest farms in the Hainan State Farms, is located in the north-central Hainan Island. The land area and natural rubber plant area of the farm are 14 367 and 5 190 hm2, respectively. It was chosen for this study. In order to improve the accuracy and real-time performance of natural rubber drought monitoring, a comprehensive model was established based on the relationship among the standardized vegetation supply water index (VSWIS), compound index (CI) and the synchronous soil moisture data measured in the study area using HJ-1A/1B-CCD data, HJ-1B-IRS data and meteorological data. The VSWIS was suitable for monitoring drought at the high density vegetation area, and the CI was suitable for real-time meteorological drought monitoring. A total of 100 sampling points in 4 sample areas was designed and the distance was 30 m between 2 adjacent points. In addition, another 25 points were randomly selected for model validation. Soil moisture of these points was measured from May 11 to May 20 in 2013. The drought was quantitatively characterized by drought index (DI). The DI value was numerically equal to the soil moisture value. By the multiple linear regression, an equation was built to estimate soil moisture based on VSWIS and CI. The model had determination coefficient of 0.67 (P<0.05). Thus, the DI value was also obtained. Then, the drought was quantitatively classified by DI values: wet with DI of 0.45-1, normal with DI of 0.30-0.45, light drought with DI of 0.20-0.30, moderate drought with DI of 0.05-0.2, and severe drought with DI of 0-0.05. The smaller DI value suggested more severe drought condition. The model validation showed the relative root mean square error (RMSEr), the Nash-sutcliffe efficiency (NSE) and the index of agreement were 11.02%, 0.40 and 0.85, respectively, indicating that model was reliable in estimating DI values. The model was used for estimating drought condition of the natural rubber drought in Yiangjiang Farm from May to July 2010. The results revealed that the natural rubber drought in Yiangjiang Farm were greatly different. Overall, the drought in the west and north were more serious than that in the east and south of the farm. It was more sever in early July. Until the late July, the drought was still serious. The natural rubber's growth was greatly affected by the drought condition. The normalized difference vegetation index (NDVI) and the net primary production (NPP) of natural rubber showed a lag effect with the length of 10 days. The drought affected the NPP of natural rubber about 30 days, longer than its influence on the NDVI of natural rubber. The rubber potential productivity and rubber yield were also affected by the drought, showing a strong synchronization, and it could last for about 30 days. The impact of drought on the yield was longer than that on the potential productivity. The research provides useful information for drought monitoring and its impacts assessment of natural rubber.