Rain Rate Prediction Research Articles

Deep Spatial Interpolation of Rain Field for U.K. Satellite Networks

This article presents two new state-of-the-art spatial rain field interpolation convolutional neural networks (SRFICNNs), referred to as learned deviation (LD) and learned interpolation (LI) models, for predicting the point rain rate at finer spatial scales. The main contribution is the successful introduction of the prior-art deep learning technique into high-resolution (HR) rainfall rate prediction with significant improvement in accuracy. This is very important for the effective implementation of fade mitigation techniques for both terrestrial and satellite networks. The comparison of the models’ performances with ground truth (radar measurements) shows that the proposed models give an excellent mean square error (MSE) and structural SIMilarity (SSIM) in rainfall field reconstruction if the network depth falls in the range of 15–25 weight layers. The final model uses 20 layers for HR point rain rate prediction. Further study shows that the LD model offers a faster convergence and yields a more accurate rain rate prediction. In particular, this article compares the rain rate exceedance distribution and Log-Normality property from the model estimates with values calculated from measured data. Results show that the LD model gives a highly accurate estimate of these two indices with the corresponding root mean square (rms) error of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$5.1709\times 10^{-4}$ </tex-math></inline-formula> and 0.0013, respectively.

Performance evaluation of some rain rate conversion models for microwave propagation studies

An important characteristic of rainfall levels at a particular place is the statistical distribution of rainfall rate. In this paper, 5-min integration time rainfall data for the Northcentral region of Nigeria was obtained from the Tropospheric Data Acquisition Network (TRODAN), Anyigba, Nigeria. Also, 1-min integration time rainfall was measured at Minna, Nigeria. In order to obtain the optimal rain rate model suitable for this region, two globally recognised rain rate models were critically evaluated and compared with the 1-min measurements. These are the ITU-R P.837-7 and Lavergnat-Gole (L-G) models. The results obtained showed that the ITU-R P.837-7 and L-G models respectively underestimated the measured rain rate by 7.3 mm/h and 9 mm/h at time percentage exceedance of 0.1%, while they underestimated the measured rain rate by 23.4 mm/h and 13 mm/h respectively at 0.01%. At 0.001%, the measured rain rate was overestimated by the ITU-R P.837-7 and L-G models by 27.4 mm/h and 3 mm/h respectively. Further performance evaluation of the predefined models was carried out using different error metrics such as sum of absolute error (SAE), mean absolute error (MAE), root mean square error (RMSE), standard deviation (STDEV) and Spearman’s rank correlation. The results obtained adjudged the Lavergnat-Gole model as the best rain rate prediction model for this region.

Modelling of Alpha and Beta for Rain Rate Prediction for Radio Propagation Systems

The effect of rain in the design of satellite and terrestrial microwave radio links is of interest to Engineers and Scientists. It is good to have a reliable design that guarantees high level of accuracy of the rain rate distribution from the lowest rain rate value to the highest. The present work proposes a model that expresses rain rate as a function of alpha and beta obtained at 0.01% of time. When tested, the results obtained with the measurement perform well for the stations considered at a rain rated between 5mm/h to 200mm/h. Thus, , the empirical models that were obtained through them could be a useful tool for the radio design engineers for high rain rate areas.

A Study on Rain Rate Prediction of Southwestern Nigeria

Rainfall parameters can be utilized to investigate the effect of climate change through scientific methods. However, data on rainfall rate exceeded for a fraction of an average year is grossly unavailable over Nigeria’s climate, thereby diminishing the capability of existing models to adequately estimate the effect of degradation due to rain. Hence, more accurate estimation is required for better predictions. Rainfall volume data for six different locations in the south-western region of Nigeria were obtained for rain rate computation using Semire and Rosmiwati model. The curve-fitted Cumulative Distribution Functions were compared with the ITU-R rain rate model (Recommendation P.837-6) and compensation function was obtained using error analysis while the performance was evaluated with respect to existing models using Chi-square, and Percentage Error and Root Mean Square Error (RMSE) metrics. The outcome of this study can be adopted for better understanding of spatial rainfall intensity in this region and other climatic zones of similar rainfall characteristics.

Comparison of 1-min rain rate derived from TRMM satellite data and raingauge data for microwave applications in Nigeria

A comparative statistical study of derived 1-min rain rate using data obtained from Tropical Rainfall Measuring Mission (TRMM) satellite (3A12 V6 and 3B43V6) and raingauge measurements are presented. 30-years data of rainfall intensities obtained from the Nigerian Meteorological Agency rain-gauge network and 9-years rainfall data obtained from TRMM satellite are used. Cumulative distributions of rain rate were obtained using a combination of an improved version of the Moupfouma–Martin model for rain rate prediction in the tropical regions. Variations of rain rate of different integration time are compared for 28 stations across Nigeria. The Kriging interpolation method was used for spatial interpolation of rain accumulation values into a regular grid in order to obtain a highly consistent and predictable inter-gauge rainfall variation. Noticeable differences between rain rate estimated from rain-gauge and TRMM data were found to be a function of period and location. From quantitative comparison over each of the region, there are reasonably positive correlations between the two sets of data for different percentages of time with correlation coefficient varying from 0.84 to 0.99 at 99.99% availability of time. Comparison with ITU-R 837-5 shows that, ITU model underestimates rain rate as high as 51% from the mean value depending on the percentages of at which time rain rate is exceeded. The overall result shows that the gauge data perform better than TRMM data in most of the sites located in the SW, SE, SS, MB and NE regions, while TRMM data provided better results for sites in the NW region from the statistical perspective with the lower RMS values.

One-Minute Rain-Rate Contour Maps for Microwave-Communication-System Planning in a Tropical Country: Nigeria

Rain attenuation is the dominant propagation impairment for high-frequency microwave systems. System design therefore requires the knowledge of the one-minute integrated rain-rate cumulative distribution function over the coverage area in order to determine the appropriate transmitter and receiver characteristics. This paper presents a set of rain-rate contour maps, useful for link and network power sizing in Nigeria. They were developed using a combination of first-order rain-rate statistical methods: the Moupfouma-Martin model for rain-rate prediction in tropical regions, and the Rice-Holmberg model.

Performances of existing rain rate models in equatorial region

The performance of the existing rain rate models in equatorial zones is still a debated issue due to the lack of measurements reported from these areas of the world to develop and validate prediction models. Therefore point rain rates measured over three years by a network of rain gauges located in Malaysia are presented. A comparison between rain rate data set with a sampling period of 1‐m and existing rain rate prediction models is presented.

Evaluation of global and CCIR models for estimation of rain rate statistics

The prediction of the statistics of attenuation by rain on a radio wave propagation path requires first the prediction of rain rate statistics for a point on the path and then the conditional prediction of the path‐averaged attenuation, given the rain rate statistics. In this paper, two models with worldwide applicability, the Global and CCIR models for effecting the first step of the prediction process, are evaluated by comparing model predictions with rain rate measurements. The use of single‐year, measured rain rate statistics as a model for the prediction of the statistics for subsequent years is also evaluated. Surface rain rate distributions for one or more years of observatins at 52 locations were employed for the model evaluations. On the basis of the available data the Global model performed best. The location‐to‐location variations in rain rate at specified probability levels and in probability at specified rain rate levels were found to be well described by lognormal distributions. No significant differences were found between year‐to‐year and location‐to‐location variability estimates relative to the two rain rate prediction models. The data show that the climate models are equivalent to inference from limited‐duration records and suggest that prediction variability can be reduced by subdivision of the rain climate zones into smaller regions.