Shai Kaplan

Object vs. pixel: a systematic evaluation in urban environments

The objective of this paper is to compare object-based and per-pixel classifiers in a systematic manner using high resolution urban imagery. The prevailing opinion is that object-based methods perform better than single-pixel classifiers, but there has been no formal investigation of this claim using multiple images and identical training samples in a detailed land-cover classification. Furthermore, there has been no standardized study of how different object-based segmentation and scale parameters improve high resolution urban classifications. We used two subsets of QuickBird over Phoenix and Scottsdale, Arizona, to test these issues. Our results show that small-scale segmentation (10) produces higher accuracy. A combination of equally balanced shape and spectral homogeneity (0.5) with compactness parameter of 0.5 is the most effective for image segmentation. The highest overall accuracy was achieved using a per-pixel Minimum distance classifier, but it was only marginally more accurate than the object-based classification.

Understanding the Impact of Urbanization on Surface Urban Heat Islands—A Longitudinal Analysis of the Oasis Effect in Subtropical Desert Cities

We quantified the spatio-temporal patterns of land cover/land use (LCLU) change to document and evaluate the daytime surface urban heat island (SUHI) for five hot subtropical desert cities (Beer Sheva, Israel; Hotan, China; Jodhpur, India; Kharga, Egypt; and Las Vegas, NV, USA). Sequential Landsat images were acquired and classified into the USGS 24-category Land Use Categories using object-based image analysis with an overall accuracy of 80% to 95.5%. We estimated the land surface temperature (LST) of all available Landsat data from June to August for years 1990, 2000, and 2010 and computed the urban-rural difference in the average LST and Normalized Difference Vegetation Index (NDVI) for each city. Leveraging non-parametric statistical analysis, we also investigated the impacts of city size and population on the urban-rural difference in the summer daytime LST and NDVI. Urban expansion is observed for all five cities, but the urbanization pattern varies widely from city to city. A negative SUHI effect or an oasis effect exists for all the cities across all three years, and the amplitude of the oasis effect tends to increase as the urban-rural NDVI difference increases. A strong oasis effect is observed for Hotan and Kharga with evidently larger NDVI difference than the other cities. Larger cities tend to have a weaker cooling effect while a negative association is identified between NDVI difference and population. Understanding the daytime oasis effect of desert cities is vital for sustainable urban planning and the design of adaptive management, providing valuable guidelines to foster smart desert cities in an era of climate variability, uncertainty, and change.

First order approximation of Broadband Directional Albedo with High Resolution Quickbird Imagery: a case study for arid urban areas

Albedo is a key forcing parameter controlling the planetary radiative energy budget and its partitioning between the surface and the atmosphere. Characterizing and developing high resolution albedo for an urban environment in arid regions is important because of the high urbanization rate in these regions and because of the high land-cover heterogeneity within urban settings. Using a Monte Carlo simulation of a multi-variable regression, we (a) correlate directional solar reflectance (albedo) ground measurements from Phoenix, AZ, with four narrowband reflectance data from QuickBird, and (b) developed a new set of coefficients for converting QuickBird narrowband reflectances to albedo. The albedo models were then applied to a second image over Las Vegas, NV, to assess their feasibility and accuracy. Two wavebands, visible-near infrared (VNIR) and total shortwave albedo, were evaluated for two reflectance models: surface and top-of-atmosphere. Results show that it is possible to accurately estimate directional albedo from high resolution imagery, specifically QuickBird, with the most accurate result from an atmospherically corrected VNIR model. The methodology presented in this paper could thus be applied in other urban areas to obtain a first order estimation of albedo. The new set of coefficients can be applied as first order albedo estimate by researchers, urban planners, developers and city managers interested in the influence of high-resolution albedo on a myriad of urban ecosystem processes.

Estimating irrigated agricultural water use through landsat TM and a simplified surface energy balance modeling in the semi-arid environments of Arizona

0099-1112/12/7808–849/

Monitoring land use and land cover changes in Turkmenistan using remote sensing

3.00/0

Assessment of pasture degradation in Turkmenistan using remote sensing

In Turkmenistan the most prominent cause for desertification is inappropriate land use practices. The natural arid pastures have limited carrying capacity and any changes of the fragile balance can lead to the destruction of this valuable resource. One of the most appropriate tools for monitoring these processes is change detection through remote sensing imagery. Accurate monitoring of changes on the Earths surface is important to understand the relationship between man and nature and to provide decision makers with relevant information. The information on vegetation change is the most important of these relationships. Vegetation cover is also a useful indicator of the magnitude of land degradation that is easily assessed by multispectral remote sensing. The reduced vegetation cover causes an increase in albedo, which can also be monitored by remote sensing. The combination of these two parameters can give us a better map of the pasture status and its degradation rate. Landsat TM and ETM+ images were processed to maps of land use/land cover changes in northern Turkmenistan. The data were further processed in GIS and revealed the shrinking and the degradation of the pasture area. From the 1970s a total of ~4000km 2 of pasture were transformed into agricultural land, increasing the grazing pressure in the remaining areas. By applying advanced techniques for image based end-member retrieval and spectral mixture analysis a sub-pixel fraction was obtained for each end-member. The fractions of soil and vegetation emphasize the most degraded/rehabilitated sectors of the study area. Our results indicate the reduction of vegetation in specific areas while most of the desert experiences an increase in the vegetation cover. Our current study focuses on combining the spectral mixture analysis products with other degradation criteria such as change detection using albedo and vegetation indices to produce a more detailed assessment and understanding of the processes leading to these changes.

Micro-scale urban surface temperatures are related to land-cover features and residential heat related health impacts in Phoenix, AZ USA

Nomadic and semi-nomadic livestock breeding is a significant income for Turkmenistans economy. Thus, natural vegetation is an important resource for the area. The natural desert pastures of Turkmenistan have limited carrying capacity, and any changes of the fragile balance can lead to the destruction of this valuable resource. Since the 1980s, no research has been carried out concerning the ongoing changes of vegetation cover despite dramatic political and economical changes that took place throughout central Asia. The primary results of this research show the potential of remote sensing in general and specifically Spectral Mixture Analysis (SMA) for vegetation mapping. The information on vegetation change is important to quantify man-nature relationship. From this information vegetation cover is a useful indicator of the magnitude of change. Landsat TM and ETM+ images were processed, and maps of land use/land cover changes in northern Turkmenistan were produced. From the 1970s about 4000 km2 of natural pastures were transformed into irrigated agricultural land, theoretically increasing the grazing pressure in the remaining areas. By applying SMA based on field end-members signatures, a sub-pixel fraction was obtained for each end-member. Our results indicate that most of the desert experiences vegetation rehabilitation.