Soil moisture is an essential climate variable with an important role in the hydrological cycle, and is critical for agricultural water management, floods & droughts monitoring, forest-fire control, locust attack prediction and many other applications. While there exist several in-situ soil moisture networks across the globe, there is an absence of a large-scale network for the Indus Basin. This paper summarizes our five years of work in the Indus basin to establish a soil moisture monitoring network, WITSMS-Network with more than 200 installations, primarily to empower farmers for real-time irrigation scheduling. The sensors were calibrated using the gravimetric method with a R2value of 0.9388 and RMSE of 0.025 m3m−3for WIT’s indigenously developed WZ-1 soil moisture probe. Both WZ-1 and commercially available soil moisture probes were integrated into an IoT hardware-software backbone for real-time data transmission, storage and sharing over the Internet. We also offer an approach to validate the soil moisture network, based on TCA metrics, against coarse resolution satellite and model-based soil moisture products including SMAP-L2, C3S SM combined, ESA CCI SM combined, and ECMWF ERA5-Land model. Our network achieved an ubRMSD 0.0460 m3m−3over 2 years of soil data records. These reproducible validation results have been generated using a community-wide accepted QA4SM validation framework. The soil moisture data is being published at ISMN under WIT-Network1.

The Indus floodplain, known for its fertile soil and traditional crop production, has been heavily affected by floods in 2022. In previous flood disaster management studies of this region, there has been a lack of focus on using remote sensing to assess the time evolution of flood water depths. This research addresses the gap using multi-temporal Sentinel-1A, 2A and ground-based validation survey methods. This has enabled the study of the spatial and temporal variations in flood water depth and agricultural damage for the 2022 flood events in Pakistan’s 21 worst-affected districts. In this work, a Google Earth Engine (GEE) based flood depth estimation model has been used. Moreover, we have used the In-SAR technique to measure changes in flood water levels and evaluated its accuracy in changed environmental situations. The results indicates that a total of 26,252 km2area was affected by the flood, where water depths reached a maximum of 7 m in the northern and western parts of the lower Indus basin. A total of 1,307,102.8 ha of agricultural land was submerged in flood water. The In-SAR approach achieved a spatial flood depth accuracy of 88%, as compared to the GEE-based Floodwater Depth Estimation Tool.

Previous studies on the cryosphere of the Upper Indus Basin (UIB) have mostly focused on glaciology and have either used remote sensing tools or sparsely deployed ground station networks at the regional or basin scales. This work is instead focused on catchment-level seasonal snowpack monitoring in the little-studied foothills of UIB, using a combination of remote sensing and in-situ measurements. To estimate the snow depth at the catchment level, we have processed Sentinel-1A (C-Band) data and validated our estimates by an indigenously developed ground-based monitoring station. We used differential interferometry for the inversion of spatial snow depth from November 2018 to May 2019. The results indicate a spatial maximum of 350 cm snow depth within the catchment, and a temporal maximum of 216 cm at the validation point. The relative error (RE) with respect to the ground station was estimated to be 90% and an RSME of 3.0 cm. The good performance of TOPS data with spatial-temporal baselines and inexpensive in-situ validation offers significant potential in future applications of snowpack monitoring and water resources management studies in the Upper Indus Basin of Pakistan.

Large-area crop classification using multi-spectral imagery is a widely studied problem for several decades and is generally addressed using classical Random Forest classifier. Recently, deep convolutional neural networks (DCNN) have been proposed. However, these methods only achieved results comparable with Random Forest. In this work, we present a novel CNN based architecture for large-area crop classification. Our methodology combines both spatio-temporal analysis via 3D CNN as well as temporal analysis via 1D CNN. We evaluated the efficacy of our approach on Yolo and Imperial county benchmark datasets. Our combined strategy outperforms both classical as well as recent DCNN based methods in terms of classification accuracy by 2% while maintaining a minimum number of parameters and the lowest inference time.

Developing countries have poor wastewater management infrastructure. Wastewater from various sources such as domestic sewage, untreated industrial discharge, agricultural runoff, and animal waste is usually dumped into surface water resources such as canals, streams and rivers. This results in the degraded quality of surface water which poses adverse effects on the ecosystem and human socio-economic needs in many different ways. Considering the socioeconomic needs of the current and future generations, steps need to be taken to restore these water bodies to their natural form. In this regard, we require some monitoring tools which can help identify the main sources of pollution. To answer such questions, the authors have developed a Lagrangian sensor system which is capable of logging and transmitting spatially distributed GPS tagged water quality data in real-time from water bodies such as drains, canals, streams and rivers. It can measure the physico-chemical parameters of water quality such as, pH, dissolved oxygen, electrical conductivity, oxidation-reduction potential, temperature and turbidity. The paper describes the design, development and deployment stages of the floating device. We report our initial successful field experiment of continuously recording six water quality parameters across 4.3 km length of an irrigation canal in Pakistan.

