Amrit, go first for gathering precipitation observations based on high altitude gauges (As Kenneth mentioned, Nepal Bureau of Standards & Meteorology would be a relibale source). The global gridded data often ingnor the orogeraphic precipitation mechanisims and approximate the values using solely interpolation methods which can cause remarkable errors.

Dear Kenneth and Mo,

I really appreciate your ideas. We are in contact with Department of Meteorology and Hydrology (DHM), Nepal. I have observations of precipitation in Nepal. But the problem is there are almost no precipitation networks at higher elevations (except few from EVK2-CNR, Italian group). I thought it would be better if we could get some gridded reanalysis product, bias correct/spatial downscale it and generate some product which could be used in operational hydrology. But i agree that the job is quite challenging.

Amrit - Since I do global precip estimates, I can say that the analysis schemes have a hard time handling the fine-scale and difficult-for-satellites structure that very mountainous scenes possess.  On the other hand, pretty much every precip technique is at its worst in this region, with different reasons for each approach.  So, you have to get creative.  The downscaling  idea has potential, but you still need to do validation.  Do you have, or maybe you can find, stream flow data?  With work, it should be possible to put bounds on what the precip should be, at least in an annual sense.

I would make a few points about "rain" gauges:

1. They have to be all-weather gauges; many are just rain gauges and don't correctly handle snow, which is a big component of the annual precip.  I recently reviewed a paper for a basin to the west of your region and the authors were quite casual about it.

2. As you interact with gauge data providers, it is always good to check whether their data are going into the GPCC and/or APHRODITE archives.  We get better answers when people share their data!

3. Raw precip gauge reports are not "truth".  Pretty much every style of gauge under-reports precipitation, more strongly in windier conditions and when the precip is snow.  The GPCC and Canadians seem to be the leaders in practical correction to this "undercatch" problem.  It would be interesting to see what DHM is doing.

I would be interested to see how you progress with this important problem!

George

Dear George,

Thank you for your information and suggestion. I really appreciate your views. Yes we also have river runoff data. I will come to you once i start my work.

I heard that APHRODITE group in coordination with DHM, Nepal is making gridded temp., precip product till 2014. Again the problem is its coarser resolution (0.25*0.25).

We really do not know what happened above 5000 or say 6000m above the sea level. We really need at least some measurements above these altitudes to better understand precipitation dynamics in Himalayas.

Thanking you once again.

With my experience, apart from station and gauge data, at the moment for Himalayan region APHRODITE and APHROTEMP is the best gridded data: both because of station information involved and its vertical and horizontal gradient involved at the time of preparing the synthetic gridded data. Please see Yatagai et al.

Thank you Prof. Dimri. I will go through recommended paper.

Dear siingh,

the TRMM resolution seems poor for our application. Do you have any other product in mind?

Over foothills of Himalayas IMD gauge observations are merged to prepare a daily satellite gauge merged products. The data is available from IMD Pune website. However, number of gauges over Himalayan region could be very limiting.

Dear Thapa

You can use ERA-Interim reanalysis data after its topographic correction as described in this paper - Elevation correction of ERA-Interim temperature data in theTibetan Plateau by Gao et al. 

you can use gauge data as a reference. 

best wishes

Dear thapa,

Could you be more clear why you need very fine resolution data in order to calculate trend?? You can get overall trend for a region even if there are few grid points. 

Dear Choudhary sir,

Yeah i went through the paper. It was good. I was able to extract the vertical distribution of temperature at different pressure level. Now the problem is how to get the corresponding elevation of atmospheric level. Do you have some experience with this?

regards from Nepal

Dear Rai,

I need high resolution data-set as i am interested in doing analysis at different elevation band. If i use course resolution data, i will end up with only few pixels at some elevation bands.

Yead i have already calculated overall trend by averaging the grids in my area of interest. But my interest is more in calculating trend at different elevation. Trying to understand the magnitude of trend at different elevation.

Regards from Nepal

Amrit,

While going through round and round-why don't you take available gridded data sets (reanalysis and observations) over the Himalayas and come up with conclusion which is best?

That will be of great help to climate researchers over the Himalayas.

