Review of Satellite Remote Sensing Data Based Rainfall Estimation Methods (original) (raw)
Related papers
2009
An attempt was made to compare TRMM 3B42 products rainfalls with surface-based rain-gauge (RNG) rainfall obtained at 31 stations over Bangladesh for the years 1998-2002. Day-by-day rainfall amounts determined by TRMM and RNG are compared for continued 274 days (from 1 March to 30 November) in each year and at every station. Out of 274 days, averaged for 5 years rainfall over 31 stations, 97.08 % and 98.91 % days are detected as rainy day by TRMM and RNG respectively. Rainy days detected by TRMM matched 95.99 % of the days detected by RNG. On an average, TRMM can determine about 98.24 % of the RNG rainfall. The TRMM overestimates rainfall during pre-monsoon and underestimates during monsoon while alike during post monsoon period. Overall, TRMM underestimates rainfall in the heavy-rainfall regions of Bangladesh. This study also presents the vertical structure using TRMM-2A25 data and diurnal variation using TRMM-3B42RT data of precipitation from pre-monsoon to post-monsoon periods in and around Bangladesh. TRMM-2A25 data analysis reveals that pre-monsoon, monsoon and postmonsoon precipitations are strong, moderate and less intensified, respectively. Strong rain rates are found at higher altitudes for pre-monsoon and relatively at lower altitudes for later periods. From averages of 4 years of data, it is found that maximum rain rate in April, July and October is 114.19, 73.88 and 49.28 mm/h, respectively. In general, pre-monsoon echoes are high compared to monsoon and post-monsoon periods. However, the maximum echo top height of about 18.25, 18.8 and 18.25 km is found during the pre-monsoon, monsoon and post-monsoon periods, respectively. Analyzing TRMM-3B42RT data, it is found that the maximum rainfall over Bangladesh and northeast of the Bay of Bengal are appearing at 06 LST (local time) during monsoon period. In the same period the maximum rainfall over India is found at 18 LST. For the entire rainy season (March-November) the maximum rainfall over Bangladesh is occurred at 06 LST with a secondary maximum peak at 15 LST. The morning maximum rainfall at 06 LST over Bangladesh is confirmed after compared with the same obtained from ground-based rain gauge data. This analysis reveals that in Bangladesh the overestimation and underestimation of rainfall by TRMM in pre-monsoon and monsoon respectively depends on different vertical structures of precipitation fields in corresponding periods. xi With the understanding of rainfall characteristics in Bangladesh, this research work also extends development of a statistical procedure of quantitative estimation of rainfall in Bangladesh from combination of radar observations and rain gauge measurements. The uncertainty bounds associated with those estimates are also evaluated at a given space and time resolution. It is identified that there are six parameters controlling the various processing stages in radar rain algorithm. Sensitivity analysis showed that there is a host of parameter values coming from different gauge clusters in the parameter space that are equally acceptable as predictors of rainfall. Consequently, a methodology is devised to assess the uncertainty arising from errors in algorithm structure and parameter selection. Within this methodology, the algorithm calibration problem is formulated into the estimation of posterior probabilities of acceptable algorithm responses, thereby avoiding the concept of determining a likelihood values associated with errors between observed and estimated precipitation amounts derived through repetitive sampling of parameter space on the basis of Monte Carlo technique. The rainfall estimation from the method of Z-R relationship yielded the largest mean relative error of 43.8 % among the selected algorithms. The Kalman filtering technique was also applied in order to adjust radar rain estimation errors. The mean relative error with average calibration, radar-gauge adjustment and Kalman filter approache is respectively 14.5 %, 10.2 % and 7.9 %. The correlations between radar and gauge rainfall for different approaches are 0.584 (not corrected i.e. before error adjustment), 0.902 (with error adjustment but without filtering), and 0.963 (corrected with Kalman filtering). It is found that there is very little error adjustment made by the Kalman filter. Finally, this study also presents development of satellite based overland rainfall estimation from Space-borne Precipitation Radar and Passive Microwave data. The Tropical Rainfall Measuring Mission (TRMM) satellite carries two sensors that are very useful to precipitation: the precipitation radar (PR) and Microwave Imager (TMI). In this study PR's standard products, that is rain profile and rain type, are used to calibrate overland rain retrievals from the TMI channels. PR-TMI calibration is done for two different regions (GBM and Southern US), and results are compared to examine the significance of differences. Coincident PR and TMI data from six summer months is used for the overland microwave (MW) algorithm calibration. The calibration scheme developed by Grece and Anagnoustoe (2001) consists of (i) rain area delineation, (ii) convective/stratiform rain classification, and (iii) a multiple linear regression model for rain rate estimation. The current algorithm is also compared with the latest [version 6 (V6)] TRMM 2A12 xv
EVALUATION OF SATELLITE BASED NEAR-REAL TIME PRECIPITATION ESTIMATION OVER URBAN AREA
The flood damage potential in urban area is high and the accurate flood/rainfall forecasting will reduce the damage caused by urban flooding. For flood forecasting, there is a need for proper knowledge on time distribution of rainfall in real time basis and this can be achieved based on near real time satellite rainfall estimates. In this paper, the Tropical Rainfall Measuring Mission (TRMM) near real time rainfall product, TRMM-3B42 RT v7, is validated using Indian Meteorological Department (IMD) hourly gauge observation of Hyderabad city, India. 3-h cumulative rainfall is calculated from hourly observed rainfall and compared with TRMM 3-h rainfall estimates. Our results indicate that the TRMM-3B42 RT v7 gridded 3 hour precipitation is overestimating the rainfall intensity and it is underestimating high intensity (>17.6 mm/3-h) rainfall events. Further, the TRMM is mostly missing the small rainfall events having intensity less than 0.2 mm/3-h.