Traversability analysis is an important aspect of autonomous navigation in robotics. In this paper, we relate the idea of traversability to safety and ease of road usage by defining a novel sensor-data driven metric called Road Traversability Index (RTI). The RTI translate the geometric interaction of vehicle with road into a distance modulated index that can be used as advice for a human driver or an autonomous agent intending to traverse a particular road segment using a specific vehicle. We present a framework in which 3D sensor data is converted into a road model, which in turn is converted into a roadmap based motion planning graph to represent the underlying configuration space. The RTI is defined as a function of the roadmap by axiomatically satisfying all required properties of road traversability. We have tested our algorithmic framework on simulated scenarios to explore safety; and real-world data sets to discover aspects of traversability for vehicles of various types. Experimental results show that RTI is a practical tool that reveals information that may be hidden to human inspection or other methods of assessment that do not explicitly model a vehicle.

Silt accumulation and sedimentation in canal beds leads to deterioration of watercourses over time. Every year a forced closure of the canals in the Indus basin is inevitable for canal cleaning, entailing a very large scale and costly operation. Silt removal precision is prone to inefficiencies due to subjective decision making in the cleaning process. In this paper, we lay out a theoretical framework to map the semi-structured (emptied) canal bed terrains with an Unmanned Aerial Vehicle (UAV) system for quantitative inspection of deposited silt. The study employs Gaussian process regression on sampled points to determine a continuous distribution of silt surface, thereby, predicting the volume of silt on canal bed. Our theoretical analysis builds upon certain mathematical bounds on the variance of estimated volume, while explicitly considering localization error and sensor noise. Essentially, we setup a framework for studying how tolerable are the process and measurement uncertainties, while achieving a desired accuracy in silt profile and corresponding volume. We demonstrate the regression results in simulations as well as lab-scaled-model (LMS151 laser scanner) with different sets of parameters. Volume estimation is verified practically and mathematical performance limits are proposed in established aerial canal inspection system.

Inspired by the need to monitor and control the world's largest irrigation and river networks such as those in the Indus River Basin, we have developed a smart water meter to gauge open channel flows. While developing this design, we have overcome challenges related to power consumption, precision, accuracy, calibration, field conditions, data communication and cost. Large scale deployment of such WSN inspired meters will ensure equity of water distribution, better management opportunities and hence reduce water scarcity.

In this paper, we focus on setting up a gunshot detection system with high detection performance, robustness to noise and low computational complexity. To achieve these objectives, we formulate a two-stage approach with a less costly impulsive event detection framework followed by a relatively more complex gunshot recognition stage. To improve detection performance of the gunshot recognition stage, we propose a template matching measure in conjunction with the eighth order linear predictive coding coefficients to train a support vector machine classifier. Using an extensive audio database, we were able to achieve a better gunshot recognition performance than with the well-known existing features used for gunshot detection.

Motion planning in high dimensional configuration space is an intractable problem. Sampling based motion planning is considered as one of the best ways to tackle the "curse of dimensionality" and to model the configuration space. In these methods (e.g. probabilistic roadmap (PRM)), a graph is produced by random nodes of valid configurations. As the number of the nodes N increases, for expansive configuration spaces the failure probability of PRM planner exponentially approaches zero; but computing expansiveness property is not feasible for most interesting problems. Another major hurdle is narrow passages detection. Therefore, we need easily computable ways to characterize the properties of configuration spaces. We propose a new framework for narrow passage detection using spectral analysis of the graph Laplacian of the PRM. We give empirical evidences to show that eigenvalues and eigenvectors reveal useful information about number and size of narrow passages, visibility and expansiveness. Simulations on various motion planning scenarios are done to verify the framework.

We consider coverage problems in sensor networks of stationary nodes with minimal geometric data. In particular, there are no coordinates and no localization of nodes. We introduce a new technique for detecting holes in coverage by means of homology, an algebraic topological invariant. The impetus for these techniques is a completion of network communication graphs to two types of simplicial complexes: the nerve complex and the Rips complex. The former gives information about coverage intersection of individual sensor nodes, and is very difficult to compute. The latter captures connectivity in terms of inter-node communication: it is easy to compute but does not in itself yield coverage data. We obtain coverage data by using persistence of homology classes for Rips complexes. These homological invariants are computable: we provide simulation results.