Please describe how you personally and professionally relate to water, space technologies and their applications to water. 

I am currently working on coastal ecosystems and their response to climate and other anthropogenic stressors in the Asia-Pacific region. An initiative called Living Indus – Investing in Ecological Restoration has become a new focus of interest for me. In this project I am particularly interested in the sustainability challenges, the need to address issues related to food security, river basin management, and efficient water use aligned with the UN Decade of Ocean objectives. Personally, my fascination with water stems from its complex and essential nature. I am captivated by its beauty in all forms and recognize its fundamental importance for life on Earth.  

Professionally, my passion for water stems from my early career with the World Wide Fund for Nature (WWF), where I was involved in Pakistan’s Wetland Program and witnessed the impact of water on associated ecosystems. Furthermore, while doing rapid response mapping for flood inundation areas. Witnessing the devastating impact of floods and the crucial role of timely information in emergency response ignited my passion for understanding and managing water resources. 

My subsequent work on watersheds and deforestation in upland and mangrove forests further deepened my water connection, recognizing its profound influence on ecosystems and human well-being. I became particularly drawn to the concept of  “More Crop Per Drop," which advocates for optimizing water use in agriculture to ensure sustainable food production for a growing population. This concept led me to explore the potential of smart sensing nexus with space technologies in enhancing water use efficiency. Space-based observations provide invaluable data for monitoring and managing water resources, enabling informed decision-making for sustainable water use and environmental protection. My research on water-related issues, including drought, ocean water quality, floods, and deforestation, highlights the critical role of these technologies in addressing global water challenges.  My work on drought in Yunnan and Sindh, floods in Pakistan, and deforestation in Swat demonstrates how space-based data can facilitates in assessing the extent and impacts of these events. For instance, satellite observations were used to track the progression of drought in Yunnan, providing early warning for potential crop failures. Similarly, satellite imagery was used to map flood inundation in Pakistan, aiding in disaster response and recovery efforts. 

My research also emphasizes the importance of spaceborne sensors in understanding ocean water quality and coastal ecosystems. Satellite data can identify and track harmful algal blooms, monitor oil spills, and measure ocean temperature, salinity, and chlorophyll levels, contributing to the protection and sustainable use of marine resources. The nexus between smart sensing and space technologies for water management practices is a transformative step towards addressing global water challenges. I am committed to furthering this integration by developing innovative solutions that promote sustainable water use, protect aquatic ecosystems, and secure water resources for future generations. 

You have worked on the monitoring of mangrove forest distribution, deforestation, water quality monitoring, drought assessment, and air pollution. Many of these are implicitly linked to water. Can you elaborate on these sometimes implicit links? What role did remote sensing play in the assessment of the above mentioned? What are the advantages of using remote sensing technology to study them? 

