Development of large scale water resource monitoring strategy using remote sensing

[Whizz]

Hi there,

I need some advice from a professional, who has experience in applying remote sensing (RS) data in hydrology and water management.

I am working on the development of large scale water resource monitoring strategy by combing remote sensing and local ground based measurements in semi-arid region. Let’s say I have a catchment of 1000 km2 and the hydrological processes of interest are: precipitation, soil moisture movements and ground water recharge.

To begin with, I need to decide on appropriate spatial and temporal resolution of RS data for this study. Considering the high spatiotemporal variability of the processes mentioned above and performance capabilities of current remote sensing sensors, I have concluded that the “trade-off solution” would be to extract quantitative information on precipitation, soil moisture movements and ground water recharge with daily temporal granularity and 500m/1 km spatial resolution.

When it comes to precipitation rates, I can simply download ready to use TRMM products (such as 3B42). However, these products at coarse resolution (25 km) pose a significant challenge in terms of accuracy assessment and unrepresentativeness of the study area. I can try to overcome this problem by using downscaling techniques based on NDVI from MODIS or AVHRR, but I am not sure as to applicability of this method, since response of NDVI to rainfall event is in the order of 1-3 weeks.

There are numerous algorithms for soil moisture retrievals using thermal, passive microwave and active microwave data with their pros and cons. The extraction of soil moisture from daily LST MODIS products with 1 km resolution is quite straightforward approach, but the quality of such soil moisture output is far from satisfactory. Another option would be to buy a few scenes from RADARSAT and use it in synergy with MODIS to get daily distribution of soil moisture. Passive microwave data is not an option for this study due to its low spatial resolution.

Finally, an extraction of groundwater recharge is a complicated task for RS community on regional scale. I’ve heard about GRACE/GOCE success in this field, but their 100 km cell size is of no use for me. I also came across a few papers where the vegetation was used as a proxy for distribution of groundwater recharge, but usually in qualitative terms. This, unfortunately, leaves me with the estimation of ground water recharge using, for example, SWAT model.

Does anyone have suggestions as to how to overcome the problems I have mentioned above or some ideas as to how to improve the methodology? The question of temporal and spatial resolution of the data for this project is still an open question. I might end up looking on monthly averages. As I said before, I will be able to buy some scenes from RADARSAT, TanDEM-X or some other imagery as well as to perform ground-truthing of soil moisture estimates.

[3dbu]

is great the chalenge that you'll talke, the first comment is that 1000 km2 is not enough big area and there you can make many process for get your data that really need, if have the money and the time you can make a set of sample for measurement the soil moisture but any time , the first is make a kind of fishnet over the area and make in field the measurement , other change for precipitation rate is get meteorology station by the zone and make analyst of this local data , will possible use SWAT too....if have global data is possible too make a downscaling analyst of it and see the potential if is possible to use....there are many treat that you can make...good luck

[Nikolaich]

RS data usually give soil moisture only for the upper 5 centimeters or even less depending on vegetation type. So this data is a main source of uncertainty in hydrology modeling.

If you have only 1000 km2 it is better use discharge measurements as an indicator of average soil moisture on the watershed and then map it back according to topography based on Topmodel theory.