Identification of physical processes inherent in artificial neural network rainfall runoff models

The emergence of artificial neural network (ANN) technology has provided many promising results in the field of hydrology and water resources simulation. However, one of the major criticisms of ANN hydrologic models is that they do not consider/explain the underlying physical processes in a watershed, resulting in them being labelled as black‐box models. This paper discusses a research study conducted in order to examine whether or not the physical processes in a watershed are inherent in a trained ANN rainfall‐runoff model. The investigation is based on analysing definite statistical measures of strength of relationship between the disintegrated hidden neuron responses of an ANN model and its input variables, as well as various deterministic components of a conceptual rainfall‐runoff model. The approach is illustrated by presenting a case study for the Kentucky River watershed. The results suggest that the distributed structure of the ANN is able to capture certain physical behaviour of the rainfall‐runoff process. The results demonstrate that the hidden neurons in the ANN rainfall‐runoff model approximate various components of the hydrologic system, such as infiltration, base flow, and delayed and quick surface flow, etc., and represent the rising limb and different portions of the falling limb of a flow hydrograph. Copyright © 2004 John Wiley & Sons, Ltd.     

An appraisal of groundwater quality in Seymour and Blaine aquifers in a major agro-ecological region in Texas,USA

Aquifer-based groundwater quality assessment offers critical insight into the major hydrochemical processes, and aids in making groundwater resources management decisions. The Texas Rolling Plains (TRP), spanning over 22 counties, is a major agro-ecological region in Texas from where highest groundwater nitrate (NO₃ ^?) levels in the state have been reported. In this study, we present a comparative assessment of major hydrochemical facies pertaining to NO₃ ^? contamination and a host of species such as sulfate (SO₄ ^2?), chloride (Cl^?), and total dissolved solids (TDS) in different water use classes in the Seymour and Blaine aquifers, underlying the TRP. Aquifer-stratified groundwater quality information from 1990 to 2010 was obtained from the Texas Water Development Board and aggregated over decadal scale. High groundwater salinization was found in the municipal water use class in the Blaine aquifer with about 100, 87 and 50 % of observations exceeding the secondary maximum contaminant level for TDS, SO₄ ^2?, and Cl^?, respectively in the 2000s (2000–2010). The NO₃-contamination was more alarming in the Seymour aquifer with 82 and 61 % of observations, respectively, exceeding the maximum contaminant level (MCL) in the irrigation and municipal water use classes in the 2000s. Salinization was more influenced by SO₄ ^2? and Cl^? in the Blaine aquifer and by NO₃ ^? in the Seymour aquifer. High NO₃ ^? (>MCL) observations in the Seymour aquifer occurred in the Ca–HCO₃ and Ca–Mg–HCO₃ facies, the domains of fresh water recharge and anthropogenic influences (e.g., agricultural activities, waste disposal). High SO₄ ^2?, Cl^? and TDS observations in the Blaine aquifer dominated the Ca–Cl, Na–Cl, and mixed Ca(Mg)–SO₄(Cl) facies indicating evaporite dissolution, mixing and solute exchange, and lack of fresh recharge.     

Stokes polarimetry at the Kodaikanal tower tunnel telescope

A Stokes Polarimeter has been developed using a masked CCD arrangement for the measurement of the vector magnetic field of sunspots. Charge shifting within the CCD is used to record near simultaneous orthogonal polarisation. The testing of the Stokes Polarimeter and the behavior of the integrated system combined with the Kodaikanal tower tunnel telescope will be discussed.     

Climate change projections for the Texas High Plains and Rolling Plains

Potential changes in future climate in the Texas Plains region were investigated in the context of agriculture by analyzing three climate model projections under the A2 climate scenario (medium–high emission scenario). Spatially downscaled historic (1971–2000) and future (2041–2070) climate datasets (rainfall and temperature) were downloaded from the North American Regional Climate Change Assessment Program (NARCCAP). Climate variables predicted by three regional climate models (RCMs) namely the Regional Climate Model Version3–Geophysical Fluid Dynamics Laboratory (RCM3-GFDL), Regional Climate Model Version3–Third Generation Coupled Global Climate Model (RCM3-CGCM3), and Canadian Regional Climate Model–Community Climate System Model (CRCM-CCSM) were evaluated in this study. Gaussian and Gamma distribution mapping techniques were employed to remove the bias in temperature and rainfall data, respectively. Both the minimum and maximum temperatures across the study region in the future showed an upward trend, with the temperatures increasing in the range of 1.9 to 2.9 °C and 2.0 to 3.2 °C, respectively. All three climate models predicted a decline in rainfall within a range of 30 to 127 mm in majority of counties across the study region. In addition, they predicted an increase in the intensity of extreme rainfall events in the future. The frost-free season as predicted by the three models showed an increase by 2.6–3.4 weeks across the region, and the number of frost days declined by 17.9 to 30 %. Overall, these projections indicate considerable changes to the climate in the Texas Plains region in the future, and these changes could potentially impact agriculture in this region.

