On
the Utility of the International Space Disaster Charter: An Analysis of Remote
Sensing Data Access Policies Using Wakabayashi Ward, Sendai as a Study Area
John
C Hayes and Gilberto Cruz 2012
University
of California Los Angeles
Introduction
Signed in
2000, the Charter On Cooperation To Achieve The Coordinated Use Of Space
Facilities In The Event Of Natural Or Technological Disasters (hereafter
referred to as the space disaster charter) provides a major channel for
international collaboration between researchers around the world to assess the
extent and effects of catastrophic events (Stryker and Jones 2009). Several
activations of the charter have occurred since the treaty’s inception in 2000,
but public and academic access to products have often been limited as the
agreement lacks any provisions that either promote or require open access to
data used in disaster relief products (International Charter Space & Major
Disasters 2012). In the wake of the Tohoku earthquake and tsunami of March
2011, the charter was activated and several disaster response products were
released by various participating agencies (International Charter Space &
Major Disasters 2012), but no raw data sets were released to the public and
only one vector data set was made publicly available by a participating agency.
Open access
has become a major factor influencing the current direction and trajectory of
research involving remote sensing (Wulder et al 20120). Changes in access
policies over the decade spanning 2000 to 2010 culminated in data from all
Landsat missions being made available on the internet for free or at the “Cost
of Fulfilling User Requests.” (National Aeronautics and Space Agency 2012)
Wulder et al have noted the way in which this “opening of the archive” has dramatically
increased the number of people using remotely sensed data in their research and
spurned a revolution in the field’s literature that has led to the development
of several new and innovative applications for the usage of satellite imagery.
(2012)
Noting the
existence of both closed and open access data policies and recognizing the
constraints of both, the following study has three primary foci: First, we will
show the differences in disaster response products that were developed in
reaction to the activation of the charter after the Tohoku earthquake and
tsunami of March 2011. Second, we will analyze using freely accessible data the
effect of the disaster upon the agriculture of the same exact region explored
in the first section. An analysis will follow in which we will explore the
constraints of closed data distribution in relation to the current study and
future research, examine what can be gleaned from the freely accessible data
provided by Landsat 7’s ETM+ sensor, and explain how the closed access nature
of the space disaster charter limits innovation in using remote sensing to
respond to various crises. We argue that the space disaster charter should open
data access to academic researchers in order to better serve their own
self-stated purpose.
Study Area
This research
focuses on the Wakabayshi ward of Sendai, Miyagi prefecture on the island of
Honshu in Japan. Sendai, with a population of over 1 million individuals is the
largest city in the Tohoku region of Japan (Central Intelligence Agency 2012).
Satellite imagery in the visual bands show the area to be home to vast
amounts of agricultural activity that we presume is mostly rice, as Sendai and
the larger Tohoku region are noted for their production of this staple grain
(United States Department of Agriculture 2012).
On 11 March
2011 a magnitude 9.0 earthquake struck off the coast of Japan and generated a
massive tsunami that inundated the coastline of the Tohoku region. Together,
the earthquake and tsunami killed approximately 230,000 individuals and caused
damage to infrastructure and property estimated to be around $10 billion
(National Oceanic and Atmospheric Administration 2011). In the immediate
aftermath of this event, the Japanese Aerospace Exploration Agency (JAXA)
requested activation of the space disaster charter and emergency response
products began to be developed by disaster response and earth monitoring
agencies around the world (International Charter Space & Major Disasters
2011).
Methods
Comparison
of Space Disaster Charter Maps and GIS Products
Emergency
response products generated from data captured by several different satellites
that were released under the space disaster charter were gathered from agencies
including the USGS, SERTIT, DLR, JAXA, and UNOSAT in their highest resolution
that was publicly available. A specific study area was determined based on a
JAXA product that depicted an unusually general inundation extent. It was
determined that the region in this product was Wakabayashi and products that
included this region released by other agencies exhibiting a clear inundation
extent were then georeferenced and digitized into a GIS using ESRI ArcMap 10.1.
The only exception to this was the UNOSAT product for which an official
shapefile format file was released. This resulted in the creation of five
separate extent maps that were then arranged in chronological order.
