Inter-comparison of retrievals of Integrated Precipitable Water Vapour IPWV) made by INSAT-3DR satellite-borne Infrared Radiometer Sounding and CAMS reanalysis data 
with ground-based Indian GNSS data

Abstract. The spatiotemporal variations of integrated precipitable water vapor (IPWV) are very important to understand the regional variability of water vapour. Traditional in-situ measurements of IPWV in Indian region are limited and therefore the performance of satellite and Copernicus Atmosphere Meteorological Service (CAMS) retrievals with Indian Global Navigation Satellite System (GNSS) taking as reference has been analyzed. In this study the CAMS reanalysis retrieval one year (2018), Indian GNSS and INSAT-3DR sounder retrievals data for one & half years (January-2017 to June-2018) has been utilized and computed statistics. It is noticed that seasonal correlation coefficient (CC) values between INSAT-3DR and Indian GNSS data mainly lie within the range of 0.50 to 0.98 for all the selected 19 stations except Thiruvanathpuram (0.1), Kanyakumari (0.31), Karaikal (0.15) during monsoon and Panjim (0.2) during post monsoon season respectively. The seasonal CC values between CAMS and INSAT-3DR IPWV are ranges 0.73 to .99 except Jaipur (0.16) & Bhubneshwar (0.29) during pre-monsoon season, Panjim (0.38) during monsoon, Nagpur (0.50) during post-monsoon and Dibrugarh (0.49) Jaipur (0.58) & Bhubneshwar (0.16) during winter season respectively .The root mean square error (RMSE) values are higher under the wet conditions (Pre Monsoon & Monsoon season) than under dry conditions (Post Monsoon & Winter season) and found differences in magnitude and sign of bias of INSAT-3DR, CAMS with respect to GNSS IPWV from station to station and season to season. This study will help to improve understanding and utilization of CASMS and INSAT-3DR data more effectively along with GNSS data over land, coastal and desert locations in term of seasonal flow of IPWV which is an essential integrated variable in forecasting applications.



Introduction 33
The vertically integrated precipitable water vapour (IPWV) content in the atmosphere is a 34 parameter of great importance in all studies of the atmosphere and its properties through the year 35 in all seasons. The assessment of IPWV is done by many ways as in situ or remote sensing 36 measurements. The in situ measurements have limited coverage, expensive and require 37 maintenance of all the time. Remote sensing instruments, especially absorption in the infrared and 38 microwave region of solar spectrum have wide coverage, cheaper, almost maintenance free but 39 needs to be validated their retrieval performance and inter comparison before applying in the 40 operational meteorological service domain. Water vapour, one of the most influential constituents 41 of the atmosphere, is responsible for determine the amount of precipitation that a region can receive 42 (Trenberth et al, 2003). Integrated precipitable water vapor (IPWV) is a meteorological factor that 43 shows the amount of water vapour contained in the coloumn of air per unit area of the atmosphere 44 in terms of the depth of liquid (Viswanadham et al., 1981). The surface radiation is completely 45 absorbed by atmospheric water vapour on its way to the satellite. Each absorbing water vapour 46 molecule emits radiation according to Planck's law, mainly depending on its temperature and the 47 extent of absorption differs depending on the wavelength, the satellite sees different levels of 48 atmosphere. 49 Geo-stationary Earth Orbit (GEO) satellites can produce data more timely and frequently. The 50 retrieved high temporal resolution, Integrated Precipitable Water vapour (IPWV) from GEO 51 satellites sensor data can be utilized to monitor pre-convective environments and predict heavy 52 rainfall, convective storms, and clouds that may cause serious damage to human life and 53 infrastructure (Martinez et  shown strong seasonal variations. 72 The behavior of coastal regions are generally different from inland and desert stations as coastal 73 regions is greatly influenced moisture advection from breezing of the seas, which is the cause of 74 the continuous increment of IPWV even after the air temperature decreased (  to application of retrieval algorithm. Based on this, retrieval algorithm has option for retrieving 125 the vertical profiles at 30 km (3 × 3 pixels) and 10 km resolution (each pixel). The Sounder has 126 eighteen narrow spectral channels in shortwave infrared, middle infrared and long wave infrared 127 regions and one channel in the visible region. The ground resolution at nadir is 10 × 10 km for all 128 nineteen channels. Specifications of sounder channels are given in Table-

