Search

In this letter, we propose a novel approach to produce attenuation forecasts for microwave links using a probabilistic approach. It uses ensembles of forecast rainfall fields to easily derive attenuation forecasts for specific frequencies. The proposed approach uses the short-term ensemble prediction system (STEPS) to generate ensembles of high, spatial and temporal, resolution forecast rainfall fields based on observed weather radar fields with lead times of 15–90 min. Attenuation forecasts could eventually be used by telecommunication operators to drive the operation of wireless networks and ensure their maintenance during severe and extreme rainfall events. This study used 109 microwave links ranging from 15 to 40 GHz to verify the results of this probabilistic attenuation forecast. Results suggest that the STEPS-based attenuation forecasts were within the narrow span of the 90% confidence region for all microwave links tested up to 30 min lead time, decreasing for longer lead times. Examples of how the proposed approach can be used to derive a detailed probabilistic attenuation forecast for multiple lead times within a domain of few kilometers, as well as probability of attenuation maps for large areas are shown.

The biophysical processes occurring in the unsaturated zone have a direct impact on the water table dynamics. Representing these processes through the application of unsaturated zone models of different complexity has an impact on the estimates of the volumes of water flowing between the unsaturated zone and the aquifer. These fluxes, known as net recharge, are often used as the shared variable that couples unsaturated to groundwater models. However, as recharge estimates are always affected by a degree of uncertainty, model–data fusion methods, such as data assimilation, can be used to inform these coupled models and reduce uncertainty. This study assesses the effect of unsaturated zone models complexity (conceptual versus physically based) to update groundwater model outputs, through the assimilation of actual evapotranspiration rates, for a water-limited site in South Australia. Actual evapotranspiration rates are assimilated because they have been shown to be related to the water table dynamics and thus form the link between remote sensing data and the deeper parts of the soil profile. Results have been quantified using standard metrics, such as the root mean square error and Pearson correlation coefficient, and reinforced by calculating the continuous ranked probability score, which is specifically designed to determine a more representative error in stochastic models. It has been found that, once properly calibrated to reproduce the actual evapotranspiration–water table dynamics, a simple conceptual model may be sufficient for this purpose; thus using one configuration over the other should be motivated by the specific purpose of the simulation and the information available.

Modelling the water transport along the soil–plant–atmosphere continuum is fundamental to estimating and predicting transpiration fluxes. A Finite-difference Ecosystem-scale Tree Crown Hydrodynamics model (FETCH3) for the water fluxes across the soil–plant–atmosphere continuum is presented here. The model combines the water transport pathways into one vertical dimension, and assumes that the water flow through the soil, roots, and above-ground xylem can be approximated as flow in porous media. This results in a system of three partial differential equations, resembling the Richardson–Richards equation, describing the transport of water through the plant system and with additional terms representing sinks and sources for the transfer of water from the soil to the roots and from the leaves to the atmosphere. The numerical scheme, developed in Python 3, was tested against exact analytical solutions for steady state and transient conditions using simplified but realistic model parameterizations. The model was also used to simulate a previously published case study, where observed transpiration rates were available, to evaluate model performance. With the same model setup as the published case study, FETCH3 results were in agreement with observations. Through a rigorous coupling of soil, root xylem, and stem xylem, FETCH3 can account for variable water capacitance, while conserving mass and the continuity of the water potential between these three layers. FETCH3 provides a ready-to-use open access numerical model for the simulation of water fluxes across the soil–plant–atmosphere continuum.

Modelling the water transport along the soil-plant-atmosphere continuum is fundamental to estimating and predicting transpiration fluxes. A tree-hydrodynamic model (SPAC-3Hpy) for the water fluxes across the soil-plant-atmosphere continuum is presented here. The model combines the water transport pathways to one vertical dimension, and assumes that the water flow through the soil, roots, and above-ground xylem can be approximated as a flow in porous media. This results in a system of three partial differential equations resembling the Richardson-Richards equation describing the transport of water through the plant system and with additional terms representing sinks and sources for the transfer of water from to the soil to the roots and from the leaves to the atmosphere. The numerical scheme, developed in Python 3, was tested against exact analytical solutions for steady state and transient conditions using simplified but realistic model parametrizations. The model was also used to simulate a previously published case study where observed transpiration rates were available in order to evaluate model performance. With the same model setup as the published case study, SPAC-3Hpy results were in agreement with observations. Through a rigorous coupling of soil, roots, and hydroactive xylem, SPAC-3Hpy can account for variable capacitance while conserving mass and the continuity of the water potential between these three layers. SPAC-3Hpy provides a ready-to-use open access numerical model for the simulation of water fluxes across the soil-plant-atmosphere continuum.