Improved rainfall estimation over the Indian region using satellite infrared technique
The GOES Precipitation Index (GPI) technique for rainfall estimation has been in operation for the last three decades. However, its applications are limited to the larger temporal and spatial scales. The present study focuses on the augmentation on GPI technique by incorporating a moisture factor for the environmental correction developed by . It consists of two steps; in the first step the GPI technique is applied to the Kalpana-IR data for rainfall estimation over the Indian land and oceanic region and in the second step an environmental moisture correction factor is applied to the GPI-based rainfall to estimate the final rainfall. Detailed validation with rain gauges and comparison with Tropical Rainfall Measuring Mission (TRMM) merged data product (3B42) are performed and it is found that the present technique is able to estimate the rainfall with better accuracy than the GPI technique over higher temporal and spatial domains for many operational applications in and around the Indian regions using Indian geostationary satellite data. Further comparison with the Doppler Weather Radar shows that the present technique is able to retrieve the rainfall with reasonably good accuracy.
Real-Time Rainfall Estimation Using Satellite Signals: Development and Assessment of a New Procedure
IEEE Transactions on Instrumentation and Measurement
This contribution presents a comprehensive methodology for the real-time estimation of the rain intensity from downlink satellite signals. The enhanced system leverages on extremely randomized tree classifiers to automatically perform rainfall detection along earth-satellite links and successively employs an improved procedure to determine the corresponding slant-path rain attenuation. The latter quantity is then exploited to yield real-time rainfall rate estimates with a 1-min time resolution. The accuracy of the proposed methodology is tested using the Ka-and Q-band propagation data, collected in two different sites (Milan and Madrid) and in the framework of the propagation experiments. The results demonstrate the reliability of the automated rain event detector, as well as a satisfactory accuracy in estimating the slant-path rain attenuation and the point rainfall rate. The accuracy is assessed both on a statistical and on an instantaneous basis through the evaluation of different error figures and by inspection of individual time series.
IEEE Transactions on Geoscience and Remote Sensing, 2000
A new technique has been developed to estimate rainfall at very fine scale (hourly rain rate at 0.05 • × 0.05 • spatial resolution) over India and associated oceanic regions (20 • S-40 • N, 40 • E-130 • E). By using infrared (IR) and 6.7-μm water vapor (WV) channel observations from Meteosat-7, a new rain index (RI) is computed. The index computation is composed of two steps. First, the IR and WV brightness temperatures are divided by their respective nonrainy thresholds to get the IR and WV rain coefficients. The product of these coefficients is defined as the RI. These RIs are collocated against rainfall from the Precipitation Radar (PR) on board the Tropical Rainfall Measuring Mission to develop a relationship between the index and the rain rate.
Rainfall Estimation method using Satellite imagery over South America
2006
The importance of the relationship between the life cycle of the mesoscale convective system (MCS) and the rainfall rate it produces has been reported in several works. In spite of that, a specific quantification of this relationship has not been found. Our aim was to find an empirical relationship between the characteristics that describe the MCS life cycle and the amount of rainfall rate it produces in order to develop a rainfall rate estimation algorithm. This paper reports a rainfall satellite estimation technique using the Precipitation Radar product (PR) onboard the TRMM Satellite, GOES IR (10.5 µm) brightness temperature (T b), an IR-VIS (0.65µm) cloud classification and radiative properties of clouds over the life cycle of deep convective systems. Numerous earlier studies focus on this subject using patch or pixel-based techniques. We use both techniques with satisfactory results when compared with the Hydroestimator technique. The algorithm first associates rain with the colder pixels belonging to a certain cloud type (convective clouds, cumulus and cold stratiform clouds). The rainfall estimation is carried out using the MCS properties (expansion and difference mean temperature among others), the internal brightness temperature (T b) variability of the pixel for every cloud type and some statistical assumptions. The method performs reasonably well in the case of convective, but also for stratiform rainfall, although it tends to overestimate rainfall rates values.
