Integration of remote sensing and meteorological data for monitoring agricultural drought.

摘要

Affecting more people than other natural hazards, drought may lead to enormous decrease in crop production and also in the amount of poultry and livestock, and thus endangering food security and economy. Developing an appropriate drought indicator and a timely and accurate drought monitoring system has been a motivation for scientists in the last two decades. Vegetation conditions valued via remotely sensed indices have been used as indicators for agricultural drought since the 1980s. However, the anomalies in vegetation performance do not always signify droughts. Wild fire, extreme temperature, flood, pesticides or lack of fertilizers can all cause the vegetation stress. One of the major goals for this dissertation is to evaluate and investigate vegetation drought stress and other vegetation stresses using remote sensing techniques. The other major goal is to estimate the root-zone soil moisture levels beneath various crop canopies using satellite data and ground observations. Since soil moisture is the primary indicator for agricultural drought, accurate and reliable soil moisture estimates have important implications for drought monitoring.;Recent technological advances in remote sensing have shown that vegetation vigor, land surface temperature (LST), vegetation water level and soil moisture (SM) can be measured by a variety of remote sensing techniques, each with its own strengths and weaknesses. This research is designed to combine the strengths of Moderate Resolution Imaging Spectroradiometer (MODIS) based visible/near-infrared (VIS/NIR), shortwave infrared (SWIR) and thermal infrared remote sensing approaches for detection of vegetation drought stress, and also to integrate VIS/NIR and microwave data from Aavanced Microwave Scanning Radiometer (AMSR-E) of the Earth Observing System (EOS) for soil moisture estimation. A vegetation drought stress estimation algorithm at moderate resolution was developed based on the existing "trapezoid" relation model by using MODIS-based LST as well as Normalized Difference Water Index (NDWI) and Normalized Difference Vegetation Index (NDVI).;A new index, the Combined Condition Index (CCI) was proposed here for monitoring vegetation drought stress from space by interpreting the relationships between LST and Normalized Difference Drought Index (NDDI). Drought Condition maps from the U. S. Drought Monitor (USDM) and other reliable agencies are used to validate the spatial patterns of drought. Also, the feasibility of constructing a library of weighing factors for vegetation overall conditions, water, and temperature for each spatial and temporal unit across the U. S. will be discussed in the dissertation. The CCI will be compared with currently used indices, for example, Vegetation Health Index (VHI), for pros and cons. Also, the departure from precipitation, Palmer Drought Severity Index (PDSI), and crop yield/progress will be used to validate this index. This new drought indicator is expected to show higher sensitivity to drought occurrences than the existing ones.;Combining the proposed methods in detecting vegetation conditions and estimating soil moisture, we can obtain time-series profiles of vegetation conditions and soil moisture of various crops at different geo-spatial situations, and thus be able to monitor agricultural drought across the whole nation.;Data, information, knowledge, and wisdom are the four basic steps humans use to perceive objects (Ackoff, 1989). Agricultural droughts being considered as objects can also be perceived in these four forms -- drought data, information, knowledge, and wisdom. Hence it is necessary to extract drought information out of related data (e.g., remotely sensed data) and discover knowledge from the extracted information. Lastly, this dissertation is to explore advantages of geospatial Web services in providing on-demand agricultural drought analysis and equipping experts, decision-makers and farmers alike with information, knowledge and even wisdom needed in the process of agricultural drought monitoring, assessment and management. Various Web services are established to support drought analysis and decision-making for the general public, which also illustrates the potential of Web services in automating geospatial knowledge discovery and dissemination within the Big Data era.