ability to
ania. U.S.
n Landsat
Linescan
f the Soil
Resources
ania State
tained by
vity class
well with
ystem was
cs. Urban
) the most
iis kind of
ing sound
and many
‚OGY
c U.S. Air
Program's
MSP/OLS)
tion Land
state of
s acquired
hough the
ion of day
capability
hotometer.
cs of light
studies of
ave given
those of
inous U.S.
(Elvidge et al.. 1997). A city light GIS coverage was
derived from a 23 1-orbit composites of the DMSP/OLS
(Figure 1). The MRLC is a GIS map of land use/cover
classes for the U.S. Federal Region III. which includes
the states of Pennsylvania. Virginia. Maryland.
Delaware and West Virginia. and was produced by
mosaiking 1992 Landsat TM scenes in conjunction
with other data such as digital elevation models
(DEM). population census data. digital line graph data.
etc. (Vogelmann et al. 1998). The resulting urban
land use layer has 30 m ground resolution (Figure 2).
The source of soils data was the STATSGO database.
STATSGO was developed by the USDA-NRCS for the
entire U.S. at 1:250.000 scale (Bliss and Revbold.
1989). Soil productivity classes consisted of soil
productivity groupings previously computed using the
SRPG model (Sinclair et al.. 1996) and those of the
USDA-NRCS LCC system (Soil Survey Division Staff.
1993). The SRPG ratings were produced using soil
characterization data from the USDA-NRCS Map Unit
Record (MUR). a dataset linked to STATSGO mapping
units. For convenience. however. these ratings were
further grouped into four soil productivity classes
(Figure 3) varying from the most productive (4) to the
least productive (1) soils. On the other hand. the LCC
groups soils that have the same limitations/risks and
therefore that respond similarly to soil conservation
and management needs. This grouping is often used as
a measure of soil productivity in land evaluations for
land use planning (Liu and Craul. 1991: Nizevimana
and Petersen, 1998). The eight USDA-NRCS LCC
classes were regrouped into four classes varving from
most suitable (4) to least suitable (1) soils for Crop
production (Figure 4).
The magnitude of soil productivity loss because of
urbanization for cach combination of urban and soil
productivity source was determined by GIS overlay of
urban thematic maps with soil productivity layers. and
analysis of the productivity level of soils within the
urban land use class.
3. RESULTS AND DISCUSSION
The urban land use areas derived from two data
sources. Landsat TM and DMSP/OLS imagery. are
presented in Table 1. The percent arca under urban
land use was higher for DMSP/OLS (4.7%) than for
Landsat TM (3.7%). This represents a 16% increase in
urban land arca when the nighttime imagery is used as
a data source compared to Landsat TM. Therefore. a
large change in resolution (from 30 m to 2600 m)
caused only a small change (16%) in urban land use
area. This indicates that the DMSP/OLS was efficient
in capturing urban land locations and extent on the
land surface when compared to Landsat TM.
Table 1. Urban land use statistics.
Data source
^ Lands IM. ; DMSP/OLS
| lotalarea (ha) © 11.733.423 ssl 11.761.614
| Urban land (ha) } 335.990 1 ....511503.
% urban land 37 4.3
The areas of soils in four levels of soil productivity
groupings obtained using SRPG and LCC modeling
schemes are presented in Table 2. The distribution of
soils in different classes of productivity differed
between the two classification svstems. For example,
the percent areas of soils in high and moderately high
soil productivitv classes increased from SRPG to
LCC. The reverse was true for moderate and low
categories. The largest difference between the two
systems occurred for the moderately high soil
productivity categorv. The percent area for this
category was 38.4% for SRPG and 60.4% for the
LCC system. This represents a 57% increase from
SRPG to LCC.
Table 2. Distribution of soil productivity classes
derived using SRPG and LCC systems.
Soil productivity Type of model’
class
SRPG (%) : LCC(99)
Meh lad de da LL
| Moderately high | ~~ B47 60 |
Er | 40.) dnl A080 d.
Low 173 0.6
' Arca = % area of soils in each productivity class.
Discrepancies in results obtained using the two
systems may be expected because of differences in
how the classes are formed. For example, SRPG
ranks soils in terms of their suitabilitv for plant
growth. Classes are determined by defining ranges of
soil and landscape properties based on the knowledge
of the effect of each parameter on plant growth. On
the other hand. the LCC system is a qualitative
measure of soil limitations as related to soil
management. Classes are defined in terms of the
number of limitations; the higher the number. the
lower the soil productivity. This type of classification
appears subjective since soil assignments to classes
Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 461