Scan Strategy of INSAT-3DR Sounder 152
The Sounder measures radiance in eighteen IR and one visible channel simultaneously over an 153 area of area of 10 km x 10 km at nadir every 100 ms. Using a two-axes gimballed scan mirror, this 154 footprint can be positioned anywhere in the FOR. A scan program mode allows sequential 155 sounding of a selected area with periodic space and calibration looks. In this mode, a 'frame' 156 consisting of multiple 'blocks' of the size 640 km x 640 km, can be sounded. The selected frame 157 can be placed anywhere within a 24º (E-W) x 19º (N-S) FOR. It takes almost three hours to sound 158 an area of 6400 km x 6400 km in size. The full aperture internal Black-body calibration is 159 performed every 30 min or on command based whenever. This enables the derivation of vertical 160 profiles of temperature and humidity. These vertical profiles can then be used to derive various 161 atmospheric stability indices and other parameters such as atmospheric water vapor content and 162 total column ozone amount. Figure  This scanning strategy is kept in such a way that sounding over an Indian land mass area will be 166 available every hour. Scan strategy and area of coverage of INSAT -3DR is shown below in the 167 Figure 1. 168

IMD IPWV observation network 169
The ground based GNSS IPWV estimated at a high temporal sampling (15 minute) data (January 170 2017-June 2018) of Indian GNSS network is processed at satellite division of India 171 Meteorological Department, Lodi Road, New Delhi. The data is processed daily by using the 172 Trimble Pivot Platform (TPP) software. The data is used operationally and archive as daily, 173 weekly, monthly as well as seasonal basis for future utilization and dissemination to the users, 174 researchers as per the official norms. Tome series of three years of GNSS data is prepared to 175 generate the diurnal variation of IPWV. An elevation angle of greater than 5° is set for all stations 176 to avoid the satellite geometry change and multipath effects. This is an optimum setting as a higher 177 cut off angle (> 5°) may introduce dry bias in the IPWV estimation and notable 0.8 mm error in 178 IPWV (Emardson et al., 1998). 179

INSAT-3DR and GNSS retrievals matchup criteria 180
The assessment of accuracy of INSAT-3DR satellite retrieved IPWV with 19 GNSS stations in 181 different geographical locations which are located in coastal, inland and desert regions over the 182 Indian subcontinent and are shown in the Table 2     3DR during pre-monsoon season. RMSE ranges between 3.5mm to 10mm (Table 4). mountainous land together along with topographically diverse terrains around these stations. 286 Similar behavior is also seen in annual analysis of IPWV in coastal stations with the above said 287 reasons. 288 It is seen that discrepancies arise because the wet mapping functions that used to map the wet delay 289 at any angle to the zenith do not represent the localized atmospheric condition particularly for 290 Narrow towering thunder clouds and non-availability of GPS satellites in the zenith direction 291 (Puviarasan et al., 2020).  (Table  298 4).The study showed differences in the magnitude and sign of bias of INSAT-3DR with respect to 299 GNSS IPWV from station to station and season to season. 300   prevails over the regions of higher moisture availability or water content in the Atmosphere. 401 (Figure 8). 402

Distribution and Variability of IPWV retrieved from INSAT03DR and CAMS reanalysis 403
The annual mean value and standard deviation of both the retrievals INSAT -3DR sounder and 404 CAMS reanalysis data sets are presented in Figure 9. The standard deviations of CAMS reanalysis 405 retrievals data set are appreciably high (0.0 to 14 mm) in both land and ocean areas as compared 406 to INSAT-3DR retrievals. This variation of higher spread from mean values is may be due to the 407 drier bias present in the CAMS reanalysis data sets (Inness et al, 2019) with coarser resolution as 408 compared to INSAT-3DR retrievals. 409 The mean IPWV values vary in the range of 0-50mm depending upon the region and prevailing 410 weather system affected throughout the year. Larger mean IPWVs occur in the coastal regions of 411 Indian Ocean regions compare to inland and desert regions due to warm air condition as compared 412 to inland and ocean. The south foothill of Himalayas has the largest PWV variation with a SD ~16 413 mm (Figure 9). This is attributed to the monsoon season that results in large changes in 414 precipitation at different seasons in these regions. The seasonal distribution of mean IPWV and 415 standard deviation of CAMS and INSAT-3DR for monsoon and post monsoon increased in CAMS 416 data as compared to INSAT -3DR retrievals due to wet bias present in the CAMS data sets ( Figure  417 10). 418 Standard deviation (SD) between CAMS reanalysis and Indian GNSS retrievals is more dispersed 419 from  This study will help to improve understanding regarding representation of uncertainties associated 507 with land, coastal and desert locations in term of seasonal flow of IPWV which is an essential 508 integrated variable in forecasting applications. 509 5. Acknowledgements: Authors are grateful to Director General of Meteorology for providing 510 data and support to accomplish this work and also thankful to the CAMS global web site data 511 (https://ads.atmosphere.copernicus.eu) link for providing the data for the above study. 512 6. References. 513