Assessment of Satellite-Based Rainfall Products Using a X-Band Rain Radar Network in the Complex Terrain of the Ecuadorian Andes

Ground based rainfall information is hardly available in most high mountain areas of the world due to the remoteness and complex topography. Thus, proper understanding of spatio-temporal rainfall dynamics still remains a challenge in those areas. Satellite-based rainfall products may help if their rainfall assessment are of high quality. In this paper, microwave-based integrated multi-satellite retrieval for the Global Precipitation Measurement (GPM) (IMERG) (MW-based IMERG) was assessed along with the random-forest-based rainfall (RF-based rainfall) and infrared-only IMERG (IR-only IMERG) products against the quality-controlled rain radar network and meteorological stations of high temporal resolution over the Pacific coast and the Andes of Ecuador. The rain area delineation and rain estimation of each product were evaluated at a spatial resolution of 11 km2 and at the time of MW overpass from IMERG. The regionally calibrated RF-based rainfall at 2 km2 and 30 min was also investigated. The validation results indicate different essential aspects: (i) the best performance is provided by MW-based IMERG in the region at the time of MW overpass; (ii) RF-based rainfall shows better accuracy rather than the IR-only IMERG rainfall product. This confirms that applying multispectral IR data in retrieval can improve the estimation of rainfall compared with single-spectrum IR retrieval algorithms. (iii) All of the products are prone to low-intensity false alarms. (iv) The downscaling of higher-resolution products leads to lower product performance, despite regional calibration. The results show that more caution is needed when developing new algorithms for satellite-based, high-spatiotemporal-resolution rainfall products. The radar data validation shows better performance than meteorological stations because gauge data cannot correctly represent spatial rainfall in complex topography under convective rainfall environments.

Estimating Rainfall with Multi-resource Data over East Asia Based on Machine Learning

The lack of accurate estimation of intense precipitation is a universal limitation in precipitation retrieval. Therefore, a new rainfall retrieval technique based on the Random Forest (RF) algorithm is presented using the Advanced Himawari Imager-8 (Himawari-8/AHI) infrared spectrum data and the NCEP operational Global Forecast System (GFS) forecast information. And the gauge-calibrated rainfall estimates from the Global Precipitation Measurement (GPM) product served as the ground truth to train the model. The two-step RF classification model was established for (1) rain area delineation and (2) precipitation grades’ estimation to improve the accuracy of moderate rain and heavy rain. In view of the imbalance categories’ distribution in the datasets, the resampling technique including the Random Under-sampling algorithm and Synthetic Minority Over-sampling Technique (SMOTE) was implemented throughout the whole training process to fully learn the characteristics among the samples. Among the features used, the contributions of meteorological variables to the trained models were generally greater than those of infrared information; in particular, the contribution of precipitable water was the largest, indicating the sufficient necessity of water vapor conditions in rainfall forecasting. The simulation results by the RF model were compared with the GPM product pixel-by-pixel. To prove the universality of the model, we used independent validation sets which are not used for training and two independent testing sets with different periods from the training set. In addition, the algorithm was validated against independent rain gauge data and compared with GFS model rainfall. Consequently, the RF model identified rainfall areas with a Probability Of Detection (POD) of around 0.77 and a False-Alarm Ratio (FAR) of around 0.23 for validation, as well as a POD of 0.60–0.70 and a FAR of around 0.30 for testing. To estimate precipitation grades, the value of classification was 0.70 in validation and in testing the accuracy was 0.60 despite a certain overestimation. In summary, the performance on the validation and test data indicated the great adaptability and superiority of the RF algorithm in rainfall retrieval in East Asia. To a certain extent, our study provides a meaningful range division and powerful guidance for quantitative precipitation estimation.

The MSG Technique: Improving Commercial Microwave Link Rainfall Intensity by Using Rain Area Detection from Meteosat Second Generation

Commercial microwave link (MWL) used by mobile telecom operators for data transmission can provide hydro-meteorologically valid rainfall estimates according to studies in the past decade. For the first time, this study investigated a new method, the MSG technique, that uses Meteosat Second Generation (MSG) satellite data to improve MWL rainfall estimates. The investigation, conducted during daytime, used MSG optical (VIS0.6) and near IR (NIR1.6) data to estimate rain areas along a 15 GHz, 9.88 km MWL for classifying the MWL signal into wet–dry periods and estimate the baseline level. Additionally, the MSG technique estimated a new parameter, wet path length, representing the length of the MWL that was wet during wet periods. Finally, MWL rainfall intensity estimates from this new MSG and conventional techniques were compared to rain gauge estimates. The results show that the MSG technique is robust and can estimate gauge comparable rainfall estimates. The evaluation scores every three hours of RMSD, relative bias, and r2 based on the entire evaluation period results of the MSG technique were 2.61 mm h−1, 0.47, and 0.81, compared to 2.09 mm h−1, 0.04, and 0.84 of the conventional technique, respectively. For convective rain events with high intensity spatially varying rainfall, the results show that the MSG technique may approximate the actual mean rainfall estimates better than the conventional technique.