Rainfall estimation for real time flood monitoring using geostationary meteorological satellite data
Advances in Space Research, 2015
Rainfall estimation by geostationary meteorological satellite data provides good spatial and temporal resolutions. This is advantageous for real time flood monitoring and warning systems. However, a rainfall estimation algorithm developed in one region needs to be adjusted for another climatic region. This work proposes computationally-efficient rainfall estimation algorithms based on an Infrared threshold rainfall (ITR) method calibrated with regional ground truth. Hourly rain gauge data collected from 70 stations around the Chao-Phraya river basin were used for calibration and validation of the algorithms. The algorithm inputs were derived from FY-2E satellite observations consisting of infrared and water vapour imagery. The results were compared with the Global Satellite Mapping of Precipitation (GSMaP) near real time product (GSMaP_NRT) using the probability of detection (POD), root mean square error (RMSE) and linear correlation coefficient (CC) as performance indices. Comparison with the GSMaP_NRT product for real time monitoring purpose shows that hourly rain estimates from the proposed algorithm with the error adjustment technique (ITR_EA) offers higher POD and approximately the same RMSE and CC with less data latency.
Comparative study of performance of real-time satellite-derived rainfall in Swat Catchment
Arabian Journal of Geosciences, 2017
Most of the conventional models require rainfall data for realistic modeling results, and where ground data is scarce, remotely sensed data plays a vital role. For monitoring, hydrological models require near real-time observations to allow for effective planning and forecasting. However, monitoring rainfall in mountainous region is difficult because of inaccessibility and sparse gauge density. However, the accurateness of these satellite estimates over different spatial and temporal scales is unknown. The study intended at carrying out a comparative analysis of satellite rainfall estimates as a substitution for ground-based rainfall observations in the Swat Catchment. Limited availability of temporally continuous available data records in Pakistan has been a problem and has effected the reliability of modeling results. As well as, data is not freely available and cost is the biggest hindrance to its usage. So, remotely sensed data plays a vital role both in terms of timely availability and its free of charge. For this region, only two remotely sensed gridded data products are freely available, i.e., NOAA RFE CA and TRMM RT. Respective two products have been analyzed by various verification statistics. RFE CA proves better probability of detection, false alarm ratio, threat score, and equitable threat score than TRMM RT. The outcome of this comparative study concludes that for hydrological modeling purposes, RFE CA data is the best choice in this region. The annual bias for RFE CA and TRMM RT is 14% (over-estimation) and 18% (under-estimation) over the years having coefficient of determination with the ground-based data of 0.87 and 0.76, respectively, on annual basis. The result shows the suitability of RFE CA for effective monthly rainfall-runoff modeling in Swat Catchment, Khyber Pakhtunkhawa, Pakistan.
Satellite Rainfall Estimation in the South-Eastern Part of Bangladesh
International Journal of Scientific & Engineering Research, Volume 4, Issue 6, page2007-2013, 2013
In case of sufficient moistures contamination from BOB, the rain forecasting is challenging for the meteorologists over south and southeastern part of Bangladesh, especially in monsoon and post-monsoon period. Real time rainfall information is necessary for early warning rainfall which triggered hazards such as floods and landslides. Slow dissemination of measured rainfall information was considered as a serious obstacle in terms of the use of such information for early warning purpose. Satellite based rainfall estimation had been considered as an alternative to fulfill that demand. This research was addressed to estimate the satellite-based rainfall by using Tropical Rainfall Measuring Mission (TRMM) satellite data. Then, estimated rainfall was made to compare with surface-based rain gauge rainfall which was acquired from 08 weather stations over Bangladesh for the years of 2001-2009. At first, daily rainfall was estimated by satellite. Then, monthly, seasonally and yearly rainfall was also determined and compared it with surface rainfall. Temporal and spatial analysis was performed by estimated (TRMM) and observed (RG) rainfall and at same time for the assessment of accuracy some statistical parameters such as correlation coefficients, regression equations, biases and mean absolute errors were deermined for all stations. The TRMM overestimated rainfall during pre-monsoon and underestimated during monsoon while alike during post-monsoon period. Overall, TRMM underestimated rainfall in the heavy-rainfall regions of Bangladesh.
Journal of Japan Society of Civil Engineers, Ser. B1 (Hydraulic Engineering), 2011
Near real time rainfall information is necessary for early warning of rainfall triggered hazard such as floods and landslides. Remote sensing based rainfall estimation has been considered to be used to fulfill that purpose. This research is addressed to use geostationary based rainfall estimation by using Multi Transport Satellite (MTSAT) data which is blended with Tropical Rainfall Measuring Mission (TRMM) 2A12 datasets in order to provide near real time rainfall information, especially for hazard study purposes over Java Island, Indonesia. Comparison to TRMM Multi Precipitation Analysis (TMPA) datasets is performed. Spatial and temporal validation of those rainfall estimations is conducted by validating them with available rain gauge data during a rainy season in December 2007. Temporal validation result shows that TMPA demonstrated better statistical performance than MTSAT blended. However for the spatial correlation, MTSAT blended shows relatively better performance than TMPA.