• Drought Assessment and Water Resources Management: Remote sensing plays a pivotal role in drought assessment by providing spatially comprehensive and timely data on drought indicators, including vegetation health, soil moisture, vegetation moisture conditions in natural forests or other ecosystems, and available water resources. Satellite sensors measure indicators of plant structure, physiological processes, and canopy/plant surface temperature, enabling the assessment of plant stress levels and the identification of moisture-deficient areas that could potentially lead to stress in a vegetated ecosystem. This information is crucial for drought monitoring, early warning systems, and water resources management strategies, ensuring ecosystem functioning, food security, and agricultural productivity. By tracking vegetation moisture conditions, remote sensing helps establish relationships between these conditions and early indications of moisture deficiency in plants, facilitating proactive measures to mitigate drought impacts and protect crops. 
• Mangrove Forest Distribution and Water Quality: Mangrove forests play a key role in maintaining water quality by acting as natural filters. They trap sediments, pollutants, and excess nutrients from water flowing through them, thereby improving water clarity, and reducing pollution levels. Mangrove also protects coastlines from erosion and storm surges. Their dense root systems act as natural barriers, trapping sediments and reducing the impact of wave action. This helps to stabilize shorelines and prevent coastal erosion, which can damage property and infrastructure. Remote sensing can be used to monitor mangrove forest distribution and assess their health, providing valuable insights into their effectiveness in water quality management and coastal protection. 
• Deforestation and Hydrology: Deforestation can significantly impact hydrological processes, including water infiltration, runoff, and streamflow patterns. Trees play a vital role in absorbing and storing water, reducing surface runoff, and promoting groundwater recharge. Deforestation can exacerbate drought conditions by reducing the amount of water that infiltrates the soil and recharging groundwater aquifers. Trees function as natural sponges, absorbing water during rainfall and releasing it slowly over time. Remote sensing can be used to monitor deforestation patterns and assess their impact on hydrology, helping to identify areas susceptible to flooding and water scarcity. 
• Water Quality Monitoring: Remote sensing can be used to monitor water quality parameters such as turbidity, chlorophyll-a concentration, and suspended sediment load. Satellite sensors can measure the reflectance of water surfaces, which can be used to infer these water quality parameters. This information is crucial for assessing the health of aquatic ecosystems as well as in-land wetlands and identifying sources of pollution and changes over time. 
• Air Pollution and Water Quality: Air pollution indirectly affects water quality by depositing pollutants into water bodies. Pollutants such as nitrogen and sulphur oxides can acidify water, while heavy metals can contaminate water sources. When acidic compounds from air pollution reach the groundwater, its quality deteriorates, often invisibly. These chemical changes in water bodies can have significant consequences, even though they may not be immediately perceivable by the human eye. As a result, we often underestimate the extent of air pollution's impact on water quality. Remote sensing plays a crucial role in monitoring air pollution levels and identifying sources of emissions. This information is essential for developing strategies to reduce air pollution and protect water quality. Understanding these relationships is essential for managing water resources effectively and minimizing the impact of air pollution on water quality. 

In conclusion, remote sensing plays a critical role in assessing and monitoring environmental parameters, including mangrove forest distribution, deforestation, water quality, drought severity, and air pollution. Its ability to provide spatially comprehensive, accessible, cost-effective, and timely data makes remote sensing an invaluable tool for environmental management and sustainable development.

What was the most exciting and relevant project you worked on related to the use of earth observations, remote sensing, and water? 

It is hard to pinpoint a single project as the most exciting, but the research on drought stands out as particularly fascinating to me. The breakthrough came with the development of the Normalized Vegetation Supply Water Index (NVSWI), an innovative measure that allowed us to comprehend the lag response of vegetation to the onset of meteorological drought or water deficiency in densely vegetated landscapes. Notably, this response varied across different vegetation cover types, adding an extra layer of complexity to our understanding. 
Additionally, delving into ocean water quality monitoring was a captivating experience. The fieldwork involved in collecting data from the ocean was not only eye-opening but also provided a unique perspective on the vastness of the oceans and the scale of phenomena within them. The impact of human pressure on ocean water quality became evident through this work. Moreover, the science of Earth observation and remote sensing data analysis in the marine and coastal environment presented distinct challenges compared to land-based systems, further enriching the breadth of my experience in this field.

After the Sindh floods hit Pakistan in 2022, you developed a framework to identify crop loss. Can you elaborate on your work and let us know if and under what conditions the framework is applicable to other regions of the world? 

The Sindh flood of 2022 was a devastating event, one of the worst natural disasters in Pakistan's history. The event was a stark reminder of the vulnerability of communities to extreme weather events, particularly in the context of climate change. Approximately 33 million people across the country were adversely affected by the flood. It is estimated that 2.5 million hectares, constituting 18% of Sindh’s total area, were submerged, with 1.1 million hectares being cropland. The remaining crop damage resulted from intense rainfall, flash floods, and deficiencies in management.  As part of the loss and damage (L&D) assessment methodologies, a comprehensive framework was developed to effectively evaluate crop-specific post-disaster production losses based on multi-sensor satellite data. This framework leverages the power of satellite data and cloud-based computing to provide valuable insights into the impact of floods on agricultural productivity. The framework's versatility stems from its ability to integrate information sourced from various satellites and its operational nature is attributed to its reliance on cloud-based computing. This combination enables the framework to assess post-flood crop losses with precision and efficiency. 

