Land Cover and Forest Formation Distributions for St. Kitts, Nevis, St. Eustatius, Grenada and Barbados from Decision Tree Classification of Cloud-Cleared Satellite Imagery

Abstract. Satellite image-based mapping of tropical forests is vital to conservation planning. Standard methods for automated image classification, however, limit classification detail in complex tropical landscapes. In this study, we test an approach to Landsat image interpretation on four islands of the Lesser Antilles, including Grenada and St. Kitts, Nevis and St. Eustatius, testing a more detailed classification than earlier work in the latter three islands. Secondly, we estimate the extents of land cover and protected forest by formation for five islands and ask how land cover has changed over the second half of the 20th century. The image interpretation approach combines image mosaics and ancillary geographic data, classifying the resulting set of raster data with decision tree software. Cloud-free image mosaics for one or two seasons were created by applying regression tree normalization to scene dates that could fill cloudy areas in a base scene. Such mosaics are also known as cloud-filled, cloud-minimized or cloud-cleared imagery, mosaics, or composites. The approach accurately distinguished several classes that more standard methods would confuse; the seamless mosaics aided reference data collection; and the multiseason imagery allowed us to separate drought deciduous forests and woodlands from semi-deciduous ones. Cultivated land areas declined 60 to 100 percent from about 1945 to 2000 on several islands. Meanwhile, forest cover has increased 50 to 950%. This trend will likely continue where sugar cane cultivation has dominated. Like the island of Puerto Rico, most higher-elevation forest formations are protected in formal or informal reserves. Also similarly, lowland forests, which are drier forest types on these islands, are not well represented in reserves. Former cultivated lands in lowland areas could provide lands for new reserves of drier forest types. The land-use history of these islands may provide insight for planners in countries currently considering lowland forest clearing for agriculture.

[1]  A. Strahler Stratification of natural vegetation for forest and rangeland inventory using Landsat digital imagery and collateral data , 1981 .

[2]  Marla Perez-Lugo,et al.  When Fields Revert to Forest: Development and Spontaneous Reforestation in Post-War Puerto Rico , 2000 .

[3]  E. Helmer,et al.  Forest conservation and land development in Puerto Rico , 2004, Landscape Ecology.

[4]  J. R. Landis,et al.  The measurement of observer agreement for categorical data. , 1977, Biometrics.

[5]  José M. C. Pereira,et al.  Land-cover Mapping in the Brazilian Amazon Using SPOT-4 Vegetation Data and Machine Learning Classification Methods , 2006 .

[6]  E. Helmer,et al.  A comparison of radiometric normalization methods when filling cloud gaps in Landsat imagery , 2007 .

[7]  R. Mittermeier,et al.  Biodiversity hotspots for conservation priorities , 2000, Nature.

[8]  T. Mitchell Aide,et al.  Urban Expansion and the Loss of Prime Agricultural Lands in Puerto Rico , 2001 .

[9]  Matthew E. Watts,et al.  Effectiveness of the global protected area network in representing species diversity , 2004, Nature.

[10]  T. Mitchell Aide,et al.  Urban Expansion and the Loss of Prime Agricultural Lands in Puerto Rico , 2001, Ambio.

[11]  C. Brodley,et al.  Decision tree classification of land cover from remotely sensed data , 1997 .

[12]  P. L. Weaver,et al.  Forest resources of Puerto Rico , 1997 .

[13]  Limin Yang,et al.  Development of a 2001 National land-cover database for the United States , 2004 .

[14]  Russell G. Congalton,et al.  A review of assessing the accuracy of classifications of remotely sensed data , 1991 .

[15]  Limin Yang,et al.  COMPLETION OF THE 1990S NATIONAL LAND COVER DATA SET FOR THE CONTERMINOUS UNITED STATES FROM LANDSAT THEMATIC MAPPER DATA AND ANCILLARY DATA SOURCES , 2001 .

[16]  Ian Mcdonald The sugar industry of the Caribbean community (caricom) : An overview , 2004 .

[17]  A. Skidmore An expert system classifies eucalypt forest types using thematic mapper data and a digital terrain model , 1989 .

[18]  Eileen H. Helmer,et al.  Mapping the Forest Type and Land Cover of Puerto Rico, a Component of the Caribbean Biodiversity Hotspot , 2002 .

[19]  E. Helmer,et al.  Cloud-Free Satellite Image Mosaics with Regression Trees and Histogram Matching. , 2005 .

[20]  R. G. Wright,et al.  GAP ANALYSIS: A GEOGRAPHIC APPROACH TO PROTECTION OF BIOLOGICAL DIVERSITY , 1993 .

[21]  G. De’ath,et al.  CLASSIFICATION AND REGRESSION TREES: A POWERFUL YET SIMPLE TECHNIQUE FOR ECOLOGICAL DATA ANALYSIS , 2000 .

[22]  Eileen H. Helmer,et al.  The Forest Types and Ages Cleared for Land Development in Puerto Rico , 2007 .

[23]  W. Cohen,et al.  Mapping montane tropical forest successional stage and land use with multi-date Landsat imagery , 2000 .

[24]  R. DeFries,et al.  Classification trees: an alternative to traditional land cover classifiers , 1996 .

[25]  Jacob Cohen A Coefficient of Agreement for Nominal Scales , 1960 .

[26]  M. Fiorella,et al.  Determining successional stage of temperate coniferous forests with Landsat satellite data , 1993 .

[27]  George V. N. Powell,et al.  Assessing representativeness of protected natural areas in Costa Rica for conserving biodiversity: a preliminary gap analysis , 2000 .

[28]  T. Farr,et al.  Shuttle radar topography mission produces a wealth of data , 2000 .

[29]  R. Lawrence Rule-Based Classification Systems Using Classification and Regression Tree (CART) Analysis , 2001 .