A reanalysis is a physically consistent set of optimally merged simulated model states and historical observational data, using data assimilation. High computational costs for modeled processes and assimilation algorithms has led to Earth system specific reanalysis products for the atmosphere, the ocean and the land separately. Recent developments include the advanced uncertainty quantification and the generation of biogeochemical reanalysis for land and ocean. Here, we review atmospheric and oceanic reanalyzes, and more in detail biogeochemical ocean and terrestrial reanalyzes. In particular, we identify land surface, hydrologic and carbon cycle reanalyzes which are nowadays produced in targeted projects for very specific purposes. Although a future joint reanalysis of land surface, hydrologic, and carbon processes represents an analysis of important ecosystem variables, biotic ecosystem variables are assimilated only to a very limited extent. Continuous data sets of ecosystem variables are needed to explore biotic‐abiotic interactions and the response of ecosystems to global change. Based on the review of existing achievements, we identify five major steps required to develop terrestrial ecosystem reanalysis to deliver continuous data streams on ecosystem dynamics., Plain Language Summary: A reanalysis is a unique set of continuous variables produced by optimally merging a numerical model and observed data. The data are merged with the model using available uncertainty estimates to generate the best possible estimate of the target variables. The framework for generating a reanalysis consists of the model, the data, and the model‐data‐fusion algorithm. The very specific requirements of reanalysis frameworks have led to the development of Earth‐compartment specific reanalysis for the atmosphere, the ocean and land. Here, we review atmospheric and oceanic reanalyzes, and in more detail biogeochemical ocean and terrestrial reanalyzes. In particular, we identify land surface, hydrologic, and carbon cycle reanalyzes which are nowadays produced in targeted projects for very specific purposes. Based on a review of existing achievements, we identify five major steps required to develop reanalysis for terrestrial ecosystem to shed more light on biotic and abiotic interactions. In the future, terrestrial ecosystem reanalysis will deliver continuous data streams on the state and the development of terrestrial ecosystems., Key Points: Reanalyzes provide decades‐long model‐data‐driven harmonized and continuous data sets for new scientific discoveries. Novel global scale reanalyzes quantify the biogeochemical ocean cycle, terrestrial carbon cycle, land surface, and hydrologic processes. New observation technology and modeling capabilities allow in the near future production of advanced terrestrial ecosystem reanalysis., European Union's Horizon 2020 research and innovation programme, Deutsche Forschungsgemeinschaft, U.S. Department of Energy, Emory University's Halle Institute for Global Research and the Halle Foundation Collaborative Research, NSF, NASA, Natural Environment Research Council, European Union'’s Horizon 2020 research and innovation programme, NSERC Discovery program, the Ocean Frontier Institute, and MEOPAR, Research Foundation Flanders (FWO), Helmholtz Association, NASA Terrestrial Ecosystems

Accurate river bathymetry is required for applications including hydrodynamic flow modelling and understanding morphological processes. Bathymetric measurements are typically a set of depths at discrete points that must be reconstructed into a continuous surface. A number of algorithms exist for this reconstruction, including spline-based techniques and kriging methods. A novel and efficient method is introduced to produce a co-ordinate system fitted to the river path suitable for bathymetric reconstructions. The method is based on numerical conformal mapping and can handle topological features such as islands and branches in the river. Bathymetric surfaces generated using interpolation over a conformal map are compared to spline-based and kriging methods on a section of the Balonne River, Australia. The results show that the conformal mapping algorithm produces reconstructions comparable in quality to existing methods, preserves flow-wise features and is relatively insensitive to the number of sample points, enabling faster data collection in the field.

Among all remote sensing missions, the Gravity Recovery and Climate Experiment (GRACE) was unique as it measured the change in total water content across all terrestrial water storages (TWS) including subsurface, deep soil moisture, and groundwater. However, its coarse resolution is a major challenge for practical applications. Ensemble Kalman filters (EnKFs) are useful tools to combine observations with models to reduce prediction errors. But due to the coarse resolution of the GRACE products, the EnKF does not work well in its usual form. Accordingly, different EnKF structures have been proposed and employed but a comparison between them has not yet been attempted. Here we assessed these structures using a synthetic problem. Alternative structures were formed using different increment calculation and updating strategies, observation operators, and the types of observation fed to the filter. It was found that all available structures led to an improvement in model performance when measured against a synthetic reference. However, the degree of improvement was strongly dependent on the assimilation strategy. Assimilating absolute TWS values (the summation of the TWS anomalies and an unbiased baseline) gave the best model performance when combined with an increment calculation strategy in which the increments are calculated and applied to all days of the month. However, without an unbiased baseline, assimilating TWS changes still leads to an acceptable improvement in model performance. Among the observation operators, those that predict the observations as an average of multiple days had the best performance.