Firstly, the flood extent or inundated area was extracted using Synthetic Aperture Radar (SAR) data from the Sentinel-1 satellite, which has the capability to penetrate through clouds and easily detect water bodies to generate maps identifying inundated regions, by comparing pre and post flood water extent maps. Subsequently, rainfall patterns and intensity data were derived from Global Precipitation Mission (GPM), providing insights into the direct impact of rainfall on standing crops. Additionally, Sentinel-2 optical imagery was processed to create pre- and post-flood NDVI (Normalized Difference Vegetation Index) maps, serving as a crucial indicators for evaluating the damage inflicted on standing crops as measure of loss in greenness.  

The cropland maps were obtained from European Space Agency (ESA)’s World Cover maps, providing a comprehensive view of the agricultural landscape, and essential information on areas actively planted with crops, enabling a targeted assessment of crop damage. By overlaying the diverse data layers derived from the satellite data sources, a comprehensive analysis was conducted to identify regions with a high likelihood of crop damage. This multidimensional approach provides a holistic view of the flood's impact on cropland, allowing for informed decision-making and resource allocation. 

The satellite-based framework developed for assessing crop loss in the aftermath of the Sindh floods offers a valuable tool for evaluating the impact of floods on croplands. Its applicability extends beyond the Sindh floods, demonstrating its potential for assessing crop loss in other regions of the world. However, its effectiveness is contingent upon the availability of satellite data, adequate vegetation cover, and access to cropland extent maps. Generally, the necessary conditions for its effective implementation are not a significant concern, provided that the required satellites are in-orbit and the possibility of cloud-free optical images immediately after the flood. Cloud free imagery prior to the flood can be obtained easily, however, it is also bounded by a time window: the crop or vegetation growing season. This data could also be updated and revised with other relevant and better datasets for more refined results. The assessment has provided valuable tools for evaluating the loss and damage of agricultural production and developing evidence-based policies to facilitate post-flood recovery, rehabilitation of affected populations, and restoration of livelihoods.  

For more in-depth details, please refer to our publication on this framework, titled " A framework for multi-sensor satellite data to evaluate crop production losses: the case study of 2022 Pakistan floods (2022) " which provides a comprehensive overview of the methodologies employed and the results obtained. 

Your publication on Characterization of Drought Development through Remote Sensing: A Case Study in Central Yunnan, China included a Normalized Vegetation Supply Water Index, can you explain what it does compared to other similar indices, and what its intended use includes?  

In the context of drought monitoring, the Normalized Vegetation Supply Water Index (NVSWI) distinguishes itself from other indices by avoiding the reliance on background soil information. While conventional indices may excel in landscapes with sparse canopies having sufficient background soil, the NVSWI proves its efficacy in dense canopies. Numerous comparative studies have validated its superior performance, particularly in environments characterized by dense vegetation cover. Understanding the response of vegetation to precipitation requires acknowledging the time lag associated with several factors including regional rainfall patterns, soil type, land cover, and vegetation type. The temporal dynamics of vegetation response play a crucial role in drought prediction, necessitating a nuanced approach that considers these time lags. 

To enhance the meaningful comparison of drought severity across different periods, the NVSWI undergoes a normalization process. This transformation results in the creation of the NVSWI, allowing for realistic and absolute comparisons within the study period. This normalization proves crucial in portraying the relative drought conditions across different years making the NVSWI a valuable tool for understanding drought severity on an absolute scale. A critical aspect of the study involves the analysis of lag times between rainfall events and vegetation response. The investigation reveals that a lag time of 64 days significantly influences soil moisture levels, particularly in cropland (CL) and shrubland (SL). Incorporating such lag time information into remote sensing-based drought indices enhances their accuracy, offering improved assessment and prediction capabilities. 