Change
Detection of Wakabayashi Agriculture Using Remotely Sensed Data
Landsat 7 ETM+
imagery of the region was acquired from the USGS Glovis online imagery
repository. Imagery of the study area used in this analysis from before the
disaster was captured on 01 September 2010, while imagery from after the
disaster was captured 05 August 2012. The acquired visible, infrared, and
panchromatic bands captured on both dates were corrected for scanline errors
present in all ETM+ data acquired since 2003 and then stacked into two separate
images using ENVI 5.0. Bands depicting red, green and blue in the visible
spectrum were sharpened using the panchromatic band to make the image clearer.
To display vegetation, NDVI indices were generated for imagery captured on both
dates using ENVI, and then analyzed for change detection using an unsupervised
isodata algorithm. An unsupervised isodata change detection analysis was also
run on the panchromatic band to see if any major changes could be detected at a
15-meter spatial resolution in black and white.
Results
Comparison
of Space Disaster Charter Maps and GIS Products
The different
emergency response products released by charter members showed quite a bit of
variation in regard to inundation extent. Much of this variation can be
ascribed to the satellites themselves, the type of data used by each respective
agency, and each map’s intended purpose. The UNOSAT map released on 12 March
2011 and the DLR map released on 16 March 2011 for example were purposed to
look at the precise extent of flooding in the aftermath of the disaster using
radar data from Radarsat-2 and TerraSAR-X (International Charter Space &
Major Disasters 2011). The maps released by SERTIT on 13 March 2011 and DLR on
18 March 2011 in contrast aimed to look at how far the wave penetrated the
coastline in a more general way using radar data from TerraSAR-X and the 2000 Shuttle
Radar Topography Mission (International Charter Space & Major Disasters
2011). This particular DLR map also had additional information regarding
population from LandScan in its original pre-digitized form (International
Charter Space & Major Disasters 2011). The JAXA product was the only
product of these five that relied upon analysis of visible inundation, and used
data from Cartosat-2 whose sensor only provides imagery in a single
panchromatic band (ISRO 2008).
Contextual Comparisons of Disaster Products
|
Agency
|
Date
|
Satellite
|
Type
|
Purpose
|
|
UNOSAT
|
12 March 2011
|
Radarsat-2
|
Radar
|
Flooding
|
|
SERTIT
|
13 March 2011
|
None; SRTM
|
Radar
|
Water Extent
|
|
JAXA
|
15 March 2011
|
Cartosat-2
|
Visual B&W
|
Flooding
|
|
DLR
|
16 March 2011
|
TerraSAR-X
|
Radar
|
Flooding
|
|
DLR
|
18 March 2011
|
LandScan
|
Radar
|
Population
|
Change
Detection of Wakabayashi Agriculture Using Remotely Sensed Data
For the
remotely sensed data acquired from Landsat 7, changes are readily apparent in
the visual spectrum and in the NDVI indices. No discernable patterns were
discovered in the change detection analysis we ran on the panchromatic bands.
From the naked eye it is possible to see that once green fields are now brown
and barren in the visual spectrum. This same brown area correlates with the
blue patches represented in the change detection map that was generated from
the two NDVIs that we created. Several red patches also appeared, suggesting
that those particular agricultural fields have seen an increase in their amount
of vegetation since the disaster hit despite being close to and sometimes
within the inundation extent.
Discussion
Constraints of Closed Systems
Of the five
products that were compared in our first analysis, only one (UNOSAT) had a
shapefile that was made publicly available by the agency that released it, and
only one used freely available data in its analysis (SERTIT). None of the
participating agencies discussed the processing of data or outlined the
algorithms used to develop their products. Raw data was also universally
unavailable to the public except for the SRTM data gathered in 2000 that SERTIT
used in their inundation model. These factors placed a couple limitations on
the accuracy and usefulness of our data in this study and further hinders
avenues of research going forward:
First, without
official vector based data sets there is no way to be sure that our digitized
extents accurately represent the boundaries outlined by each product.
Digitization in this study was limited by three primary factors: the reduced
resolution of publicly available products, errors of omission and commission
intrinsic to converting raster data into vector data, and other inevitable
human errors. Numerically quantifiable areal comparison between the different
products, which we initially intended to perform, would therefore be inaccurate
and misleading.
