Analysis of Agricultural Land Losses Using DMSP Nighttime Lights - - PowerPoint PPT Presentation

Analysis of Agricultural Land Losses Using DMSP Nighttime Lights

Kanichiro Matsumura, Kwansei Gakuin University Christopher D. Elvidge, NOAA Aug 9, 2010

Abstract

Nocturnal light data, collected by multiple satellites in 1992, 2000, and 2008, are used to analyze agricultural land losses. Overlaying DMSP datasets on vegetation datasets. We attempted to calculate agricultural land losses by using spatial information for 1992, 2000, and 2008.

GDP (nominal value) and light volume in 2002

Regression Coefficient Constant Correlation Coefficient Significance 0.87 12.01 0.823 1%

Demand for Land

Economic development and increase in the population result in the demand for lands for industrial areas, roads and

• infrastructure. Land use and cover change

have

• ccurred.

Agricultural production plays an important role in coping with the population increase and is calculated from yield and harvested area. The Food and Agricultural Organization (FAO) provides a time series of the harvested area by country.

Reclassifying night time data

Where light is observed as “value 1”, and the areas where light is off “value 0” for the datasets in 1992, 2000, and 2008.

2008 1992 2000

Multiplying reclassified datasets

We multiplied 1 for the dataset in 1992, multiplied 10 for the dataset in 2000 and multiplied 100 for the dataset in 2008. The points where light is observed in 1992, 2000, and 2008 were individually expressed as “1,” “10,” and “100” respectively.

Multiplying reclassified datasets

The points where light is observed in 1992, and 2000, as “11,” those in 1992, and 2008, as “110,” and those in 2000 and 2008, as “110.”

Multiplying reclassified datasets

The obtained datasets for western part of Japan and China.

Reclassifying vegetation data

Vegetation data sets developed by the Global Mapping Project conducted by Geographical Survey Institute (GSI).

Reclassifying vegetation data

Cropland, paddy field and cropland/other vegetation as agricultural land and reclassified these 3 legends as value “1”.

Crop Land Paddy Filed Vegetation and Other

Analysis

Assuming that night time lights

• n

agricultural land express agricultural land losses. We multiply the night-time datasets by the agricultural land. Subsequently, we can estimate the change of agricultural land losses in 1992, 2000, and 2008. We used the zonal statistics tools

• f

ArcGIS software to calculate the cell number

• f

the agricultural land with night-time.

Analysis

The vegetation value for 1992 is obtained using the following formula: total – 1 – 11 – 111: For 2000, using the following formula: total – 10 – 11 – 110-111: For 2008, using the following formula: total – 100 – 101 – 110 -111. We assume the vegetation for 1992= to be 100.

Comparing Statistical Data

Crop land change (x-axis: statistical value, y-axis:gridded value) for 1992, 2000 , and 2008

Comparing Statistical Data

Paddy field change (x-axis: statistical value, y-axis: gridded value) for 1992, 2000 , and 2008

Comparing Statistical Data

Cropland/vegetation change (x-axis: statistical value, y-axis: gridded value) for 1992, 2000 , and 2008

Comparing Statistical Data

Cropland/vegetation change (x-axis: statistical value, y-axis: gridded value) for 1992, 2000 , and 2008

Conclusion

Overlaying global scale night time light data for 1992, 2000, and 2008 on the vegetation map, we tried to calculate the agricultural land losses. The results were compared with the country based statistical value. The areas of crop and crop/vegetation change drastically year by

• year. According to farmers, it is important

to plant crops on the basis of producer

• prices. The change does not appear as

big with respect to rice.

Future works

For future research, we should consider the coherence of association between gridded vegetation and FAO-STAT more precisely. we need to consider a newly developed agricultural area for a type bush in Brazil. We also take into account the intensity of

• lights. The possibilities of using DMSP

datasets on agricultural land losses are investigated.

Acknowledgements

Night time image and data processing was performed by NOAA’s National Geophysical Data Center. The DMSP data were collected by the US Air Force Weather Agency

References

• 1. Night time data,Available at

http://www.ngdc.noaa.gov/dmsp/global_composites_v2.html

• 2. Elvidge, C.D., Sutton,P.S., Baugh, K.E., Tuttle, B.T., Howard, A.T., Erwin, E.H., and

Bhaduri, B., A Global Poverty Map Derived From Satellite Data, 2006, Available at:http://www.ngdc.noaa.gov/dmsp/download_poverty.html

• 3. Nakaya, T.: A Grid-surface Projection of Urban and Rural Population in China, 1990-

2050, in Otsubo, K. ed.: Study on the Processes and Impact of Land-use Change in China-Final Report of the LU/GEC Second Phase (1998-2000) - , Centre for Global Environmental Research, 2002, pp. 89-98.

• 4. Imhoff, M. L., Lawrence, W. T., Stutzer, D. C. and Elvidge C. D.: A Technique for

Using Composite DMSP/OLS “City lights” Satellite Data to Map Urban Area, Remote Sensing Environment 61, 1997, pp. 361-370

• 5. FAO STAT (2009) Available at: http://faostat.fao.org/site/567/default.aspx
• 6. GSI, Chiba University, Collaborating Organizations. Original data set Available

at:http://www1.gsi.go.jp/geowww/globalmap-gsi/gm-gaiyo.html