The study's insights extend to the diverse responses observed in different land cover types. Evergreen forests (EF) demonstrate a delayed response to rainfall-induced drought, attributed to its characteristic ability to enter a semi-dormant state. In contrast, crops exhibit heightened sensitivity to recent rainfall and near-surface soil moisture, both being crucial factors for initiating a rapid response. To conclude, the NVSWI and its derivatives prove reasonable for drought monitoring, especially in environments with dense vegetation cover. By considering lag times and normalizing the index, researchers can enhance the accuracy of drought assessments and predictions, gaining valuable insights into the complex dynamics of vegetation response to associated deriving factors. 

For more in-depth details, please refer to our publications, titled “Characterization of Drought Development through Remote Sensing: A Case Study in Central Yunnan, China”, and “Integration of remote sensing datasets for local scale assessment and prediction of drought” 

What were the challenges and your lessons learned applying machine learning algorithms to monitor water quality based on satellite data in the subtropical region?  

Applying machine learning for water quality monitoring in subtropical regions comes with various data-related challenges. The scarcity of in-situ data hampers model training and validation, potentially causing overfitting. Diverse landscapes and/or seascapes and weather patterns necessitate meticulous data normalization and preprocessing to accommodate spatial and temporal variations in water quality parameters. Frequent cloud cover and atmospheric interference in humid regions significantly reduce usable optical remotely sensed satellite data availability, hindering continuous monitoring efforts. Moreover, the spectral resolution and characteristics of satellite sensors may not be optimal for water quality analysis, demanding careful selection of relevant spectral wave bands and features (visible and near-infrared spectral wave bands are used mostly). Applying atmospheric corrections to minimise influence of atmosphere for obtaining near-true water leaving reflectance is also a challenging task while associating water quality parameters with spectral characteristics.  Innovative approaches to data collection, analysis, and model development are crucial for overcoming these challenges and unlocking the full potential of machine learning in subtropical water quality monitoring. 

To develop successful machine learning applications it is crucial to consider numerous interconnected challenges. When developing machine learning applications, there are many algorithms available, each with its strengths and weaknesses. Some algorithms may be very accurate but computationally expensive, meaning they require a lot of computational resources and time to train and execute. On the other hand, some algorithms may be less accurate but much more computationally efficient. Therefore, selectin of an optimal algorithm requires a delicate balance between accuracy and computational efficiency for real-time monitoring. Achieving this balance involves meticulous feature engineering, selecting the most relevant features from satellite data and fostering a deep understanding of the relationships between water quality parameters and spectral bands.  

Beyond algorithms, interpreting complex models poses additional hurdles, making it challenging to unravel predictions and potential biases. Rigorous model validation through independent datasets is crucial, with particular attention to overfitting, where models may become overly tailored to training data. The regional specificity of models can limit broader applicability, necessitating continuous evaluation and updating for adaptability to changing conditions. Collaboration with domain experts and authorities is essential for valuable insights, ground truth data access, and model validation. Ensemble modelling, which is the integration of predictions of multiple models, improves robustness under diverse environmental conditions. Integration of multiple data sources enhances understanding of water quality dynamics. Communicating uncertainties associated with predictions is necessary for responsible decision-making. 

Innovative strategies for model adaptation and transferability can maximise the impact of machine learning in water quality monitoring across regions. Addressing interconnected challenges through innovative data collection methods (use of emerging technologies such as drones, space technologies or remote sensing, IoT devices, or advanced sensor networks, to collect water quality data), interpretable model design, and robust validation unlocks machine learning's full potential, paving the way for accurate, efficient, and transparent monitoring systems and improved water quality management in critical ecological areas. 

Understanding the specific water quality challenges and the influence of climate zones in subtropical regions is key for effective strategies to monitor and manage water resources. Machine learning can play a vital role in this effort by analysing complex data sets, identifying trends, and predicting water quality changes. However, addressing data limitations, choosing appropriate algorithms, and interpreting model outputs with caution is essential for ensuring the reliability and effectiveness of machine learning solutions for water quality monitoring in these dynamic and challenging environments. 

 What is specific to water quality in the sub-tropical region and how do climate zones affect water quality?  

