A scaling technique was investigated to extrapolate the local information on land cover derived from high spatial resolution data (LANDSAT/TM) to more extensive area through low spatial resolution data (NOAA/AVHRR). The error of overlay between AVHRR data and TM data was evaluated to ensure the following analysis. Over a combined data set of TM and AVHRR, at first, land cover types were classified into three categories including vegetation, water, and dried non-vegetation with TM data, and next, AVHRR image density was statistically regressed with category mixture conditions derived from TM pixels in each AVHRR pixel. Based on the regression model, the land cover category mixing ratio was estimated from AVHRR data by extrapolate the model to the whole coverage of AVHRR data. As a result, we found that end-members of AVHRR data for each land cover category was under estimated. It is partially because that a large part of TM pixels are also mixture pixels. Taking this into account, we used the vegetation-soil-water index (VSWI) as a key to estimate the end-member of AVHRR data from related TM data. AVHRR VSW index is difficult to determine from AVHRR data only, because there are very few pure pixels for vegetation, soil or water to determine accurate end-members for each land cover category. In this study, we investigated a scaling method to estimate AVHRR VSW index by means of the regression analysis relating NOAA/AVHRR CCT counts in both of ch.1 and 2 with LANDSAT/TM VSW index.
A new prediction method for time series data with high missing data ratio is proposed. The method does not require the assumption that time series data is derived from a linear process. It is found that the prediction error of the proposed method is 8.4 to 90.0% smaller than that of the existing method of RLS (Recursive Least Square) method for DMSP (Defence Meteorological Satellite Program) /SSM/I (Special Sensor of Microwave/Imager) data.
The field campaign for the vicarious calibration of ASTER/VNIR and SWIR was conducted at Tsukuba test site on December 13 and 14 1997. A comparison of the measured surface reflectance of the grass field and asphalt at the test site, optical depth as well as Top of Atmosphere (ToA) radiance was made among four institutes, University of Arizona, Geophysical Survey of Japan, Kanazawa Institute of Technology and Saga University. Approimately 8% of descrepancy on the mesured reflectance is found among the institutes while about 3% of the estimated optical depth and around 2% of the estimated ToA radiance are also found. Further improvement on the surface reflectance measurement, in particular, is highly required.
A preliminary field experiment for ASTER vicarious calibration (VC) was held in Tsukuba, Japan, on December 13 and 14 in 1997. Activities for ASTER/TIR (channels 10 to 14) in the experiment were water temperature measurements on Lake Kasumigaura at NOAA14 overpass times and four radiosonde observations per day. In the present paper, the effectivity of the VC of ASTER/TIR using the lake in winter is evaluated based on results from these activities. First, the top-of-atmosphere brightness temperatures (TOA-BTs) for AVHRR channels 4 and 5 are calculated from the VC with the in-situ measurements, and show very good agreements (<0.3K) with the TOA-BTs from the onboard calibration. Then, the TOA-BTs for ASTER/TIR channels are calculated with radiosonde profiles and NCEP GDAS (global data assimilation system) profiles around the lake during the experimental period, and show very good agreements (<0.3 K) with each other ; because the atmosphere around the lake was stable during the period; and because the GDAS profiles around the lake have high accuracy owing to a nearby permanent weather station. Finally, the sensitivity of input parameters to the VC of ASTER/TIR is analyzed under conditions around the lake in winter. The results show that the measurement errors of water vapor and air temperature profiles by a radiosonde are small enough for VC, the ozone profile and visibility (aerosols) errors degrades the VC of channel 12 and transparent channels respectively, and the accuracy of VC is most sensitive to water skin temperature. In the future, the VC of actual ASTER/TIR channels using the lake in winter will be executed based on the results from the present paper.