ERS Wind Scattermeter Data For Vegetation Monitoring In The Thar Desert
Haroon Stephen, Robert Schumann, Volkmar, Kiyoshi Honda and Keaw Nualchawee
Space Technology: Application and Research Program,
Asian Institute Of Technology,
P.O.# Klong Luang 10120, Pathum Thani, Thailand.
Tel (66)-2-524-5579 Fax (66)-2-524-5596
E-mail:haroom@ait.ac.th
Abstract
Since summer 1991 the Scattermeter aboard European Remote sensing satellite ERS-1 have provide global radar backscatter measurements of the Earth's surface. Over the oceans these data are used operationally for wind field retrieval and possible land applications have been under investigation. Due to the global coverage, the continuity of the data the wide range of incidence angle these data have been found to provide valuable information despite the coarse resolution. Furthermore, the day/night and all-wether operation capability enables data acquisition possible round the clock and under cloudy and rainy conditions. The geographical parameters of the land are reflected in the incidence angle diversity of the ERS Scattermeter. In this research we analyze normalized radar cross section (NRCS) measurement obtained over Thar Desert in Pakistan for the year 1994. Spatial variations in the radar cross section are compared with vegetation parameters obtained from AVHRR data. Seasonal as well as inter- annual variations are investigated by correlating the radar backscatter with NDVI data. Vegetation signatures from Scattermeter are analyzed which will help vegetation monitoring in the area which lack AVHRR data due to adverse wetland conditions.
Introduction
The first European Remote Sensing Satellite (ERS-1) WITH c-BAND (5.3 GHz) wind Scattermeter on board has been observing oceans for global wind measurements since July 1991. The launch of its sister ERS-2 in April 1995 enabled the uninterrupted supply of the global normalized radar cross section (NRCS) data to be continued towards the turn of the century. The ERS Scattermeter operator at vertical polarization (VV) and provides quasi-simultaneous views using three antenna which look 450 forward, perpendicular and 450 backward with respect to the satellite flight direction. The incidence angle varies from 250to590 for the fore-and aft-beam and from 180 to 470 for the mid-beam antenna. Different incidence angles and look directions provide enhanced geometric sensitivity of the backscatter measurements. The sensor illuminates a swath of 500 km wide to the right hand side of the satellite ground track. The data has an along track and cross track resolution of 50 km which is resample to a 25 km grid. Scattermeter acquisition is sun independent and is not hampered by clouds. The instrument achieves global coverage within 3-4 days (Wiseman, et.al.1996).
Thar desert is situated between 650-750E and 250-300 N. With a very low annual rainfall of less than 250 mm, it has a hot arid climate. Fig 1 shows the study area for this research, which lies at the border og\f Pakistan and India. In the west of the desert the river Indus with a wide patch of vegetation on both sides, and in the south of the desert is the coastal area of Arabian sea. Small amounts of unevenly distributed rains parts of the year cause
vegetation changes which are cyclic in the nature.
furthermore, the flucuations in the discharge of the
river Indus passing through the desert also affect the
semi desert part (Stephen, et.al1981).
Figure 1: Map of study area
Data set
The investigations are made with following
Available data:
Meteorological Data
Rainfall data for 16 ground stations in the study area from 1991 to 1995 was used which was collected from the Pakistan Meteorological Centers.
Radar backscatter data
Backscatter data of the study area is extracted from global NRCS data set distributed on CD-ROM by Center ERS archiving et de Treatment (CERSAT). Spatial radar backscatter maps of 45*40 pixels were prepared. Monthly averages of fore-beam radar backscatter for far range incidence angles (400-570).
Vegetation Cover Data
Atles of Pakistan (Survey of Pakistan,1981) and 12NOAA AVHRR images for the year 1994 were used to calculate Normalized Difference Vegetation Index (NDVI) for determination of vegetation cover variations in the study area. The NDVI data was resampled to a 25 km* 25 km grid scale by averaging the data within the cell in order to make it consistent with NRCS data.
Methodology
On the basis of NDVI variations in the area, the ground stations were categorized into following zones:
-
Desert
- Semi Desert
- Vegetated area
- Mountains
Investigations were made on point as well as spatial basis and in first case relationships between different parameters i.e. radar backscatter, incidence angle and rainfall were studied. Time series analysis if backscatter and rainfall was done for four zones in the study area. Spatial analysis of the scatterometer data was made comparison with NDVI data. While investigating spatial behavior the information extracted from point based analysis was utilized wherever applicable. Spatial time series and Hovmoeller diagrams were prepared which is the time series of a slice cut the study area and concatenated together.