Accurately simulating floodplain inundation is absolutely vital to minimize damage to life and property. The giant strides made in advanced computing, now allow running increasingly complex models at reasonable resolutions over large areas, with the promise of further improvement in the near future. However, the uncertainty contributed by input, boundary, and forcing data, often leads to highly erroneous predictions. As spatially distributed Earth Observations of flood extent and water level become increasingly available, they pave the way for further constraining, and hence, improving the accuracy of hydraulic flood forecasting models. Effectively using these datasets requires an in-depth understanding of the impacts that the resolution, accuracy, location, timing, and frequency of acquisition of observations may have on model-data integration efforts. This chapter presents a review of the current capabilities in the field of flood data assimilation. The challenges and opportunities of using Earth Observation data for operational flood inundation forecasting are also discussed.

The monitoring of wildfire smoke is important to help mitigate impacts on people such as by sending early warnings to affected areas. Received signal levels (RSLs) from radio links have been used as an opportunistic way to accurately measure rainfall and humidity. Radio links provide integrated measurements along their paths and are an exceptional untapped resource to complement air quality stations in areas affected by smoke events, or in developing countries without air quality monitoring capability. This study analyzed radio link signal fluctuations during smoke events associated with the 2019–2020 Australian wildfires. Concurrently, the atmospheric boundary layer was characterized using atmospheric soundings and surface observations, as well as air quality proxies such as particulate matter concentrations less than 2.5 μm (10 μm), or PM2.5 (PM10). Observations showed that dry air containing large amounts of smoke within a surface layer above the ground acted as a lid, reducing dispersion, trapping and maintaining high ground-level concentrations of smoke. These conditions also created anomalous propagation conditions for radio links and operational weather radars. Power-law relations between signal fluctuations and PM10 and PM2.5 were derived based on the link data collected and the closest air quality station observations. While there was variability in retrieval performance across smoke events, the best performance determination coefficients exceeded 0.5, with an RMSE on the order of less than 50 μg m−3 for concentrations of more than 400 μg m−3. Mid-range link lengths (5–20 km) provided the best results.

Australia is the driest of all inhabited continents, and is prone to both droughts and severe floods. In order to mitigate the impact of these phenomena, accurate flood forecasts and a reliable estimate of the land surface water storage are needed. This chapter provides an overview of three studies developing innovative methods to merge satellite data with models to achieve these goals. The overall conclusion from this work is that satellite data provide extremely valuable information for an improved water management, and that innovative methods are required to fully utilize their potential.

This study presents the first evaluation of using commercial microwave link (CML) data for rainfall measurements in Australia, with the test site being the greater Melbourne Metropolitan area. More than 100 CMLs with microwave frequency ranging between 10 and 40 GHz have been used for the rainfall retrieval. The 15-minute received signal levels (RSLs) for each CML based on two sampling strategies (average and minimum/maximum) collected for 2 years provided a unique dataset to compare performances of rainfall retrievals. The open source algorithm RAINLINK was used for deriving rainfall from the 15-minute RSL data. From two years of data, a subset of 30 rainy days distributed across this period were used for calibrating the RAINLINK parameters, with the remaining data used for validation. For this study, only path-averaged rainfall intensities were validated based on a gauge-adjusted radar product serving as the reference. The result of the wet-dry classification showed that the minimum and maximum RSL data performed better, with lower probability of false detection and higher Matthews correlation coefficient than average RSL data. For the rainfall retrieval, both datasets showed similar correlation with the gauge adjusted radar product. However, based on other statistics (RMSE, bias and CV) minimum and maximum RSL data outperformed average for the rainfall retrieval. Overall, this study highlights the robust accuracy of commercial microwave links for rainfall retrieval while using only minimum and maximum RSL data.