Second, a
complete lack of discussion on data processing and algorithms combined with the
unavailability of raw data makes future post-event comparisons using supervised
classification methods and ground control data impossible. There is no way to
compare the accuracy of the data produced by algorithms utilized by one agency
to those of another because explanations are not given and equations are not
expressed. This issue is particularly frustrating and problematic when looking
at the data provided by high-resolution radar satellites like Radarsat-2 and
TerraSAR-X. These two satellites, with their extremely high spatial resolutions
under 3 meters (Canadian Space Agency 2012, DLR 2012), appear to have captured
and displayed the extent of inundation with the most accurate and precise
detail, and arguably show the most promise for accurately modeling similar
future events.
The Promising Results From Open Systems
It appears
that the change in vegetation shown in our NDVI analysis was the result of salt
deposits left behind in the rice paddies as a result of the tsunami. High
levels of salt severely retard plant growth and decrease the overall yields of
crop producing plants. Salt hinders a plant’s ability to take up sufficient
water by negatively affecting vital processes such as photosynthesis. Although
many plants have a moderate tolerance to sodium, rice crops happen to be among
species that are most sensitive to changes in soil salinity. Once salt dissolves
into the soil, it is difficult to remove because of the way it attaches itself
to other elements (Plett 2009).
Rice is the
prime agricultural crop in Japan, and the harvest of the crop in Miyagi
prefecture during 2010 generated $818 million for the local economy. Vast
swaths of the region’s agricultural land was essentially poisoned by the salt
present in the water that the tsunami washed ashore and as a result rice yields
have dropped. According to the Japanese agriculture ministry, the difference in
harvested rice acreage between 2010 and 2011 was 11,400 acres, and as a result
of this drop in productivity the disaster has continued to affect the local
economy in negative ways by leaving fields barren. The tsunami not only wiped
out traditional rice crops, but also induced conditions unfit for future
cultivation because of the amount of salt that was deposited into the ground.
It has been determined that in total, 11 percent of the prefecture’s farmland
was damaged by this event. (Nagano 2012)
The blue patches
within the change detection represent a loss in vegetative cover since the
tsunami and the red patches represent an increase over the same time. It is
likely that the areas where vegetation is increasing are growing crops that are
known to be salt-resistant. We speculate that these areas could actually be
cotton fields, as news reports from around the world have noted that farmers in
the tsunami inundated zones have begun to cultivate this salt resistant crop
(Villemez 2011, Kaname 2011). Our analysis identifies land in Wakabayashi that
would be best suited for cultivating rice and that which would be best suited
for salt-resistant crops; an application that will be explored in further
detail in another section. In addition, a future direction for research in this
specific area has been identified and could revolve around quantifying the
amount of agricultural land that has been converted from rice to cotton fields.
Given that each crop gives off a separate spectral signature, these signatures
could be determined and then mapped out within the spatial subset. This should
be possible using the same freely available satellite imagery captured by
Landsat 7.
 |
| This image is of the study area and was taken by photgrapher Tetsuji Asano and published in the Asahi Shimbun newspaper on 11 September 2012. You can see barren fields on the right that were damaged by salt water. |
The Self-Contradictory Nature of Closed Access in Disaster Response
Situations
Having run
into several constraints with space disaster charter data during our study, we
believe that the closed nature of data access that occurs within its current
framework runs counter to one of the agreement’s major reasons for existing. As
stated by its founding signatory organizations in the preamble of its text, the
driving belief necessitating international cooperation was the ambitious idea
that such collaboration could, “increase the efficiency of services that may be
provided to crisis victims and to the bodies called upon to help them.”
(International Charter Space & Major Disasters 2012) By limiting data
access primarily to government organizations; thereby excluding the vast
majority of academic researchers from participation and avoiding the rigors of
peer review, the ability to refine current methods and the capacity to develop
new ones supportive of this noble purpose has been unwisely impeded. Given the
potential benefits that the field of remote sensing has to offer in supporting
disaster relief efforts across the globe, we feel that a modification to the
charter encouraging open access is both necessary and demanded by the ultimate
aim of the charter itself.
Open access of
data created under the charter can be beneficial to science and humanitarian
efforts in a variety of ways. As elucidated upon earlier, flexible use of data
can pose an opportunity in which algorithms can be juxtaposed and compared for
accuracy, opening paths to new and innovative approaches. As the current closed
access nature of the charter offers no way to test the accuracy of algorithms
used in previous disaster responses, future targeted rescue operations face the
increased possibility of running into unforeseen difficulties caused by errors
on the part of photogrammetrists and cartographers.