Water quality in subtropical regions is shaped by various environmental factors. The temperature fluctuations with significant variations that are characteristic of these areas, can accelerate biological and chemical processes within water bodies. This can manifest in the form of algal blooms and alterations in dissolved oxygen levels. High humidity and periodic heavy rainfall, common in subtropical climates, contribute to elevated runoff from land surfaces. This runoff often carries pollutants like sediments, nutrients, and pesticides into water bodies, exerting an influence on overall water quality. Intense storm events, which are common in subtropical regions, can lead to flash floods and increased erosion, facilitating the rapid transport of contaminants from land to water bodies. Additionally, the diverse vegetation and land use patterns in subtropical areas, coupled with distinct seasonal variability, further contribute to the complexity of water quality dynamics. 

The impact of climate zones on water quality is significant and multifaceted. Temperature variations associated with different climates influence the solubility of gases and the metabolic rates of aquatic organisms. Elevated temperatures can promote the proliferation of certain bacteria and algae, impacting water quality parameters. Precipitation patterns dictated by climate zones play a crucial role, with increased rainfall leading to runoff and the transportation of pollutants into water bodies. Conversely, drought conditions concentrate pollutants due to reduced dilution. Extreme weather events, such as hurricanes, typhoons, and cyclones, are influenced by climate patterns or regimes. Different regions experience distinct and recurring climate patterns, which can affect the occurrence and intensity of extreme weather and can cause substantial disturbances to water bodies. Climate change, associated with specific climate regions, contributes to sea-level rise, impacting freshwater resources through saltwater intrusion. Furthermore, climate-induced changes in hydrology alter the availability and distribution of water resources, affecting pollutant concentrations and overall water quality. Understanding these climate-related factors is vital for implementing tailored water quality management strategies in different regions. 

Compared to traditional methods, what are the advantages and limitations of applying machine learning algorithms to water quality monitoring based on satellite data? Where do you see unharnessed potential in this field of research? 

Machine learning algorithms have revolutionized water quality monitoring, offering distinct advantages over traditional techniques. Firstly, by analysing large-scale satellite data, these algorithms provide comprehensive and accurate monitoring across vast areas, eliminating the limitations of geographically dispersed sampling points. Additionally, machine learning enables near real-time monitoring, promptly detecting changes in water quality and allowing for timely interventions to address problems. This automation also translates to cost-effectiveness and significantly reduces the need for manual labour and enabling efficient data processing. Finally, machine learning can facilitate information extraction at superior temporal resolution, allowing for continuous tracking of water quality changes over time, a crucial aspect for understanding long-term trends and predicting future conditions. 

Despite its undeniable advantages, machine learning applications in water quality monitoring face certain challenges. The accuracy of model predictions relies heavily on data quality. Inaccurate or biased satellite data can significantly impact results. In remote areas, the limited availability of ground truth data, essential for training and validating models, remains a challenge. Additionally, complex models can be difficult to interpret and are prone to overfitting, requiring careful development and evaluation.  

The field of machine learning for water quality monitoring holds immense untapped potential. Integration with ground-based sensors offers a promising avenue for enhancing model accuracy and reliability by combining the strengths of both satellite and in-situ data. Research into advanced feature engineering techniques tailored to specific water quality parameters could further improve model performance and enable more precise monitoring. Combining machine learning with physical models in hybrid approaches could also enhance interpretability and provide deeper insights into the underlying processes. Engaging communities in data collection and validation efforts could address the limitations of ground truth data and make models more applicable to diverse regions. 

Furthermore, the emergence of new data sources unlocks exciting possibilities for water quality monitoring. Geostationary satellites offer near-continuous monitoring, enabling the detection of rapid changes and improving understanding of diurnal variations. Hyperspectral satellites provide significantly more spectral bands than traditional multispectral sensors, allowing for more precise identification and quantification of various water quality parameters. Additionally, high temporal resolution multispectral datasets offer frequent observations with high spectral wave bands, capturing dynamic changes and improving model training and validation. Developing machine learning algorithms that can effectively exploit these diverse data sources will be crucial to unlocking their full potential. Deep learning architectures, particularly convolutional neural networks, are well-suited to manage the high dimensionality of hyperspectral data and extract meaningful features for water quality prediction. Additionally, transfer learning techniques can leverage knowledge gained from one data source to improve performance on another, further enhancing the accuracy and generalizability of water quality models. 