This study presents an assessment of a pre-operational soil moisture product at 1 km resolution derived from satellite data acquired by the European Radar Observatory Sentinel-1 (S-1), representing the first space component of the Copernicus program. The product consists of an estimate of surface soil volumetric water content (SSM) [m/m] and its uncertainty [m/m], both at 1 km. The retrieval algorithm relies on a time series based Short Term Change Detection (STCD) approach, taking advantage of the frequent revisit of the S-1 constellation that performs C-band Synthetic Aperture Radar (SAR) imaging. The performance of the S-1 SSM product is estimated through a direct comparison between 1068 S-1 SSM images against in situ SSM measurements acquired by 167 ground stations located in Europe, America and Australia, over 4 years between January 2015 and December 2020, depending on the site. The paper develops a method to estimate the spatial representativeness error (SRE) that arises from the mismatch between the S-1 SSM retrieved at 1 km resolution and the in situ point-scale SSM observations. The impact of SRE on standard validation metrics, i.e., root mean square error (RMSE), Pearson correlation (R) and linear regression, is quantified and experimentally assessed using S-1 and ground SSM data collected over a dense hydrologic network (4 - 5 stations/km) located in the Apulian Tavoliere (Southern Italy). Results show that for the dense hydrological network the RMSE and correlation are ~0.06 m/m and 0.71, respectively, whereas for the sparse hydrological networks, i.e., 1 station/km, the SRE increases the RMSE by ~0.02 m/m (70% Confidence Level). Globally, the S-1 SSM product is characterized by an intrinsic (i.e., with SRE removed) RMSE of ~0.07 m/m over the SSM range [0.03, 0.60] m/m and R of 0.54. A breakdown of the RMSE per dry, medium and wet SSM ranges is also derived and its implications for setting realistic requirements for SAR-based SSM retrieval are discussed together with recommendations for the density of in situ SSM observations.

The diversity in hydrologic models has historically led to great controversy on the “correct” approach to process-based hydrologic modeling, with debates centered on the adequacy of process parameterizations, data limitations and uncertainty, and computational constraints on model analysis. In this paper we revisit key modeling challenges, outlined by Freeze and Harlan nearly 50 years ago, on requirements to (1) define suitable model equations, (2) define adequate model parameters, and (3) cope with limitations in computing power. We outline the historical modeling challenges, summarize modeling advances that address these challenges, and define outstanding research needs. We illustrate how modeling advances have been made by groups using models of different type and complexity, and we argue for the need to more effectively use our diversity of modeling approaches in order to advance our collective quest for physically realistic hydrologic models.

The bio-physical processes occurring in the unsaturated zone have a direct impact on the water table dynamics. Conceptual models, with a simplified representation of the unsaturated zone dynamics, are often selected for coupling to groundwater models, while physically-based models are widely used, particularly at the field scale, for an accurate representation of the water transport. The recharge rates estimated by these Unsaturated Zone Models (UZMs) can then be used as input for groundwater models. Because recharge estimates are always affected by uncertainty, model-data fusion methods, such as data assimilation, can be used to reduce the uncertainty in the model results. In this study, the required complexity (i.e. conceptual versus physically-based) of the unsaturated zone model to update groundwater models through the assimilation of evapotranspiration (ET) rates is assessed for a water-limited site in South Australia. ET rates are assimilated because they have been shown to be related to the groundwater table dynamics, and thus form the link between remote sensing data and the deeper parts of the soil profile. It has been found that, under the test site conditions, a conceptual UZM can be used to improve groundwater model results through the assimilation of ET rates.

