Since many had asked this, Writing a short thread on the formation between India and Sri Lanka
Usually many get deceived by "static" satellite images and believe that the Tombolo section as permanent relics of a man made bridge.
Deeper ocean currents don't enter the section between Sri Lanka and India because of the Cont Shelf. The sea surface in this section is dominated by longshore currents in two directions - one from Gulf of Mannar and other from Palk strait and they are in opposite directions
Longshore currents bring a lot of sediments. And at the meeting point of the currents they settle and form these islands. Notice in the satellite images below how these islands change a lot owing to current direction, tides, etc. This is just a short term timelapse (<4 years)
Though there are articles on Koshi River n flooding in Bihar, I prepared this thread to visually explain it!
Image: Topography of #Bihar
Despite being far from sea most of Bihar people live at an elevation lesser than 100m similar to Kerala
R.Kosi is a tributary of Ganga. It's watershed covers Mt.Everest but at it's confluence with Ganges, the elevation is just 25 m above sea level. All this within an aerial distance of 300kms! Along with seasonal rainfall, this has led to massive erosion as well as deposition
Cross-sectional profile of the watershed shows this variation
After flowing in higher gradients cutting through the himalayas, Kosi river exits n meets flatter plains section with very less gradient. This leads to millions of tonnes of silt deposition in this section every year
Currently elevation data like SRTM have a vertical error i.e height data created by satellites don't represent ground height very accurately due to buildings, trees etc. So while doing analysis there's a possibility of understimating areas affected by floods, storm surge etc
Inorder to overcome this issue, the authors of the paper had tried to create an AI model that can correct the elevation bias in the data. The AI model is built with ~23 parameters like population density, vegetation etc to explain the variation in the satellite data.
@rapiduplift@JamwalNidhi SRTM does have vertical bias however the correction methodology is not sound enough to overcome that in a more accurate manner.
The problem I see is with ML algorithm which uses LIDAR data and ~23 parameters from developed countries as training and testing datasets
@rapiduplift@JamwalNidhi These parameters come in coarse resolution datasets (some of them as coarse as a km). Also, The relationship between those parameters and Elevation bias would vary between developed and developing countries. However the training data doesn't account this.
@rapiduplift@JamwalNidhi For eg, population density is related to Building heights to some extent. But it cannot be assumed as the same for US and India. So it is likely to exaggerate bias in India
Params like Vegetation r also problematic as they report difficulty in estimating for places like Florida
Sol 1: #RainWaterHarvesting - Chennai has urbanized pretty fast n percolation into ground is getting reduced. Building wise RWH is one,but we need RWH in public spaces like bus stands etc; Every year Corporates spend a lot of CSR which could be used
Sol 2: #WasteWaterReuse - Currently most of the sewage is going to the sea via Cooum,Adyar rivers, Buckingham canal or ending in the lakes. This water could be reused through decentralized sewage treatment and reuse system