A deep learning model for reconstructing centenary water storage changes in the Yangtze River Basin.

Since 2002, the Gravity Recovery and Climate Experiment (GRACE) and its Follow-On mission (GRACE-FO) have facilitated highly accurate observations of changes in total water storage anomalies (TWSA). However, limited observations of TWSA derived from GRACE in the Yangtze River Basin (YRB) have hindered our understanding of its long-term variability. In this paper, we present a deep learning model called RecNet to reconstruct the climate-driven TWSA in the YRB from 1923 to 2022. The RecNet model is trained on precipitation, temperature, and GRACE observations with a weighted mean square error (WMSE) loss function. The performance of the RecNet model is validated and compared against GRACE data, water budget estimates, hydrological models, drought indices, and existing reconstruction datasets. The results indicate that the RecNet model can successfully reconstruct historical water storage changes, surpassing the performance of previous studies. In addition, the reconstructed datasets are utilized to assess the frequency of extreme hydrological conditions and their teleconnections with major climate patterns, including the El Niño-Southern Oscillation, Indian Ocean Dipole, Pacific Decadal Oscillation, and North Atlantic Oscillation. Independent component analysis is employed to investigate individual climate patterns' unique or combined influence on TWSA. We show that the YRB exhibits a notable vulnerability to extreme events, characterized by a recurrent occurrence of diverse extreme dry/wet conditions throughout the past century. Wavelet coherence analysis reveals significant coherence between the climate patterns and TWSA across the entire basin. The reconstructed datasets provide valuable information for studying long-term climate variability and projecting future droughts and floods in the YRB, which can inform effective water resource management and climate change adaptation strategies.

Influence of coupled ocean-atmosphere phenomena on the Greater Horn of Africa droughts and their implications.

Drought-like humanitarian crises in the Greater Horn of Africa (GHA) are increasing despite recent progress in drought monitoring and prediction efforts. Notwithstanding these efforts, there remain challenges stemming from uncertainty in drought prediction, and the inflexibility and limited buffering capacity of the recurrent impacted systems. The complexity of the interactions of ENSO, IOD, IPO and NAO, arguably remains the main source of uncertainty in drought prediction. To develop practical drought risk parameters that potentially can guide investment strategies and risk-informed planning, this study quantifies, drought characteristics that underpin drought impacts management. Drought characteristics that include probability of drought-year occurrences, durations, areal-extent and their trends over 11 decades (1903-2012) were derived from the Standardized Precipitation Index (SPI).Transient probability of drought-year occurrences, modelled on Beta distribution, across the region ranges from 10 to 40%, although most fall within 20-30%. For more than half of the drought events, durations of up to 4, 7, 14 and 24months for the 3-, 6-, 12- and 24-month timescales were evident, while 1 out of 10 events persisted for up to 18months for the short timescales, and up to 36months or more for the long timescales. Apparently, only drought areal-extent showed statistically significant trends of up to 3%, 1%, 3.7%, 2.4%, 0.7%, -0.3% and -0.6% per decade over Sudan, Eritrea, Ethiopia, Somalia, Kenya, Uganda and Tanzania, respectively. Since there is no evidence of significant changes in drought characteristics, the peculiarity of drought-like crises in the GHA can be attributed (at least in part) to unaccounted for systematic rainfall reduction. This highlights the importance of distinguishing drought impacts from those associated with new levels of aridity. In principle drought is a temporary phenomenon while aridity is permanent, a difference that managers and decision-makers should be more aware.

Accounting for dynamics of mean precipitation in drought projections: A case study of Brazil for the 2050 and 2070 periods.

Changes in drought around the globe are among the most daunting potential effects of climate change. However, changes in droughts are often not well distinguished from changes in aridity levels. As drought constitutes conditions of aridity, the projected declines in mean precipitation tend to override changes in drought. This results in projections of more dire changes in drought than ever. The overestimate of changes can be attributed to the use of 'static' normal precipitation in the derivation of drought events. The failure in distinguishing drought from aridity is a conceptual problem of concern, particularly to drought policymakers. Given that the key objective of drought policies is to determine drought conditions, which are rare and so protracted that they are beyond the scope of normal risk management, for interventions. The main objective of this Case Study of Brazil is to demonstrate the differences between projections of changes in drought based on 'static' and '30-year dynamic' precipitation normal conditions. First we demonstrate that the 'static' based projections suggest 4-fold changes in the probability of drought-year occurrences against changes by the dynamic normal precipitation. The 'static-normal mean precipitation' based projections tend to be monotonically increasing in magnitude, and were arguably considered unrealistic. Based on the '30-year dynamic' normal precipitation conditions, the 13-member GCM ensemble median projection estimates of changes for 2050 under rcp4.51 and rcp8.52 suggest: (i) Significant differences between changes associated with rcp4.5 and rcp8.5, and are more noticeable for droughts at long than short timescales in the 2070; (ii) Overall, the results demonstrate more realistic projections of changes in drought characteristics over Brazil than previous projections based on 'static' normal precipitation conditions. However, the uncertainty of response of droughts to climate change in CMIP5 simulations is still large, regardless of GCMs selection and translation processes undertaken.

Hydrogeological characterisation of groundwater over Brazil using remotely sensed and model products.