Floods can have extremely expensive and often fatal consequences, thereby necessitating accurate flood inundation forecasts for increased preparedness and resilience. In an operational forecasting scenario, inflow uncertainties propagated from precipitation forecasts dominate and lead to inherently erroneous predictions of flood inundation, frequently impeding their application to rescue and response operations. Thus, it is expected that assimilating independent inundation observations, from Synthetic Aperture Radar (SAR) sensors for example, may reduce the inherent uncertainty in hydraulic modelling. The increasing number of SAR satellites, with their all-weather/all-day imaging capabilities, have increased the probability of monitoring flood dynamics from space. SAR-based flood extents were previously used to indirectly retrieve floodplain water levels in conjunction with digital elevation models. However, studies highlighted this process as an additional source of uncertainty, leading to the development of algorithms for the direct assimilation of flood extent into hydraulic flood inundation forecasting chains. The efficiency of flood extent assimilation is keenly sensitive to the spatiotemporal observation characteristics, and so the expected improvement in the forecast strongly depends on the acquisition timing with respect to the position of the flood wave. In this study, numerical experiments were used to simulate multiple spatiotemporal SAR acquisition scenarios, to identify the optimum measurement design for targeted satellite acquisition, to best facilitate flood extent assimilation. A particle filter based flood extent assimilation framework was developed using the hydraulic model LISFLOOD-FP, and implemented for the 2011 flood event in the Clarence Catchment, Australia. An operational forecasting scenario was emulated for the open loop model ensemble, with the consideration of temporally correlated, variance changing uncertainties in inflows, simulating hydrological model forecasts. The impact of assimilating flood extent at reaches exhibiting uniform flow behaviour, with different combinations of first visit and revisit intervals were investigated. Results indicate that the optimum timing and frequency of targeted SAR acquisitions differs with respect to reach hydraulic characteristics and that images acquired after the peak is observed in the channel are most informative for the forecast. Note that the maximum inundation extent in the floodplain always follows the channel peak, and therefore, post-peak images with respect to the within reach flood wave could improve predictions during the peak in the floodplain. Moreover, a single image assimilated at a reaches exhibiting more diffusive flow behaviour just after the peak, could result in improvements comparable to the assimilation of multiple images elsewhere. Findings from the study will allow the optimal utilization of SAR imagery to overcome localized model uncertainties, and help to maximize the accuracy of inundation forecasts. Keywords: Flood inundation modelling, flood extent assimilation, SAR, data assimilation, hydraulic modelling, forecast uncertainty

Flooding is one of the most common and costly natural hazards at global scale. Flood models are important in supporting flood management. This is a computationally expensive process, due to the high nonlinearity of the equations involved and the complexity of the surface topography. New modelling approaches based on deep learning algorithms have recently emerged for multiple applications.This study aims to investigate the capacity of machine learning to achieve spatio-temporal flood modelling. The combination of spatial and temporal input data to obtain dynamic results of water levels and flows from a machine learning model on multiple domains for applications in flood risk assessments has not been achieved yet. Here, we develop increasingly complex architectures aimed at interpreting the raw input data of precipitation and terrain to generate essential spatio-temporal variables (water level and velocity fields) and derived products (flood maps) by training these based on hydrodynamic simulations.An extensive training dataset is generated by solving the 2D shallow water equations on simplified topographies using Lisflood-FP.As a first task, the machine learning model is trained to reproduce the maximum water depth, using as inputs the precipitation time series and the topographic grid. The models combine the spatial and temporal information through a combination of 1D and 2D convolutional layers, pooling, merging and upscaling. Multiple variations of this generic architecture are trained to determine the best one(s). Overall, the trained models return good results regarding performance indices (mean squared error, mean absolute error and classification accuracy) but fail at predicting the maximum water depths with sufficient precision for practical applications.A major limitation of this approach is the availability of training examples. As a second task, models will be trained to bring the state of the system (spatially distributed water depth and velocity) from one time step to the next, based on the same inputs as previously, generating the full solution equivalent to that of a hydrodynamic solver. The training database becomes much larger as each pair of consecutive time steps constitutes one training example.Assuming that a reliable model can be built and trained, such methodology could be applied to build models that are faster and less computationally demanding than hydrodynamic models. Indeed, in with the synthetic cases shown here, the simulation times of the machine learning models (< seconds) are far shorter than those of the hydrodynamic model (a few minutes at least). These data-driven models could be used for interpolation and forecasting. The potential for extrapolation beyond the range of training datasets will also be investigated (different topography and high intensity precipitation events).

The fourth and fifth Soil Moisture Active Passive Experiments (SMAPEx-4 and -5) were conducted at the beginning of the SMAP operational phase, May and September 2015, to: 1) evaluate the SMAP microwave observations and derived soil moisture (SM) products and 2) intercompare with the Soil Moisture and Ocean Salinity (SMOS) and Aquarius missions over the Murrumbidgee River Catchment in the southeast of Australia. Airborne radar and radiometer observations at the same microwave frequencies as SMAP were collected over SMAP footprints/grids concurrent with its overpass. In addition, intensive ground sampling of SM, vegetation water content, and surface roughness was carried out, primarily for validation of airborne SM retrieval over six $\sim 3\,\,\text {km} \times 3$ km focus areas. In this study, the SMAPEx-4 and -5 data sets were used as independent reference for extensively evaluating the brightness temperature and SM products of SMAP, and intercompared with SMOS and Aquarius under a wide range of SM and vegetation conditions. Importantly, this is the only extensive airborne field campaign that collected data while the SMAP radar was still operational. The SMAP radar, radiometer, and derived SM showed a high agreement with the SMAPEx-4 and -5 data set, with a root-mean-squared error (RMSE) of ~3 K for radiometer brightness temperature, and an RMSE of $\sim 0.05~\text{m}^{3}/\text{m}^{3}$ for the radiometer-only SM product. The SMAP radar backscatter had an RMSE of 3.4 dB, while the retrieved SM had an RMSE of 0.11 m3/m3 when compared with the SMAPEx-4 data set.

