Quality Assessment of MODIS Water Vapor Products in IR and Near-IR Bands over Iran

Water vapor, as one of the most important greenhouse gases in the atmosphere, plays a key role in hydrological cycles, climate change, and the global climate. Many parameters for the expression of water vapor in the atmosphere have been proposed by meteorologists, one of which is Precipitable Water Vapor (PWV). There are many ground-based and space-based methods to measure PWV. Meanwhile, radiosonde is considered as one of the most common and traditional tools for measuring this parameter. However, low temporal resolution, high cost, and lack of uniform coverage across the globe are some of the limitations of this technique. In the last two decades, GPS Meteorology due to unique features such as usability in any weather conditions, long-term stability, continuous observations with very high resolution, low cost, and PWV estimation with an accuracy level of about 2 millimeters has received a lot of attention. Although radiosonde and GPS are precise methods for estimating water vapor in the atmosphere, their observations are limited to the land. While satellite remote sensing methods can provide continuous observations of the distribution of water vapor on a regional and global scale. MODIS is one of the sensors capable of measuring atmospheric water vapor measurements, which is onboard the Terra and Aqua satellites. However, PWV products obtained from remote sensing data should be evaluated with respect to the reliable in situ data before application. The main purpose of this study was to use PWV estimates obtained from ground-based GPS receivers in order to statistically evaluate the accuracy of MODIS water vapor products in IR and Near-IR bands and different times of the day over Iran. The MODIS sensor, which is on board of the Terra and Aqua satellites, is able to provide water vapor products in the IR (both night and day) and Near-IR (day-only) bands. In order to evaluate MODIS PWV products over Iran, one year data of high temporal resolution GPS PWV values in 38 different stations in the country were considered as reliable values. For statistical analysis, water vapor values were extracted from the pixels with cloud-free conditions. Also, among the cloud-free pixels, that with the closest distance to the GPS station was selected. Moreover, the corresponding PWV values of GPS and MODIS with a maximum time difference of 10 minutes were selected for comparison. Initially, Near-IR PWV products were assessed separately for Terra and Aqua satellite data. The results showed a good agreement between the two sets of PWV measurements. The correlation values between the GPS PWV and the corresponding values of the MODIS Near-IR products varied in the range of 0.90 to 0.98. Average bias values indicated that MODIS Near-IR overestimated PWV in comparison with GPS over Iran. In addition, a comparison of Near-IR water vapor values extracted from Terra and Aqua datasets separately showed that the data quality of both satellites in this band is almost at the same level in terms of the correlation coefficient, average bias, and RMSE. In the next step, the MODIS IR PWV products were evaluated separately during the day and night with respect to the corresponding values obtained at the GPS stations. The maximum correlation between GPS and IR PWV products during the day and night was 0.7 and 0.64, respectively. Furthermore, the average bias of MODIS IR PWV data in the study area for day and night was found to be -0.38 and 3.11 mm, respectively. In other words, MODIS IR PWV products in the study area had, on average, a positive bias with a small amount during the day and a significant negative bias during the night. On the other hand, a comparison of daytime MODIS IR and Near-IR water vapor products revealed that the quality of IR PWV data was significantly lower than the Near-IR band and requires a suitable calibration method. The results of this study indicate that the MODIS Near-IR water vapor products had a high agreement with GPS PWV values with an average correlation coefficient of 0.95 in the study region. The mean bias and RMSE error of (GPS-MODIS Near-IR) PWV differences were -2.2 and 3.3 mm, respectively. A similar analysis of MODIS Near-IR PWV data from the Terra and Aqua satellites showed that almost both sets of water vapor data had the same accuracy. The average bias values of the MODIS IR PWV data compared to the GPS PWV for day and night were also investigated. Results showed that in the study area, MODIS IR products had a small positive bias during the day and significant negative bias at night. Examining the efficiency of the daytime MODIS water vapor products during the day, we found that the accuracy and precision of these data in the Near-IR band are much better than the IR band. Therefore, proper calibration should be made before employing the IR band.