By embracing these diverse data sources and advancements in artificial intelligence and data science, we can unlock a new era of water quality monitoring. Near-real-time, precise, and comprehensive monitoring will enable early detection of contamination events, facilitate proactive water management interventions, and ultimately contribute to a healthier planet for all. 

Please explain what are the key machine learning algorithms relevant for Earth observation and remote sensing of water? How can people new to machine learning identify if a chosen approach is a good fit? 

Navigating the vast landscape of machine learning algorithms can be daunting, especially for those new to the realm of water remote sensing. Yet, certain algorithms stand out as invaluable tools for tackling diverse challenges.  

• Random Forest, with its exceptional ability to handle large datasets and robustly resist overfitting, excels in land cover classification and identifying key features within the data. This insightful analysis makes it a valuable asset for researchers seeking to understand the composition of land surfaces.  
• Support Vector Machines (SVM) boast remarkable versatility, thriving in high-dimensional spaces (data with many attributes or variables) and delivering effective performance across various data types. Their prowess in land cover classification, change detection, and mapping tasks makes them valuable for monitoring and analysing land cover dynamics.  

For tackling complex tasks like water quality analysis and parameter prediction, Gradient Boosting Machines, such as XGBoost and LightGBM, rise to the challenge. Their ability to handle non-linear relationships and resist outliers, coupled with their high predictive accuracy, makes them instrumental in understanding the intricate dynamics of water systems.  

When dealing with image-based analyses like object detection and segmentation, Convolutional Neural Networks (CNNs) come into play. Their exceptional capability to recognize spatial patterns and learn hierarchical features directly from the data makes them ideal for extracting valuable insights from imagery.  

For those looking to understand time-varying phenomena in water bodies, Recurrent Neural Networks (RNNs) offer a powerful solution. Their ability to capture temporal dependencies in sequential data, such as water level measurements over time, proves invaluable for uncovering trends and patterns in dynamic systems.  

Finally, for exploring unlabelled data and uncovering hidden structures, Unsupervised Learning algorithms like K-Means and Hierarchical Clustering play a crucial role. These techniques allow researchers to identify patterns and clusters within data, providing valuable insights into complex systems without prior labelling. 

By carefully considering these factors and beginning with simpler models, beginners can effectively navigate the realm of machine learning and identify the most suitable approach for their specific water-related earth observation and remote sensing tasks. Approach to solve a problem with machine learning involves a simple pathway:

• Understand your objective, it emphasizes the importance of understanding the specific goal. This helps to choose relevant approach whether it's classification, regression, or clustering.  
• Data characteristics, including spatial, spectral, and temporal domain, should be carefully analysed, as certain algorithms perform optimally with specific data features. 
• Begin with simple models and structures, for beginners, starting with simpler models like Random Forest or Support Vector Machines (SVM), as their simplicity often leads to easier interpretation and implementation. 
• Experimentation with different algorithms and comparison of their performance is encouraged, recognizing that an ensemble or combination of algorithms can enhance results.  
• Training and testing machine learning models are crucial for evaluating their performance and generalization ability. Sufficient training and testing data ensure that models can learn effectively and make accurate predictions across different datasets. 
• Validation of models, implementing cross-validation techniques is essential for assessing the generalization of the chosen approach and identifying potential overfitting or underfitting issues. 
• Community support and online learning, considering community support and available resources for the chosen algorithm guarantees access to abundant learning materials and problem-solving assistance. Exploring resources, such as documentation, webinars, tutorials, and books with practical applications, helps to understand algorithm parameters and their application in earth observation and remote sensing.  

Recently, scholars have been assessing the suitability of deep learning (DL), a subset of machine learning (ML), for us in diverse hydrological disciplines and water-related applications. This surge in interest has resulted in several review articles exploring the current situation (Ghobadi and Kang 2023).  