For an accurate estimation of land surface state variables through remote sensing data assimilation, it is important to estimate the forecast and observation biases as well. This study focuses on the evaluation of a methodology to estimate land surface state variables, together with model forecast and observation biases. Two conceptual rainfall-runoff models (HBV and GRKAL) are used for this purpose. Soil moisture data, retrieved by the Soil Moisture Ocean Salinity (SMOS) mission, are assimilated into these models for 59 unregulated sub-basins of the Murray-Darling basin in Australia. When both models simulate similar soil moisture values, the methodology results in similar forecast and observation bias estimates for both models. The same behavior is obtained when the temporal evolution of the soil moisture simulations is different, but with a similar long-term mean climatology. However, when the long-term mean climatology of both models is different, but with a similar temporal evolution, the bias estimates from both models have a different climatology as well, but with a high temporal correlation. The overall conclusion from this paper is that observation bias estimation is of key importance when updating internal state variables in a conceptual rainfall-runoff system that is calibrated to produce realistic discharge output for possibly biased internal state variables, and that the relative partitioning of bias into forecast and observation bias remains a model-dependent challenge.

Synthetic Aperture Radar (SAR) data are currently the most reliable resource for flood monitoring, though still subject to various uncertainties, which can be objectively represented with probabilistic flood maps. Moreover, the growing number of SAR satellites has increased the likelihood of observing a flood event from space through at least a single SAR image, but generalized methods for flood classification independent of sensor characteristics need to be developed, to fully utilize these images for disaster management. Consequently, a neuro-fuzzy flood mapping technique is proposed for texture-enhanced single SAR images. Accordingly, any SAR image is first processed to generate second-order statistical textures, which are subsequently optimized using a dimensionality reduction technique. The flood and non-flood classes are then modelled within a fuzzy inference system using Gaussian curves. Parameterization is achieved by training a neural network on the image through user-defined polygons. The results of the optimized texture-based neuro-fuzzy classification were compared against the performance of the SAR image alone and that of SAR enhanced with randomly selected texture features. This approach was tested for a COSMO-SkyMed SAR image at two validation sites, for which high resolution aerial photographs were available. An overall accuracy assessment using reliability diagrams demonstrated a reduction of 54.2% in the Weighted Root Mean Squared Error (WRMSE) values compared to the stand-alone use of SAR. WRMSE values estimated for the proposed method varied from 0.027 to 0.196. A fuzzy validation exercise was also proposed to account for the uncertainty in manual flood identification from aerial photography, resulting in fuzzy spatial similarity values ranging from 0.67 to 0.92, with higher values representing better performance. Results suggest that the proposed approach has demonstrated potential to improve operational SAR-based flood mapping.

Conventional neuro-fuzzy systems used for rainfall-runoff (R-R) modelling are generally dependent on expert knowledge. In these models, not only the structure is designed by the expert user, but also all the required knowledge for fuzzy partitioning of the input–output space and rule base need to be provided by the expert. To move towards NFS with a more flexible rule base and structure, efforts are made to integrate the self-reliant mechanisms into the learning algorithm that enable the model to identify the position and distribution of fuzzy labels in input–output space and generate the required rule base. In this study, the self-adaptive fuzzy inference network (SaFIN) is used for the R-R application. SaFIN employs a new clustering technique known as Categorical Learning-Induced Partitioning (CLIP) which allows the model to adapt to the new incoming tuple by consistently updating the model. SaFIN is also equipped with a rule-pruning mechanism that can exclude inconsistent and obsolete rules. In this study, SaFIN R-R models are developed in three different catchment types and sizes where the results are compared against a benchmark NFS model and few physical models including URHM, HBV, GR4J. Results shows that SaFIN is a capable and robust tool for R-R modeling under varying catchment conditions. Moreover, SaFIN produced comparable if not superior results to the benchmark models. It was concluded that SsFIN with its self-reliant learning and rule generation mechanism equipped with rule pruning can make it a competent tool for R-R modelling in catchments where the data may contain some inconsistencies.