It is apparent that the number of environmental disasters and the human and property loss resulting from them have been on the rise in the recent decade due to the fact that urban centers get more and more densely developed while human populations and property are highly concentrated in areas physically vulnerable to natural and technological hazards. In particular, human vulnerability to floods is one of the major natural hazards leading to damage and loss of life when it occurs. According to Federal Emergency Management Agency [1] , approximately 17,000 communities are situated in the areas which are at risk of flooding while about 9.6 million households and property worth $390 billion face a flood hazard that is characterized by a high probability or that has a 1% chance of occurring per year; however, only 20% to 30% of building at risk is said to be covered by flood insurance.

Since the Industrial revolution, human populations all across the world have been increasing and will continue to do so in the future. Especially, the increase of the population is generally happening in our urban environment, meaning that more and more people and property concentrate on cities. Urbanization is a major global trend simply because people want to get their jobs, raise their children, and live their lives in cities. As urbanization always transforms the natural environment into a man-made one, attracting people and materials, thus contributing to change not only in land use and land cover but also landscape and hydrological systems in that area. This change, in turn, negatively influences the natural environment, which can lead to an increase in disasters and crises as well as vulnerability. For example, changing a certain floodplain into a built-up area used to develop a subdivision neighborhood makes that particular area and adjacent ones less permeable to precipitation and more vulnerable to flooding. As a result, a heavy rain, for example, can flood those areas, claiming loss of lives and property.

There has been some research done to investigate relationships between land use/land cover change and environmental hazards [2] [3] [4] [5] [6] . However, little research has been conducted to test the direct effects of land cover change on environmental disasters such as floods, hurricanes, hazardous material releases. Therefore, this research aims to analyze the effect of land cover change on environmental hazards by using GIS and remote sensing technologies. More specifically, its main goal is to test whether land cover change is related to floods in the state of Texas from 1993 to 2012 by using such technologies. The reason why the time frame (i.e., 1993 to 2012) was chosen is the remote sensing data obtained from the National Land Cover Database (NLCD) were available only for the time period when this research was conducted.

This research aims to analyze the effect of land cover change in the state of Texas on environmental disasters (i.e., flooding). To do so, property damage caused by floods was used as a proxy to environmental disasters. More specifically, the goal of the research is to test whether land cover change is related to a flood disaster in Texas from 1993 to 2012.

To achieve the objective, two research hypotheses were developed. The main point of the hypotheses is that the increase in the spatial extent of built environment and the decrease in that of the natural environment contribute to the increase of loss of life and property resulting from flooding events. These hypotheses and their rationales are as follows:

- Hypothesis 1: The change in forest area within a county from 1992 to 2012 is correlated with its total property damage resulting from flooding.

- Rationale for Hypothesis 1: Forest area can contain much precipitation through an infiltration process instead of letting the runoff flow into the urban built environment, rivers or lakes during rainfall because the area is pervious. However, if forest area is converted into impervious built-up area, then flooding is more likely to happen than ever. Therefore, the decrease in spatial extent of forest is expected to contribute to flooding, which can claim loss of life and property damage, especially in the urban areas.

- Hypothesis 2: The change in built-up area within a county from 1992 to 2012 is correlated with its total property damage resulting from flooding.

- Rationale for Hypothesis 2: Impervious areas that comprise built-up sites (e.g., buildings, roads, streets, and parking lots) make runoff stay longer mostly in the roads and streets, which causes the water flow level to increase in urban areas, resulting in flooding. Therefore, the increase in spatial extent of built-up area is expected to contribute to flooding, which is highly likely to claim greater loss of life and property damage.

The study area is the state of Texas. The state is situated in the south central part of the country and bordered by Louisiana on the east, Oklahoma on the north, New Mexico on the west, and the Gulf of Mexico and Mexico on the south. This second largest state in terms of its population (about 25.7 million in 2011) [7] and spatial extent (268,820 square miles) has three metropolitan areas such as Houston, Dallas-Fort Worth, and San Antonio. The year 2010 United Census statistics show that the state’s racial and ethnicity consist of White (70.4%; Non-Hispanic Whites: 45.3%), Black or African American (11.8%), Native American (0.7%), Asian (3.8%), and Pacific Islander (0.1%) with other race (10.5%) [8] . Much of Texas has been plagued with various natural (e.g., floods, hurricane, tornadoes, wildfire, and extreme weather) and technological disasters (e.g., hazardous material releases). Since the state has Tornado Alley along its northern region, the most tornadoes occur in this state every year. Hurricanes are also one of the salient natural events in this state. A hurricane which struck Galveston in 1900 in the southern part of Texas claimed over 8000 deaths, which ranked the first in the most death tolls in the U.S. history [9] . Also, the state has a high risk from hazardous material accidents because the state has a large number of oil refining and manufacturing facilities along the coast line [10] . If a hurricane of category 3, 4 or 5 hit oil-gas facilities located in the east coast, the very deadliest and costliest extreme disaster would happen in the country. In addition, Texas has seen a lot of flooding events. According to National Weather Service, it is floods and flash floods that claim an average of 15 causalities every year in Texas. Especially, flash floods with little or no lead time for emergency warnings rapidly trigger runoff, which can lead to more damage and loss of life than regular floods that usually happen within or near floodplains.

As mentioned earlier, the economic damages triggered by these natural and technological disasters in the state are likely to increase as a consequence of population growth and its concentration, especially in fifteen counties along the Gulf coast region. In reality, according to the Census Bureau data, 10 out of 13 counties in the Gulf coast region have experienced rapid population growth rates from 2000 to 2008, ranging from 10% to 50%. Additionally, it is more important to note that this increased physical vulnerability is likely to be related to land cover change over the last decade. This issue will be dealt with in details in the section of research results.

One of the main datasets used for this research is the National Land Cover Database (NLCD), which can publicly be downloaded from http://www.mrlc.gov/index.php. NLCD is Lansat TM-based land cover maps of the United States. Landsat TM sensor produces seven spectral bands of imagery, each of which has a spatial resolution of 30 meters, and one far-infrared band of imagery, which has a spatial resolution of 120 meters. These remote sensing images can be downloaded at no charge from the USGS Earth Explorer website (http://earthexplorer.usgs.gov).

NLCD products currently available include land cover maps for 1992, 2001, and 2006, and land cover change between 1992 and 2001, all of which have metadata including spatial reference and land cover classification systems. These datasets were developed by the Multi-Resolution Land Characteristics Consortium (MRLC) consisting of a host of federal government agencies. MRLC is a partnership to conduct a national land cover mapping project. In 1992 participants of MRLC included the Environmental Protection Agency, the United States Geologic Survey, National Oceanic and Atmospheric Administration, the United States Forest Service, the National Aeronautics and Space Administration, and the Bureau of Land Management while the list of 2001 and 2006 was added to the list of the 1992 participants, which include LANDFIRE, the Natural Resources Conservation Service, the National Park Service, the US Fish and Wildlife Service, and the Office of Surface Mining.

In 1992, MRLC partners started to develop the first land-cover data, the NLCD 1992 for the continuous 48 United States by using 30-meter Landsat TM images which were obtained in 1991, 1992 and 1993 [11] . This land cover map consists of 21 different classes of land cover. In 2006, MRLC began conducting a third-time land-cover mapping project. This land cover classes follows the land cover classification method used in 2001, which features 16 different classes. All the different land classes are shown in Table 1 below.

Even though the 1992 and 2006 NLCD vary with classification methods, this research used only the classes found in both NLCDs (e.g., open water, forest, and wetlands). In addition, some classes were merged so that each NLCD could have