     

Estimation of spectral reflectance of snow from IRS-1D LISS-III sensor over the Himalayan terrain

The sensor onboard the satellite views the earth as a plain surface and consequently the satelliteobtained spectral radiances cannot represent true values over a mountainous terrain. The relative magnitudes of terrain slope and its aspect with respect to the sun's position will determine the amount of direct solar radiation incident on an undulating surface. Estimation of spectral reflectance from satellite data forms an important component in many of the snow and glacier studies. The spectral reflectance of snow is influenced by its various parameters. The changes in snowpack characteristics as a result of various metamorphic processes, with age, can cause variations in its spectral reflectances. Since, the terrain geometry also modifies the amount of reflected radiation from a rugged surface, one has to correct the estimated spectral reflectances for terrain topography so as to use them in deriving the snowpack characteristics accurately. Also, the amounts of melt runoff originating from glaciers having different slopes and orientations will not be the same. Considering these aspects, a model has been developed to estimate the terrain corrected spectral reflectances over the Himalayan terrain using the Linear Imaging Self Scanner-III data of the Indian Remote Sensing Satellite. The model computes spectral reflectances from satellitebased radiance measurements and includes the effect of the terrain topography on the incident solar radiation. The terrain slope and its aspect are generated from the digital elevation model of the region. The analysis carried out over the Beas Basin, Himachal Pradesh, India, indicated a variation of 22% in the amount of incident solar radiation for an increase of 10‡ in terrain slope. Further, the terrain with south-east aspect received maximum amount of solar radiation. The large differences observed between the uncorrected and terrain corrected reflectances in the shortwave infrared band (B5), which is not saturated over the snow covered region, suggest that the terrain slope and its aspect cannot be neglected in estimating the accurate spectral reflectances over the Himalayan terrain.     

Pointing of HAGAR telescope mirrors

An array of seven atmospheric Cherenkov telescopes was commissioned at a high altitude site in Hanle in the Ladakh region of the Himalayas. The array called HAGAR has been designed to observe celestial γ-rays of energy >100 GeV. Each telescope is altitude-azimuth mounted and carries seven parabolic mirrors whose optic axes are co-aligned with the telescope axis. The telescopes point and track a celestial source using a PC-based drive control system. Two important issues in positioning of each HAGAR telescope are pointing accuracy of telescope axis and co-alignment of mirrors’ optic axes with the telescope axis. We have adopted a three pronged strategy to address these issues, namely use of pointing models to improve pointing accuracy of the telescopes, RA-DEC scan technique to measure the pointing offsets of the mirrors and mechanical fine-tuning of off-axis mirrors by sighting a distant stationary light source. This paper discusses our efforts in this regard as well as the current status of pointing and monitoring of HAGAR telescopes.     

Geochemistry and quality of groundwater of Gummanampadu sub-basin, Guntur District, Andhra Pradesh, India

Groundwater survey has been carried out in the area of Gummanampadu sub-basin located in Guntur District, Andhra Pradesh, India for assessing the factors that are responsible for changing of groundwater chemistry and consequent deterioration of groundwater quality, where the groundwater is a prime source for drinking and irrigation due to non-availability of surface water in time. The area is underlain by the Archaean Gneissic Complex, over which the Proterozoic Cumbhum rocks occur. The results of the plotting of Ca²⁺ + Mg²⁺ versus HCO₃ ^? + CO₃ ^2?, Ca²⁺ + Mg²⁺ versus total cations, Na⁺ + K⁺ versus total cations, Cl^? + SO₄ ^2? versus Na⁺ + K⁺, Na⁺ versus Cl^?, Na⁺ versus HCO₃ ^? + CO₃ ^2?, Na⁺ versus Ca²⁺ and Na⁺: Cl^? versus EC indicate that the rock–water interaction under alkaline condition is the main mechanism in activating mineral dissociation and dissolution, causing the release of Ca²⁺, Mg²⁺, Na⁺, K⁺, HCO₃ ^?, CO₃ ^2?, SO₄ ^2? and F^? ions into the groundwater. The ionic relations also suggest that the higher concentrations of Na⁺ and Cl^? ions are the results of ion exchange and evaporation. The influences of anthropogenic sources are the other cause for increasing of Mg²⁺, Na⁺, Cl^?, SO₄ ^2? and NO₃ ^? ions. Further, the excess alkaline condition in water accelerates more effective dissolution of F^?-bearing minerals. Moreover, the chemical data plotted in the Piper’s, Gibbs’s and Langelier–Ludwig’s diagrams, computed for the chloro-alkaline and saturation indices, and analyzed in the principal component analysis, support the above hypothesis. The groundwater quality is, thus, characterized by Na⁺ > Ca²⁺ > Mg²⁺ > K⁺: HCO₃ ^? + CO₃ ^2? > Cl^? > SO₄ ^2? > NO₃ ^? > F^? facies. On the other hand, majority of groundwater samples are not suitable for drinking with reference to the concentrations of TDS, TH, Mg²⁺ and F^?, while those are not good for irrigation with respect to USSL’s and Wilcox’s diagrams, residual sodium carbonate, and magnesium hazard, but they are safe for irrigation with respect to permeability index. Thus, the study recommends suitable management measures to improve health conditions as well as to increase agricultural output.     

