The Research Of Remote Sensing Drought Prediction Model Based On EOS MODIS Data

Drought is a very important phenomenon that affects not only agricultural production but also society. Among all natural disasters, drought occurs the most frequently, with the longest duration, the largest area, and the greatest losses in agricultural production. In addition to its direct influence on grain yields, it also has a potential long-term effect on the environment, desertification, and other disasters. Especially combining with human activities, it is a great menace to environment. Effective drought early warning system is an integral part of efforts worldwide to improve drought preparedness and mitigation. Timely and reliable data and information must be the cornerstone of effective drought policies and plans. In China, the density of meteorological and hydrological stations is insufficient to provide adequate coverage for drought prediction and monitoring. Meteorological and hydrological data could not often be shared among agencies of government. This restricts early assessment of drought and retards data utilization in drought preparedness, mitigation, and response. Remote Sensing techniques are the most appropriate for detecting the status of soil moisture, evapotranspiration, crop growth, land-cover type and drought. The MODIS (MODerate Resolution Imaging Spectrometer) instrument is flying on the EOS TERRA-1 and AQUA-1, with a daily global coverage. MODIS is a passive, imaging spectroradiometer carrying 490 detectors, arranged in 36 spectral bands that cover the visible and infrared spectrum (from 0.415 to 14.235 mm). It is a high signal-to-noise instrument designed to satisfy a diverse set of oceanographic, terrestrial, and atmospheric science observational needs. It will make global moderate-resolution narrow-band radiance observations over 36 spectral regions. The spectral stability is expected to be better than 2 nm and the instantaneous field of view varies from 250m, 500m, and 1000m. Geo-location is better than the same types of sensors before. Much information could be retrieved from MODIS data, such as cloud, Land-Surface Temperature (LST), Vegetation Index (VI), Land-Cover Type, snow and ice cover, etc. These kinds of information have potential capabilities in drought early warning and monitoring.This Dissertation covers: (1) The background of EOS MODIS is systematically introduced, including the EOS project, the TERRA and AQUA satellites, the characteristics of MODIS sensor, the technical parameters and band distribution of MODIS data, the standard products and potential applications of MODIS data; (2) The algorithms of drought-related parameters from MODIS data are researched, including automatic geometry correction, temperature calculation, MODIS-NDVI vi AVHRR-NDVI researching, cloud detection, and three cloud parameters calculating, i.e., which are Continuous Cloud-free Days (CCFD), Cloud-free Days Ratio (CFDR), and Continuous Cloud Days (CCD); (3) The concept, classification, and characters of drought are discussed, and the drought distribution and affection in China are introduced; (4) The theories and methods of drought monitoring based on in-situ and remote sensing data around world are summarized; (5) A new drought prediction and monitoring model based on cloud method and traditional methods is proposed, and the full automatic prediction techniques based on this model and MODIS data is explored; (6) This model is verified with the in-situ data. The research shows that the model is very useful in drought early warning and monitoring, especially with the ground truth data.According to the researches in this dissertation, some conclusions could be derived as follow:(1) Drought is a very complicated phenomenon, with its natural attributes and societal ones. It is necessary to identify and make sure your aim and stage in drought research. Remote sensing means is mainly used in drought early warning.(2) With its many bands and advantages, MODIS data could be used to retrieve many drought-related parameters, such as, clouds, snow cover, land surface temperature, water body and vegetation information, which makes it the best potential data source in drought monitoring for large-scale so far.(3) The relationship between MODIS-NDVI and AVHRR-NDVI of the same region at different time, based on Histogram and Feather Space methods are analyzed. The results show that they shapes are the same, but MODIS NDVIs have more sensitivity to vegetation, and the range of their values are more width than NOAAAVHRR ones. They have no strong correlations. It is not a good idea to use AVHRR NDVI directly as MODIS history NDVI in drought monitoring.(4) In this dissertation, a new Drought Prediction and Monitoring Model based on cloud parameters and other traditional parameters is first proposed and verified. It is validated as a good and robust model in large-scale, short period drought prediction and monitoring with very useful application and research values.