Effect of rain gauge density over the accuracy of rainfall: a case study over Bangalore, India
© Mishra; licensee Springer. 2013
Received: 23 May 2013
Accepted: 9 July 2013
Published: 11 July 2013
Rainfall is an extremely variable parameter in both space and time. Rain gauge density is very crucial in order to quantify the rainfall amount over a region. The level of rainfall accuracy is highly dependent on density and distribution of rain gauge stations over a region. Indian Space Research Organisation (ISRO) have installed a number of Automatic Weather Station (AWS) rain gauges over Indian region to study rainfall. In this paper, the effect of rain gauge density over daily accumulated rainfall is analyzed using ISRO AWS gauge observations. A region of 50 km × 50 km box over southern part of Indian region (Bangalore) with good density of rain gauges is identified for this purpose. Rain gauge numbers are varied from 1–8 in 50 km box to study the variation in the daily accumulated rainfall. Rainfall rates from the neighbouring stations are also compared in this study. Change in the rainfall as a function of gauge spacing is studied. Use of gauge calibrated satellite observations to fill the gauge station value is also studied. It is found that correlation coefficients (CC) decrease from 82% to 21% as gauge spacing increases from 5 km to 40 km while root mean square error (RMSE) increases from 8.29 mm to 51.27 mm with increase in gauge spacing from 5 km to 40 km. Considering 8 rain gauges as a standard representative of rainfall over the region, absolute error increases from 15% to 64% as gauge numbers are decreased from 7 to 1. Small errors are reported while considering 4 to 7 rain gauges to represent 50 km area. However, reduction to 3 or less rain gauges resulted in significant error. It is also observed that use of gauge calibrated satellite observations significantly improved the rainfall estimation over the region with very few rain gauge observations.
KeywordsPrecipitation Rain gauge Remote sensing Satellite Hydrology
Rainfall is one of the most discontinuous atmospheric parameters due to its temporal and spatial variability. Indian economy is highly dependent on agriculture. Accurate rainfall estimates are essential for agricultural purposes. Chief source of rainfall over Indian region is Monsoon. India Meteorological Department (IMD) use rain gauge based gridded rainfall product developed by Rajeevan et al. (2005), to monitor rainfall over India. Rain gauges are conventional tools to quantify area averaged precipitation over land surface. Dense network of uniformly distributed rain gauge stations are used to estimate rainfall for a particular area (Mishra et al. 2011). The problem of installing optimum rain gauge network has been the subject of research over the years. Insufficient gauge density leads to error in representing the areal rainfall of a region. It is also found that rainfall is also affected by the distance of rain gauge stations from the grid point (Bhowmik and Das, 2007). The purpose of this study is to analyse the effect of rain gauge density over the accuracy of the areal daily accumulated rainfall over a region in Bangalore. Use of gauge calibrated satellite observations to fill the gap over poor gauge density region is studied in the present paper. Variation in the rainfall observations as function of inter-gauge distances is also studied in this paper.
Meteosat is a geostationary satellite launched in 1997 by the European Space Agency. It provides thermal infrared (TIR, 10.5-12.5 μm) and water vapor (WV, 5.7-7.1 μm) images every half an hour with a spatial resolution of 5 km. For the present study, data from 2009 to 2102 are used to study the impact of gauge calibrated satellite observations on areal rainfall estimation.
In the present study, Shepard (1968) inverse distance weighted interpolation technique has been used. It is based on the assumption that the interpolating region should be influenced most by the nearby points and less by the farthest points.
where p = 2, hi is the distance between the interpolation point and rain gauge location.
Rainfall variation with rain gauge spacing
361 cases of rain events were identified during rainy season of 2009–2012 to study the variation of rainfall with rain gauge spacing. Short lived intense rainfall events are defined as those with minimum hourly rainfall rate 15 mm and maximum life time of 3 hours in a day.
Rainfall study using gauge calibrated satellite observations
Apart from the southern part of Indian region, ISRO AWS rain gauge density over India is poor. The density over some places is such that only 1 rain gauge station (and sometimes no rain gauge) falls in 50 km × 50 km region. It is observed from the present study that rainfall values may change significantly within 15 km area. So, it is very difficult to quantify the rainfall on the basis of rain gauge observations over a region having poor rain gauge density. In this section, possibility of using rain gauge calibrated satellite observations to fill the gaps of missing rain gauge stations is analyzed. Past study (Mishra et al. 2010) shows that satellite estimates of rainfall matches well with that from rain gauge observations over well populated rain gauge area. These satellite rainfall estimates are based on a matchup between ground truth rainfall and rain signature from satellite.
Impact of rain gauge density in rainfall estimation
It is found from section 1 that rainfall in a region of 50 km box shows considerable variability. The estimate is affected by number of gauge stations in the area of study. Effect of rain gauge density over the accuracy of the rainfall estimation is studied in this section. For this purpose, total number of 274 rainy cases were considered during the years 2009–2012.
In the present paper, ISRO AWS rain gauge stations over southern part of India having a good rain gauge density are used to study the effect of rain gauge density and gauge spacing on rainfall estimation. Possibilities of using rain gauge calibrated satellite observations to represent vacant rain gauge stations are also explored in this study. Significant variations are observed even among stations located within about 15 km of each other. Error increases with increase in rain gauge spacing. It may also be concluded that 4–6 rain gauges give reasonable accuracy in daily rainfall estimation in a 50 km × 50 km area. There are scopes to use rain gauge calibrated satellite observations to represent the rain gauge station in area with poor rain gauge density. The technique described here may be used to estimate the rainfall over the area having insufficient number of rain gauges. Homogeneous distribution of rain gauges having sufficient number of equally spaced gauges form a perfect network to monitor the rainfall accurately over a region.
I acknowledge the MOSDAC for providing ISRO AWS rain gauge data. Meteosat data from ESA used in this study is also thankfully acknowledged. Useful discussions with Prof J. Srinivasan of Divecha Centre for climate change, Indian Institute of Science (IISc), India, are appreciated. Significant part of the work presented in this paper was done while the author was with Divecha Centre for Climate change, IISc. Financial support from National Science Council of Taiwan under grants NSC96-2111-M-001-005-MY3 is thankfully acknowledged. The author is thankful to the anonymous reviewers for their useful comments to enhance the quality of this paper.
- Bhowmik SKR, Das AK: Rainfall analysis for Indian monsoon region using the merged rain gauge observations and satellite estimates: Evaluation of monsoon rainfall features. J. Earth Syst. Sci 2007, 116(3):187-198. 10.1007/s12040-007-0019-1View ArticleGoogle Scholar
- Mishra A, Gairola RM, Varma AK, Agarwal VK: Remote sensing of Precipitation over Indian land and oceanic regions by synergistic use of multi-satellite sensors. J Geophys Res 2010., 115: D08106 10.1029/2009JD012157Google Scholar
- Mishra A, Gairola RM, Varma AK, Agarwal VK: Improved rainfall estimation over Indian land oceanic regions using satellite infrared technique. Adv Space Res 2011, 48: 49-55. 10.1016/j.asr.2011.02.016View ArticleGoogle Scholar
- Rajeevan M, Bhate J, Kale JD, Lal B IMD Met Monograph No: Climatology 22/2005, 27. Development of a high resolution daily gridded rainfall data for the Indian region 2005. Available from National Climate Centre, IMD Pune. email@example.com Google Scholar
- Shepard D: A two dimensional interpolation function for irregularly spaced data, Proc. ACM Nat. Conf 1968, 517-524.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.