An Approach To Estimate Multiple Cropping Index From Remotely Sensed Data In Eastern Hilly Region Of Sichuan

Multiple Cropping Index (MCI), one important component of cropping system, is the average number of tillage in a field during a period of time. It is very important for researches on Carbon Sequestration at global or regional scales, Crop Yield Estimation, or Cultivated land Capacity Estimation and so on to know the exact MCI information of one region. Considering the advantages of objectivity, high efficiency, low cost and better spatial distribution attribute, the remote sensing technology is widely used to monitor the MCI at regional scale to make up for the disadvantages of the traditional method based on statistical calculation.Many methods for monitoring the MCI have been developed based on the remote sensing technology and different theories. However, the monitoring precisions of those existing methods are generally poor in some ragged areas where lots of cultivated lands are located (e.g., the Hilly Region in Eastern Sichuan), if the common remotely sensed data is used in those methods. One idea to improve the monitoring capacities of those methods is to introduce or develop some data fusion algorithms for generating the remotely sensed data with high spatial and temporal resolutions (i.e.,"both high") characteristics, which can improve the spatial resolutions of the common data source. Taking a case study for Yanting County in Mianyang for example, the study intends to explore the feasibility to monitoring MCI in some ragged regions based on "both high" time-series VI data by high spatial and temporal resolution data fusion algorithms.Through introducing a patch-based ISODATA classification method, the sliding window technology and the temporal-weight concept, this study firstly improves one data fusion algorithm■STDFM (Spatial and Temporal Data Fusion Model); and the test results show that the prediction precision of the improved algorithm—ESTDFM (Enhanced Spatial and Temporal Data Fusion Model) is better than the original one. This study then compares the different schemes (direct scheme and indirect scheme) based on another data fusion algorithm—STARFM (Spatial and Temporal Data Fusion Model) to generate time-series NDVI data with high spatial and temporal resolutions, and the test results present that it is almost no difference for the two schemes to predict mid-values, the direct scheme, however, is superior to the indirect one for predicting high or low values. The study finally selects the direct scheme of STARFM to generate time-series NDVI data with high spatial and temporal resolutions for Yanting County, and acquires the MCI information of2002and2011in the county in use of the second order difference algorithm, comparing advantages and disadvantages of different data fusion models. Validated results show that:the MCI extracted by the monitoring method is nearly equal to the statistical data at the scale of county (the statistical data vs. the remotely sensed extraction data:1.69vs.1.67); the overall accuracy is73.97%validating by the survey data in2013. The result has proved that it is feasible to monitoring MCI in some ragged regions based on "both high" time-series VI data by high spatial and temporal resolution data fusion algorithms. Additionally, monitoring results show that:the MCI fell by about0.3from2002to2011in Yanting County. About31.04%of the cultivated lands in the county have changed the cropping intensity. The main change is the conversion from double cropping system to single cropping system, which about29.93%; and the main change areas concentrate on the western and southern of the county.The innovation of this study is the creative researches on high spatial and temporal resolution data fusion algorithms, which contants that it puts forward an Enhanced Spatial and Temporal Data Fusion Model (ESTDFM) on the basis of the STDFM algorithm, which promotes the researches on the unmixing based data fusion algorithms; and it compares the different schemes based on STARFM to generate time-series NDVI data with high spatial and temporal resolutions, which supplies the scientific basis for selecting which one scheme of STARFM in application.