Structural appraisal of the Gadag schist belt from gravity investigations

Semi-detailed gravity investigations were carried out over an area of approximately 2750 sq km with maximum N-S and E-W extents of 55 and 50 km respectively in the Gadag region in the Dharwar craton with a view to obtain a clearer perception of the structural configuration of the region. From qualitative analysis of the gravity data, several tectonic features are inferred: the high density Gadag schist belt is characterized by a gravity high and occurs in two discontinuous segments — the main N-S trending segment, and its thinner NW-SE trending extension, the two separated by a NE-SW trending deep seated fault. While the N-S trend of the Gadag schist belt is bounded on its east by the NW-SE trending Chitradurga thrust fault and on its west by another major NNWSSE trending fault, the NW-SE extension is likewise bounded by two other NW-SE major faults. Quantitative evaluation from forward modeling/inversion of five profiles in the region, assuming a density contrast of 0.29gm/cc of the anomalous schistose body with the gneissic host rocks indicated a synclinal structure plunging to the southeast along its axis for the Gadag schist belt. The maximum width and depth from surface of the schist belt are 22 km and 5.6 km respectively.     

Field based spectral reflectance studies to develop NDSI method for snow cover monitoring

Snow is highly reflective in the visible region of the electromagnetic spectrum making it possible to easily distinguish on a satellite image. However, cloud cover and mountain shadows pose a serious problem in the identification of snow in a mountainous region. Therefore, to identify snow in such an environment, a Normalized Difference Snow Index (NDSI) has been applied. The NDSI is based on the high reflectance of snow in the visible region and its low reflectance in the SWIR region, whereas, reflectance of cloud remains high compared to snow in the SWIR region. Efforts have been made to carry out field observations on reflectance of various land features near Manali in Himachal Pradesh (HP) to develop NDSI values for identifying snow. Field data have been collected using three field radiometers, viz., Multi-band Ground Truth Radiometer (GTR) operating in the 12 spectral bands ranging from visible to near-infrared wavelengths, Near-Infrared Ground Truth Radiometer (NIGTR) operating in the SWIR range, and Ratio-Radiometer (RR) operating in two spectral bands, one in the visible range, and another band in the SWIR range. All these three field radiometers have been designed and developed indigenously at the Space Applications Centre (ISRO), Ahmedabad. NDSI values for all types of snow, such as, fresh, clear, patchy and wet, have been found to be in the range 0.9 to 0.96. In addition, the NDSI value for snow under mountain shadow is found to be more than 0.9. This suggests the use of NDSI method for snow cover monitoring under mountain shadow. NDSI values for other land features such as soil, vegetation, and rock were substantially different than snow. However, water bodies have NDSI values close to snow and they need to be masked during snow cover delineation using NIR band.     

Delineation of groundwater potential zones and electrical resistivity studies for groundwater exploration

The electrical resistivity method is widely used in groundwater exploration studies. The method is particularly well suited to carry out groundwater studies in hard-rock terrains. The popular Schlumberger method of electrode configuration was applied in the study area. In all, 112 vertical electrical soundings (VESs) were carried out at different locations in the study area. VES interpretations were used to generate a top-layer contour map and a depth-to-basement map. Finally, a groundwater quality/potential map for the study area was generated using the depth-to-bedrock map, a water-level fluctuation map, and the second-layer thickness. Furthermore, this groundwater potential map was classified into three sectors, i.e. poor, moderate and good zones with respect to the prospects of finding groundwater in the study area. This map has a practical value, as it provides guidance to farmers in selecting favourable well sites, thereby avoiding unnecessary financial losses.