34
R. Madhavan et al.
Part of it will infiltrate into the soil, part of it will run off, and part of it will evaporate.
In a high slope spot covered by concrete, the water will run off. In a spot covered
by a forest on a sandy soil at a low slope, water is going to infiltrate easily. So if a
forest is replaced by agriculture, or even worse, by concrete, less water will infiltrate
and, consequently, more water will run off. If water flows on the surface, it carries
soluble substances, grains, small pieces of rock, big pieces of rock, cars, houses, and
people. However, it is not enough to think in terms of a single spot, but we need to
consider a region instead, and each one of these dimensions (vegetation, land use,
soil, geology, slope, etc.) as superimposed layers.
GIS is the conventional platform to cross as many layers of information as needed
so we can detect and model the spatial relationships between parameters. Once we
formulate the (map) algebra, we can then generate digital maps depicting runoff and
flow accumulation which accumulation, which, in turn, indicate landslide and flood
risk. As noted earlier, spatial criteria change through time. We can use the same
geology and soil maps as long as we live because they will not change for hundreds
or thousands of years. But the same cannot be said about land use and cover. Forests
turn into pasture. Natural fields turn into rice fields. Human activities are changing
land cover ever faster. So fast that when a land use and cover map is done it is already
outdated!
Remote sensing enables us to be quick in assessing this process. Sensors carried
by satellites give us quantitative measures of the way a material interacts with the
electromagnetic energy so we can differentiate targets based on what we call spectral
response pattern. If a satellite carries a radar sensor, or yields stereo pairs, we can
also obtain topographic data. This type of information comes as images, so we can
mathematically manipulate them and find ways to automatically detect land use and
cover changes. However, when we need very detailed spatial information, sometimes
urgently, satellites present several constraints: low spatial resolution of free/low-cost
images, high costs related to high spatial resolution images, a fixed revisit frequency,
and requirements to schedule an observation, the time lapse between an observation
and the image supply, or the simple presence of clouds.
What to do when we need to be very sure whether an extreme meteorological
event will be hazardous or not for specific infrastructures or populations? Highresolution land use and cover maps, as well as 3D models, are required for hydrological, hydraulic, and meteorological models at a very local scale. In the imminence/occurrence of an extreme event, rapid assessment of high-risk/affected areas
is needed. It would be way too cumbersome and labor-intensive to regularly map land
cover and use and build 3D models with a high resolution (in the order of 2–5 m)
for a region as large as 60,000 km
2 . This is the actual area of the Littoral Basin in
Rio Grande do Sul where our efforts are focused (see Fig. 3.1). Besides, when it
comes to map land use and cover, which is driven by human activities, we are pretty
much always out of time. GIS models based on satellite data are great. They give us
a broad view of vast regions so we can detect high-risk areas as well as where flood
and landslide risks are increasing over time.
Our current work, thus, focuses on enhancing the regional models using precision
and high-resolution images acquired using UAVs (Fig. 3.2), especially for those
R. Madhavan et al.
Part of it will infiltrate into the soil, part of it will run off, and part of it will evaporate.
In a high slope spot covered by concrete, the water will run off. In a spot covered
by a forest on a sandy soil at a low slope, water is going to infiltrate easily. So if a
forest is replaced by agriculture, or even worse, by concrete, less water will infiltrate
and, consequently, more water will run off. If water flows on the surface, it carries
soluble substances, grains, small pieces of rock, big pieces of rock, cars, houses, and
people. However, it is not enough to think in terms of a single spot, but we need to
consider a region instead, and each one of these dimensions (vegetation, land use,
soil, geology, slope, etc.) as superimposed layers.
GIS is the conventional platform to cross as many layers of information as needed
so we can detect and model the spatial relationships between parameters. Once we
formulate the (map) algebra, we can then generate digital maps depicting runoff and
flow accumulation which accumulation, which, in turn, indicate landslide and flood
risk. As noted earlier, spatial criteria change through time. We can use the same
geology and soil maps as long as we live because they will not change for hundreds
or thousands of years. But the same cannot be said about land use and cover. Forests
turn into pasture. Natural fields turn into rice fields. Human activities are changing
land cover ever faster. So fast that when a land use and cover map is done it is already
outdated!
Remote sensing enables us to be quick in assessing this process. Sensors carried
by satellites give us quantitative measures of the way a material interacts with the
electromagnetic energy so we can differentiate targets based on what we call spectral
response pattern. If a satellite carries a radar sensor, or yields stereo pairs, we can
also obtain topographic data. This type of information comes as images, so we can
mathematically manipulate them and find ways to automatically detect land use and
cover changes. However, when we need very detailed spatial information, sometimes
urgently, satellites present several constraints: low spatial resolution of free/low-cost
images, high costs related to high spatial resolution images, a fixed revisit frequency,
and requirements to schedule an observation, the time lapse between an observation
and the image supply, or the simple presence of clouds.
What to do when we need to be very sure whether an extreme meteorological
event will be hazardous or not for specific infrastructures or populations? Highresolution land use and cover maps, as well as 3D models, are required for hydrological, hydraulic, and meteorological models at a very local scale. In the imminence/occurrence of an extreme event, rapid assessment of high-risk/affected areas
is needed. It would be way too cumbersome and labor-intensive to regularly map land
cover and use and build 3D models with a high resolution (in the order of 2–5 m)
for a region as large as 60,000 km
2 . This is the actual area of the Littoral Basin in
Rio Grande do Sul where our efforts are focused (see Fig. 3.1). Besides, when it
comes to map land use and cover, which is driven by human activities, we are pretty
much always out of time. GIS models based on satellite data are great. They give us
a broad view of vast regions so we can detect high-risk areas as well as where flood
and landslide risks are increasing over time.
Our current work, thus, focuses on enhancing the regional models using precision
and high-resolution images acquired using UAVs (Fig. 3.2), especially for those