Efficient, robust, and accurate early flood warning is a pivotal decision support tool that can help save lives and protect the infrastructure in natural disasters. This research builds a hybrid deep learning (ConvLSTM) algorithm integrating the predictive merits of Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) Network to design and evaluate a flood forecasting model to forecast the future occurrence of flood events. Derived from precipitation dataset, the work adopts a Flood Index (), in form of a mathematical representation, to capture the gradual depletion of water resources over time, employed in a flood monitoring system to determine the duration, severity, and intensity of any flood situation. The newly designed predictive model utilizes statistically significant lagged , improved by antecedent and real-time rainfall data to forecast the next daily value. The performance of the proposed ConvLSTM model is validated against 9 different rainfall datasets in flood prone regions in Fiji which faces flood-driven devastations almost annually. The results illustrate the superiority of ConvLSTM-based flood model over the benchmark methods, all of which were tested at the 1-day, 3-day, 7-day, and the 14-day forecast horizon. For instance, the Root Mean Squared Error (RMSE) for the study sites were 0.101, 0.150, 0.211 and 0.279 for the four forecasted periods, respectively, using ConvLSTM model. For the next best model, the RMSE values were 0.105, 0.154, 0.213 and 0.282 in that same order for the four forecast horizons. In terms of the difference in model performance for individual stations, the Legate-McCabe Efficiency Index (LME) were 0.939, 0.898, 0.832 and 0.726 for the four forecast horizons, respectively. The results demonstrated practical utility of ConvLSTM in accurately forecasting and its potential use in disaster management and risk mitigation in the current phase of extreme weather events.

Both fluvial and pluvial floods are a common occurrence in Fiji, with the fluvial floods causing significant economic consequences for this island nation. To investigate flood risk on a daily basis, the Flood Index (𝐼 𝐹 ) is developed in this study, based on the rationale that the onset and the severity of a flood event on any given day is based on the current and the antecedent day's precipitations. The mathematical methodology considers the notion that the impact of daily cumulative precipitation on a particular flood event arising from a previous day's precipitation, which decreases gradually over time due to the interaction of hydrological factors (e.g., evaporation, percolation, seepage, surface run-off, drainage, etc.,). These are accounted for, mathematically, by a time-reduction weighted precipitation influencing the magnitude of 𝐼 𝐹 with the gradual passage of time. Considering the duration, severity and intensity of all identified flood events, the applicability of daily 𝐼 𝐹 is tested at 9 study sites in Fiji using a 30year precipitation dataset (1990 to 2019) obtained from Fiji Meteorological Services. The newly developed 𝐼 𝐹 is also applied at several flood prone sites, with results demonstrating flood events were common throughout the country, mostly notable between the months of November to April (the wet season). Upon examining changes in daily 𝐼 𝐹 , the subsequent flood properties were determined, showing that most severe events generally start in January. The flood event with the highest severity was recorded in Lautoka (𝐼 𝐹 𝑎𝑐𝑐 (Flood Severity) ≈ 149.14, 𝐼 𝐹 𝑚𝑎𝑥 (Peak Danger) ≈ 3.39, 𝐷 𝐹 (Duration of Flood) ≈ 151 𝑑𝑎𝑦𝑠, 𝑡 𝑜𝑛𝑠𝑒𝑡 (Onset Date) = 23𝑟𝑑 𝐽𝑎𝑛𝑢𝑎𝑟𝑦 2012 ) , followed by Savusavu (𝐼 𝐹 𝑎𝑐𝑐 ≈ 141.65, 𝐼 𝐹 𝑚𝑎𝑥 ≈ 1.75, 𝐷 𝐹 ≈ 195 𝑑𝑎𝑦𝑠, 𝑡 𝑜𝑛𝑠𝑒𝑡 = 27𝑡ℎ 𝑁𝑜𝑣𝑒𝑚𝑏𝑒𝑟 1999) and Ba (𝐼 𝐹 𝑎𝑐𝑐 ≈ 131.57, 𝐼 𝐹 𝑚𝑎𝑥 ≈ 3.13, 𝐷 𝐹 ≈ 113 𝑑𝑎𝑦𝑠, 𝑡 𝑜𝑛𝑠𝑒𝑡 = 9𝑡ℎ 𝐽𝑎𝑛𝑢𝑎𝑟𝑦 2009). These results clearly illustrate the practicality of daily 𝐼 𝐹 in determining the duration, severity, and the intensity of flood situations in Fiji, aswell as its potential application to small island nations. The use of daily 𝐼 𝐹 to quantify the flood events can therefore enable a cost-effective and innovative solution to study historical floods in developing and first world countries with methodology being particularly useful to their governments, private organizations, non-governmental organizations and individual communities to help develop more community-amicable policy and strategic plans for flood impact preparation and its risk mitigation.