Rain Area Detection in South-Western Kenya by Using Multispectral Satellite Data from Meteosat Second Generation

This study presents a rain area detection scheme that uses a gradient based adaptive technique for daytime and nighttime rain area detection and correction from reflectance and infrared (IR) brightness temperatures data of the Meteosat Second Generation (MSG) satellite. First, multiple parametric rain detection models developed from MSG’s reflectance and IR data were calibrated and validated with rainfall data from a dense network of rain gauge stations and investigated to determine the best model parameters. The models were based on a conceptual assumption that clouds characterised by the top properties, e.g., high optical thickness and effective radius, have high rain probabilities and intensities. Next, a gradient based adaptive correction technique that relies on rain area-specific parameters was developed to reduce the number and sizes of the detected rain areas. The daytime detection with optical (VIS0.6) and near IR (NIR1.6) reflectance data achieved the best detection skill. For nighttime, detection with thermal IR brightness temperature differences of IR3.9-IR10.8, IR3.9-WV73 and IR108-WV62 showed the best detection skill based on general categorical statistics. Compared to the Global Precipitation Measurement (GPM) Integrated Mult-isatellitE Retrievals for GPM (IMERG) and the gauge station data from the southwest of Kenya, the model showed good agreement in the spatial dynamics of the detected rain area and rain rate.

Topographic amplification of crustal subsidence by the rainwater load of the 2018 heavy rain in SW Japan

<p>The huge amount of water vapor in the atmosphere caused disastrous heavy rain and floods in early July 2018 in SW Japan. Here I present a comprehensive space geodetic study of water brought by this heavy rain done by using a dense network of Global Navigation Satellite System (GNSS) receivers. </p><p>First, I reconstruct sea level precipitable water vapor above land region on the heavy rain. The total amount of water vapor derived by spatially integrating precipitable water vapor on land was ~25.8 Gt, which corresponds to the bucket size to carry water from ocean to land. I then compiled the precipitation measured with a rain radar network. The data showed the total precipitation by this heavy rain as ~22.11 Gt.</p><p>Next, I studied the crustal subsidence caused by the rainwater as the surface load. The GNSS stations located under the heavy rain area temporarily subsided 1-2 centimeters and the subsidence mostly recovered in a day. Using such vertical crustal movement data, I estimated the distribution of surface water in SW Japan. </p><p>The total amount of the estimated water load on 6 July 2018 was ~68.2 Gt, which significantly exceeds the cumulative on-land rainfalls of the heavy rain day from radar rain gauge analyzed precipitation data. I consider that such an amplification of subsidence originates from the selective deployment of GNSS stations in the concave places, e.g. along valleys and within basins, in the mountainous Japanese Islands.</p>

Machine Learning Prediction of Precipitation in Metro Manila, Philippines

<p>It is difficult to predict the occurrence and rain volume of torrential rainfalls, such as guerrilla rain, rain band with typhoon and linear precipitation zone. As heavy rain area is spatially localized and the parent thunderstorm tends to develop within a short time, it makes difficult to accurately predict the occurrence location/time and rain volume. Recently, the machine learning technique is remarkably developed with the improved processing speed of computers and with a huge amount of the data. In addition to this, the application of the machine learning methods to the meteorological fields is intensively tried in the world. Since 2017, we started installing the automatic weather observation system (AWS) named as P-POTEKA in Metro Manila, the Philippines, which is one of the cities suffering from the torrential rainfall and related flood. So far, we installed 35-P-POTEKAs in Metro Manila and continue the weather observations (rain volume, temperature, air pressure, humidity, wind speed, wind direction and solar radiation) with the time resolution of 1 min. In this study, we used both P-POTEKA rain volume data and machine learning model (ConvLSTM: Convolutional Long-Short Term Memory) in order to predict the near future (< 1hour) rain volume and distribution. At the presentation, we will show the results derived from the machine learning prediction of the rain volume and distribution more in detail.</p>