How do you best identify if machine learning has advantages to traditional approaches for a given problem? 

Determining whether machine learning provides advantages over traditional approaches for a given problem involves a careful examination of a range of factors. Firstly, consider the complexity of the problem—machine learning excels in handling intricate patterns and non-linear relationships. The volume and nature of available data play a crucial role; machine learning is particularly advantageous when dealing with large datasets and high-dimensional feature spaces (data with many attributes or variables). Assessing the adaptability of the problem to change, as machine learning models excel in dynamic environments.  

Additionally, weigh the interpretability requirements, computational resources, and existing domain knowledge. Understanding the specific performance metrics and conducting rigorous model validation help determine the suitability of machine learning versus traditional methods. It's a meticulous evaluation where factors such as problem complexity, data characteristics, and interpretability need to guide the choice between machine learning and traditional approaches. 

Can you imagine that we will be able to monitor more parameters or indicators of water quality from space in the future? If so, which ones? And which role do you see in machine learning. 

Advancements in technology, particularly machine learning, hold the promise of significantly expanding water quality monitoring capabilities from space. This evolution will be propelled by the growing availability of data from geostationary satellites and hyperspectral/high spectral and spatial resolution images, offering unprecedented temporal and spatial resolutions. The synergy between these data sources and advanced machine learning algorithms, including self-learning models and cloud computing platforms, will be instrumental in scaling up water quality monitoring. This will facilitate the assessment of critical parameters like Chlorophyll-a, Turbidity, Dissolved Organic Matter (DOM), Harmful Algal Blooms (HABs), microbial communities, water temperature, water depth, marine debris, and plastic pollution. This expanded monitoring is particularly crucial in the face of climate change and extreme sea surface temperatures. 

Excitingly, a promising array of upcoming space-based missions including Surface Water and Ocean Topography (SWOT), Plankton, Aerosol, Cloud, ocean Ecosystem (PACE), Sentinel-6B, Geosynchronous Littoral Imaging and Monitoring Radiometer (GLIMR), Landsat Next, and AquaWatch program.  More details about these upcoming missions could be found at https://science.nasa.gov/earth-science/missions/earth-missions/earth-mi…. These missions are set to significantly boost our water monitoring ability, bridging the gap between ocean and land and advancing our comprehension of the global water cycle.  

Advancements in technology help us better understand and manage our water resources. Deeper insights will empower us to manage it more effectively amidst the challenges of climate change. Artificial Intelligence together with data science will revolutionize water quality monitoring from space by efficiently analysing huge datasets. It will unveil hidden (unexplored) patterns, predict future changes, and automate tasks, freeing experts for strategic endeavours. This integration will enhance data collection, analysis, and model development, ultimately leading to more effective protection and management of our vital water resources.

How do you keep abreast of the rapid developments in remote sensing, machine learning and all the fields of application you work on?  

In this era, we possess abundant resources, yet keeping up with the rapid developments in various fields poses challenges. As a spatial data scientist, navigating the nexus between smart digital technologies, earth observations, climate, and development requires staying abreast of updates in these dynamic domains. To meet this challenge, I employ several methods to stay at the forefront of Earth Observation, Smart Sensing Nexus, Data Science, and their diverse applications. 