Furthermore, elaborating upon our
second research focus, researchers and policy makers often lack reliable
information capable of measuring recovery rates accurately. In the case of our
second study, a product using open access data was developed that has direct
policy implications. Using our product, it could be possible for Japanese
authorities to implement new land use regulations that will help accelerate the
recovery of agricultural production in Wakabayashi. A policy for example, that
reserves land unaffected by the inundation of salt water for rice cultivation
and requires salt-resistant crops to be grown in the inundation zone could
possibly reduce the length and severity of long term economic effects upon the
region by better targeting, and therefore maximizing its agricultural output.
Open access in the context of this
study has been equated with the policies of the United States Geological Survey
in allowing nearly free and unrestricted access to Landsat series data, but the
authors feel that it is unnecessary to rigidly tie the concept to these
practices. Open access in terms of the disaster charter could hypothetically
mean a private embargo between member organizations for a year before data is
released. Such a compromise would be gladly welcomed as it addresses several of
the points we have made in regard to why closed access hinders disaster relief
efforts. Given the inevitable nature of natural disasters, member organizations
of the space disaster charter in determining which direction they want to go on
this important issue literally hold the lives of hundreds of thousands of
people in their hands. They would be wise to remain constantly mindful of this
fact with every activation.
References
Canadian
Space Agency 2012
RADARSAT-2 innovation (http://www.asc-csa.gc.ca/eng/satellites/radarsat2/innovations.asp)
Accessed 11 December 2012
Central
Intelligence Agency 2012 The
world factbook (https://www.cia.gov/library/publications/the-world-factbook/geos/ja/html)
Accessed 10 December 2012
DLR 2012 TerraSAR-X – Germany’s radar eye in space
(http://www.dlr.de/eo/en/desktopdefault.aspx/tabid-5725/9296_read-15979/)
Accessed 12 December 2012
Indian
Space Research Organization 2008
Cartosat-2 (http://www.isro.org/satellites/cartosat-2.aspx)
Accessed 10 December 2012
International
Charter Space & Major Disasters 2011 Earthquake in Japan (http://www.disasterscharter.org/web/charter/activation_details?p_r_p_1415474252_assetId=ACT-359)
Accessed 23 November 2012
International
Charter Space & Major Disasters 2012 Charter On Cooperation To Achieve The Coordinated Use Of
Space Facilities In The Event Of Natural Or Technological Disasters Rev.3
(25/4/200).2 (http://www.disasterscharter.org/web/charter/charter)
Accessed 11 December 2012
Kaname,
Kakuta 2011 Takashimaya
to market goods made of cotton grown on tsunami-swamped rice paddies The Asahi
Shimbun (http://ajw.asahi.com/article/economy/business/AJ201109038917)
Accessed 13 December 2012
Nagano,
Yuriko 2012 Rice Farmers
Seek to Save Their Crops From Salt The New York Times
(http://www.nytimes.com/2012/04/23/business/global/rice-farmers-seek-to-save-their-crops-from-salt.html)
Accessed 11 December 2012
National Aeronautics and Space
Administration 2012 Landsat 7 (http://landsat.gsfc.nasa.gov/about/landsat7.html) Accessed
13 December 2012
National
Oceanic and Atmospsheric Administration 2011 Great Tohoku, Japan Earthquake and Tsunami, 11 March 2011 (http://www.ngdc.noaa.gov/hazard/honshu_11mar2011.shtml.)
Accessed December 1, 2012.
Plett,
Darren 2009 Na+
transport in glycophytic plants: what we know and would like to know Plant
Cell and Environment 33 612-626
Stryker,
Timother and Jones, Brenda 2009 Disaster
Response and the International Charter Program Photogrammetric Engineering
& Remote Sensing December 2009 1342-1343
Villemez,
Jason 2011 Cotton
Replaces Rice in Japan’s Salt-Soaked Fields PBS Newshour (http://www.pbs.org/newshour/updates/world/july-dec11/cotton_09-16.html)
Accessed 13 December 2012
Wulder,
Michael A, Masek, Jeffrey G, Cohen, Warren B, Loveland, Thomas R, and Woodcock,
Curtis E 2012 Opening
the archive: How free data has enabled the science and monitoring promise of
Landsat Remote Sensing of Environment 122 2-10
No comments:
Post a Comment