For Brazil, a country frequented by droughts and whose rural inhabitants largely depend on groundwater, reliance on isotope for its monitoring, though accurate, is expensive and limited in spatial coverage. We exploit total water storage (TWS) derived from Gravity Recovery and Climate Experiment (GRACE) satellites to analyse spatial-temporal groundwater changes in relation to geological characteristics. Large-scale groundwater changes are estimated using GRACE-derived TWS and altimetry observations in addition to GLDAS and WGHM model outputs. Additionally, TRMM precipitation data are used to infer impacts of climate variability on groundwater fluctuations. The results indicate that climate variability mainly controls groundwater change trends while geological properties control change rates, spatial distribution, and storage capacity. Granular rocks in the Amazon and Guarani aquifers are found to influence larger storage capability, higher permeability (>10-4 m/s) and faster response to rainfall (1 to 3months' lag) compared to fractured rocks (permeability <10-7 m/s and lags > 3months) found only in Bambui aquifer. Groundwater in the Amazon region is found to rely not only on precipitation but also on inflow from other regions. Areas beyond the northern and southern Amazon basin depict a 'dam-like' pattern, with high inflow and slow outflow rates (recharge slope > 0.75, discharge slope < 0.45). This is due to two impermeable rock layer-like 'walls' (permeability <10-8 m/s) along the northern and southern Alter do Chão aquifer that help retain groundwater. The largest groundwater storage capacity in Brazil is the Amazon aquifer (with annual amplitudes of > 30cm). Amazon's groundwater declined between 2002 and 2008 due to below normal precipitation (wet seasons lasted for about 36 to 47% of the time). The Guarani aquifer and adjacent coastline areas rank second in terms of storage capacity, while the northeast and southeast coastal regions indicate the smallest storage capacity due to lack of rainfall (annual average is rainfall <10cm).

When every drop counts: Analysis of Droughts in Brazil for the 1901-2013 period.

To provide information useful in policy formulation and management of drought impacts in Brazil, in this study, a sequence of drought events based on monthly rainfall of 1901-2013 on ~25 km x 25 km grid are derived at 4 timescales that include short-timescales (3-month and 6-month) and medium to long-timescales (12-month and 24-month). Subsequently, probability of drought occurrences, intensity, duration and areal-extent are calculated. The probabilities of occurrence of severe and extreme droughts at short-timescales are 1 in 12 and 1 in 66 years, respectively, all over the country. At medium to long-timescales, the probability of severe droughts is about 1 in 20 years in northern Brazil, and 1 in 10 years in the south. The probabilities of extreme droughts are 1 in 9 and 1 in 12 years over northern Brazil and in the south, respectively. In general, no evidence of significant (α =0.05) trend is detected in drought frequency, intensity, and duration over the last 11 decades (since 1901) at all the 4 timescales. The drought areal-extent show increasing trends of 3.4%/decade over Brazil for both 3-month and 6-month timescales. However, the trend increases for the 12-month and 24-month timescales are relatively smaller, i.e., 2.4%/decade and 0.5%/decade, respectively.

On the potentials of multiple climate variables in assessing the spatio-temporal characteristics of hydrological droughts over the Volta Basin.

Multiple drought episodes over the Volta basin in recent reports may lead to food insecurity and loss of revenue. However, drought studies over the Volta basin are rather generalised and largely undocumented due to sparse ground observations and unsuitable framework to determine their space-time occurrence. In this study, we examined the utility of standardised indicators (standardised precipitation index (SPI), standardised runoff index (SRI), standardised soil moisture index (SSI), and multivariate standardised drought index (MSDI)) and Gravity Recovery and Climate Experiment (GRACE) derived terrestrial water storage to assess hydrological drought characteristics over the basin. In order to determine the space-time patterns of hydrological drought in the basin, Independent Component Analysis (ICA), a higher order statistical technique was employed. The results show that SPI and SRI exhibit inconsistent behaviour in observed wet years presupposing a non-linear relationship that reflects the slow response of river discharge to precipitation especially after a previous extreme dry period. While the SPI and SSI show a linear relationship with a correlation of 0.63, the correlation between the MSDIs derived from combining precipitation/river discharge and precipitation/soil moisture indicates a significant value of 0.70 and shows an improved skill in hydrological drought monitoring over the Volta basin during the study period. The ICA-derived spatio-temporal hydrological drought patterns show Burkina Faso and the Lake Volta areas as predominantly drought zones. Further, the statistically significant negative correlations of pacific decadal oscillations (0.39 and 0.25) with temporal evolutions of drought in Burkina Faso and Ghana suggest the possible influence of low frequency large scale oscillations in the observed wet and dry regimes over the basin. Finally, our approach in drought assessment over the Volta basin contributes to a broad framework for hydrological drought monitoring that will complement existing methods while looking forward to a longer record of GRACE observations.

Application of GNSS-RTK derived topographical maps for rapid environmental monitoring: a case study of Jack Finnery Lake (Perth, Australia).

In environmental monitoring, environmental impact assessments and environmental audits, topographical maps play an essential role in providing a means by which the locations of sampling sites may be selected, in assisting with the interpretation of physical features, and in indicating the impact or potential impact on an area due to changes in the system being monitored (e.g., spatially changing features such as wetlands). Global Navigation Satellite Systems (GNSS) are hereby presented as a rapid method for monitoring spatial changes to support environmental monitoring decisions and policies. To validate the GNSS-based method, a comparison is made of results from a small-scale topographic survey using radio-based real-time kinematic GNSS (GNSS-RTK) and total station survey methods at Jack Finnery Lake, Perth, Australia. The accuracies achieved by the total station in this study were 2 cm horizontally and 6 cm vertically, while the GNSS-RTK also achieved an accuracy of 2 cm horizontally, but only 28 cm vertically. While the GNSS-RTK measurements were less accurate in the height component compared to those from the total station method, it is still capable of achieving accuracies sufficient for a topographic map at a scale of 1:1,750 that could support environmental monitoring tasks such as identifying spatial changes in small water bodies or wetlands. The time taken to perform the survey using GNSS-RTK, however, was much shorter compared to the total station method, thereby making it quite suitable for monitoring spatial changes within an environmental context, e.g., dynamic mining activities that require rapid surveys and the updating of the monitored data at regular intervals.