Best

Dimri

Dear Thapa,

You can calculate or better say estimate pressure-level elevations (in metres above sea level) in a well known standard way by normalising geopotential heights by gravity at sea level. you can use era-interim geopotential field for this purpose. do you remember phi(h) = phi/9.8

best wishes

Thank you Prof. Dimri and Anubhav sir 

I will proceed and let you know if something sounds interesting.

We have a comparison study of gridded rainfall products over India. Please see this article. http://www.sciencedirect.com/science/article/pii/S0022169417300240. A comparison of reanalyses products (a different study) reveals that MERRA-2 is better in capturing precipitation pattern over India. I hope the results could be similar. 

There are plenty of precipitation data sets available for Himalayan region. These data sets include satellite, rain gauge based as well as combined satellite and rain gauge products. However, accuracy of these data sets are questionable due to complex topography of the region. A suitable data set can be suggested if you can explain the exact nature of application you are interested in (climate study or meteorological application???).

Dear Anoop Jee,

My interest is more on elevation wise precipitation trend and trying to understand its relation with snow.

Dear Amrit,

Reanalysis data is not reliable for precipitation monitoring. Rain gauge based gridded products are also not very accurate due to poor density and limitations of precipitation monitoring over complex terrain. I think satellite based  global precipitation product GPCP can be used for your study. Refer to the following page

ftp://meso.gsfc.nasa.gov/pub/1dd-v1.2/

One study was conducted over Hindu-Kush using data from Satellite, reanalyses and rain gauges. http://onlinelibrary.wiley.com/doi/10.1029/2012JD018697/pdf

Just be careful when you say precipitation and/or snow. There in itself great uncertainty when a liquid becomes solid precipitation over the Himalayas.

Hi,

May be this article provide you a bit more clarity!

http://dx.doi.org/10.1002/2015GL063392

Reanalysis data may show large errors in precipitation. Primarily due to lack of observational network and complex topography of the Himalaya. High resolution simulation of precipitation over the Himalaya show large errors.

Dear D Singh, It will be good if you share some reference for your statement that "Reanalysis data may show large errors in precipitation."

Dear Amrit,

Precipitation is very hard to assess in high mountain ranges, basically because of the substantial effects of snow drifting by wind, a process which substantially exacerbates direct measurement, reanalysis, and modeling.

The "High Asia Refined Analysis" is a sophisticated data set focusing on your region of interest (s. link 1, data is freely available after registration), also including estimations of the amount of snowfall. Curio and Scherer (2016, link 2) demonstrated that relatively detailed inferences regarding precipitation patterns in High Asia are possible basing on this data set.

Nevertheless, keep in mind that there are huge uncertainties in any such precipitation estimation and be very careful with trend analyses basing on reanalysis data (which tends to be biased, e.g. link 3).

Best wishes,

David

http://www.klima.tu-berlin.de/index.php?show=forschung_asien_tibet_har&lan=en

https://doi.org/10.5194/esd-7-767-2016

http://onlinelibrary.wiley.com/doi/10.1029/2004JD004536/full

Dear Dr Pankaj,  One of the key issues in understanding precipitation variability over the Himalaya is lack of adequate observational network besides complexity of topography.  Without assimilating observed precipitation data  into reanalysis data  may lead to large errors. Many papers have been published on the issue of precipitation simulation on the mountainous topography. Even if precipitation data is assimilated, the errors still can be large due to topographic variations.

Hi Amrit,

You can try MSWEP data(http://www.gloh2o.org/)

It's a merge product(satellite, gauge, re-anlaysis). Hence, i am expecting less uncertainty as compared with each individual product.

Dear Ashish Jee,

Thank you for your suggestion. I had a look at this product some time back and running some hydrologial model with this dataset. Ashish jee, just one more query. Have you done bias correction for reanalysis/ satellite based precipitation. If so, how you handle zero values during the process.

Hi Amrit,

Several methods are present to correct reanalysis/satellite product such as Quantile mapping, Bayesian approach, Moving average....One of the major problem to correct the product is the presence of Miss/ False Precipitation (Zero value) present in the data-sets . I would suggest you to reduce the Miss/ False Precipitation before the correction method applied. You can reduce Miss/ False Precipitation by Thresholding.

The other option is to consider only Hit Event (Both satellite as well Observation indicates the rainfall) for the bias correction. If you have any other queries mail me on this id

[email protected].

Dear Ashish Jee,

Thank you for the information. I will write you soon.