• Immersing myself in academic literature is a fundamental approach. Regularly exploring leading journals and conference proceedings through renowned databases enables me to seek out relevant publications and stay informed about the latest advancements.  
• Reports from reputable bodies such as the IPCC (Intergovernmental Panel on Climate Change), IPBES (Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services), GEOS (Group on Earth Observations), ADB (Asian Development Bank), World Bank, UN portals, and others are instrumental in keeping me updated with the latest literature and guiding my research. Additionally, insights from reports related to UN SDGs (Sustainable Development Goals) play a crucial role in shaping the direction of my research, ensuring alignment with global sustainability objectives. 
• Providing review services to top journals in different target domains also contributes to my knowledge. Staying connected to leading researchers and institutions through email alerts and social media platforms, including Twitter, LinkedIn, and Facebook, ensures continuous awareness of cutting-edge research. Review articles and surveys serve as invaluable summaries, offering a concise roadmap for further exploration. 
• Harnessing the power of online resources is crucial. Numerous websites, blogs, and online seminars provide a wealth of information. Following relevant blogs and news websites keeps me updated on developments, while online seminars and webinars offer direct insights from experts. Working with real-world data through online datasets provides practical insights, bridging the gap between theory and application. 
• Building a robust network of professionals is another key strategy. Connecting with researchers and practitioners in the field fosters a vibrant learning environment. Conferences and workshops provide opportunities to gain experience, network, and gain valuable feedback. Engaging in online communities and forums facilitates collaborative discussions and shared learning experiences. 
• Embracing active learning is essential for targeted knowledge expansion. Identifying specific topics relevant to current projects ensures focused and efficient learning. Experimenting with innovative technologies and contributing to open-source projects enhances hands-on experience and skill development. Engaging with research students keeps my perspective fresh and allows me to keep an eye on technical advancements. 
• Engaging with the next generation is critical for fostering learning and innovation in this era of rapid technological development. Engaging with young brains, students, and researcher, is not just about transferring knowledge; it is about understanding their needs and encouraging them to delve into innovative thinking and adapt new tools. By doing so, we not only transfer our knowledge and new developments but also instil in them a passion for learning. This interaction helps me foster my own learning in the domain, as I am constantly challenged to think in new ways and stay updated with the latest trends and technologies. 
• Fostering a culture of continuous learning is a core principle. Dedication to learning, even in short daily periods, ensures consistent investment in knowledge expansion. Viewing learning as a lifelong journey promotes openness to innovative ideas and perspectives, ensuring adaptability to the ever-changing landscape of remote sensing and machine learning. 
• Combining these approaches enables effective navigation of the rapidly evolving world of remote sensing, machine learning, and their applications. Active engagement with academic literature, online resources, professional networks, and a commitment to active learning facilitates meaningful contributions to the advancement of these fields. 

What do you need to innovate? 

Understanding, predictive modelling, and identification of risks and vulnerabilities are crucial for proactive strategies to transform uncontrollable forces of nature in to manageable challenges. Innovation requires a comprehensive approach across various interconnected spheres. The development of advanced algorithms that integrate diverse datasets, including environmental, satellite, socio-economic, and field observations, is a key focus. These tools aim to illuminate complex interactions in river basins, enabling more effective management strategies. Interactive platforms for visualizing natural resource data enhance accessibility, fostering informed decision-making and collaboration. 

Continuous improvements in remote sensing technologies, such as high-resolution imagery, drones, LiDAR, and artificial intelligence, contribute to accurate monitoring of water availability and environmental changes. Strengthening water governance involves participatory approaches empowering local communities and developing water-saving technologies for sustainability. Innovative financing mechanisms bridge funding gaps for water infrastructure, ensuring long-term viability. 

What is your favourite aggregate state of water, and why?  

My favourite aggregate state of water is solid. Appreciating the distinct properties of each aggregate state of water, the focus should shift to recognizing the versatility of water and its vital role in Earth's natural systems. In the context of global warming, emphasizing the preservation of the solid state, specifically snow and ice cover, becomes crucial. This preservation holds numerous benefits, including regulating global temperatures by reflecting sunlight and acting as a natural coolant. Preserving ice cover is essential for freshwater storage, as glaciers and ice sheets serve as significant reservoirs, ensuring a stable water source for future generations. Additionally, the preservation of ice and snow supports ecosystem stability, and biodiversity, and plays a role in disaster mitigation by regulating water availability and preventing floods. Beyond these ecological advantages, there is a cultural significance in preserving ice and snow, honouring the traditions and identities of Indigenous communities globally. In summary, safeguarding the solid state of water is paramount for mitigating the impacts of global warming and ensuring a sustainable future, contributing to temperature regulation, freshwater storage, ecosystem support, disaster mitigation, and cultural preservation. Recognizing and acting upon the importance of this natural phenomenon is imperative for effective conservation efforts.