My adviser has studied this topic in several different species. He has many of this publications posted for public access on his personal webpage: http://calmit.unl.edu/people/agitelson2/

My experience is that you can use visible/NIR reflectance to estimate yield (although you can use others as suggested above). As for spatial resolution, I am not knowledgeable of your crop (e.g. how large it is, are there plants planted underneath the canopy, how homogenous an area containing cocca, etc.). All these factors influence the necessity of spatial resolution.

I don't know what kind of data you have access to, but generally you can download MODIS data for free. It has decent spatial and spectral resolution (but there are better sensors). Here is a paper looking at estimating Maize yield which might give you an idea on what to try: http://calmit.unl.edu/people/agitelson2/pdf/2012/2012-AFM-LAI.pdf

As for specific vegetation indices (VI) there are many to choose from. My experience has been that VIs like NDVI are great at low values of biomass/green leaf area index. However, they do saturate and you can use a different vegetation index like simple ratio. You can see my publication for more details along with some other VIs not listed in the paper above: https://www.msu.edu/~vina/2012_AJ_CVI.pdf

If you have trouble accessing these pdfs, send me a message and I can email them to you.

I repeat Yunpeng's question. What do you mean by shaded? Are you trying to determine yield of a crop planted under another? This complicates matters and may require more specialized bands that can penetrate deeper into the canopy (e.g. red edge) thus spectral resolution may also be important. Is it artificially shaded? This might require understanding the spectral characteristics of the shading apparatus. As you can see there are many factors that must be considered.

To my understanding, it is possible. Remote sensing has been used to monitor productivity of many crops like rice and corn. The method is similar for cocoa. What do you mean "shaded"? do you mean the cocoa is planted in shadow areas?

OK

Various RS Techniques can be used to monitoring productivity of crops. Hyperspectral and optics images generated from various spacial platform are very interesting techniques.

Best regard from Valparaíso en Chile

2.5 meter resolution multispectral remote sensing data helps to map on 1 to 12,500 scale or better. "What is the plot size" is the issue and should fall with in the scope of resolution mentioned above. Productivity is an issue related to modeling. Ground truth plays very important issue in monitoring the plantation and estimating the productivity. .

I've worked, also during my Ph.D., using optical data but I know also the active sensors (microwave sensors). From my point of view, the vegetation (COCOA PLANTS in this cases) can be monitored using data from meteorological sensors, combining reflectance channels (creating VEGETATION INDEXES). Actually, I'm working with a collegue for a project and we are using these vegetation indexes. I think, therefore, that it's possible to do also to COCOA PLANTS. If you wish to knew something more, you can ask me a consulting and I'll help you (not for free). Best regards. Dr. Casciello Daniele.

Yield estimation by remote sensing is efficient to the crops that the yield is related to plant biomass, that I don´t know if it is the case of cocoa crop. You have to find this kind of information and have in mind that field sampling are offen necessary.

My adviser has studied this topic in several different species. He has many of this publications posted for public access on his personal webpage: http://calmit.unl.edu/people/agitelson2/

My experience is that you can use visible/NIR reflectance to estimate yield (although you can use others as suggested above). As for spatial resolution, I am not knowledgeable of your crop (e.g. how large it is, are there plants planted underneath the canopy, how homogenous an area containing cocca, etc.). All these factors influence the necessity of spatial resolution.

I don't know what kind of data you have access to, but generally you can download MODIS data for free. It has decent spatial and spectral resolution (but there are better sensors). Here is a paper looking at estimating Maize yield which might give you an idea on what to try: http://calmit.unl.edu/people/agitelson2/pdf/2012/2012-AFM-LAI.pdf

As for specific vegetation indices (VI) there are many to choose from. My experience has been that VIs like NDVI are great at low values of biomass/green leaf area index. However, they do saturate and you can use a different vegetation index like simple ratio. You can see my publication for more details along with some other VIs not listed in the paper above: https://www.msu.edu/~vina/2012_AJ_CVI.pdf

If you have trouble accessing these pdfs, send me a message and I can email them to you.

I repeat Yunpeng's question. What do you mean by shaded? Are you trying to determine yield of a crop planted under another? This complicates matters and may require more specialized bands that can penetrate deeper into the canopy (e.g. red edge) thus spectral resolution may also be important. Is it artificially shaded? This might require understanding the spectral characteristics of the shading apparatus. As you can see there are many factors that must be considered.

Yes, It is possible to use remote sensing data (satellite and ground sensors) Vegetation indices such as NDVI from satellite will be related to ground based sensors data of Leaf Area Index (LAI), to track inter annual phenology of cocoa then multivariate time series analysis of data derived from satellite and ground based sensors will be related to annual cocoa yield and annual meteorological data to forecast the inter annual productivity /yield.

Also you need to take into consideration the temporal and spatial resolution of the satellite data and in this case MODIS NDVI product will suffice and it is available on NASA website free of charge.

Please change the topic title to-Using remote sensing techniques to monitor cocoa productivity...... in a selected plantation or selected plantations

Please remove the word shaded because no plant will grow normally under shade

Like coffee, cocoa wants to grow in the shade. The cocoa trees are rather squat and unassuming looking and need the tall canopy of the larger trees. (http://www.candyblog.net/blog/item/cocoa_plantations_the_cornerstone_of_preserving_biodiversity/).

You may use hyperspectral remote sensing images for your study, if your plot size is small.

If the cocoa plants are under shade, it would be difficult to use satellite based data for information of the plants. Resolution will depend on the size of the plots.

SAR may be better, but I do not have any experience with radar imagery so I will not make any suggestions regarding this type of remote sensing. I do know that bands like the Red Edge (e.g. 709 nm in MERIS) have deeper penetration into the canopy, unfortunately few satellite sensors have a red edge band (not to mention the ENVISAT carrying MERIS died earlier this year). The good news is that there are several that are being launched in the next couple of years (Sentinel-2 with MSI and Sentinel-3 with OLCI).

Hyperspectral imagery may work, but it can be very expensive, especially for long-term monitoring. Unfortunately for predicting yield is is better to have larger data sets since yield is influenced by many factors and you only get 1 number per season. I would suggest a study utilizing close range sensors either spectral (Ocean Optics or ASD) or multispectral (SKYE or LED instruments you can build yourself) if you use the correct bands. This way you can determine if you can 'see' through the canopy (and possibly use satellite data) or if you will need to have a system of sensors monitoring the cocoa below the canopy which will require either a large number of instruments or manpower to monitor large plantations. Of course you could minimize this need if you knew phenology and the time period to best monitor the plants for an accurate yield estimate (e.g. around tasseling for maize). I think you have your work cut out for you. Good luck!

I wish to appreciate all scholars who have contributed to this question. Indeed, your suggestions have gone a long way to improve my understanding of the proposed work. I must however wish to confirm whether spectral signatures in remote sensing data could discriminate cocoa plantation of small sizes from other land use systems. In developing nations, most cocoa agroforestry systems/plantations are under natural forest which serve as shade to young cocoa plantations. Largely, cocoa plantations are owned mostly by smallholder farmers and these plots are scattered across agricultural landscape. Which vegetation index would be most appropriate to discriminate between these scattered plantation , protected area, fallow ground, annual cultivated land etc.

NDVI will be appropriate however SAVI is better but when doing a research you need to take into consideration the technical details and the availability of equipment plus financial resources at your disposal/country where the research will be carried out, i suspect the research will be carried out in Nigeria. For ground based measurement of LAI, hemispherical photography will be OK.Though there are other equipments like CASI, Sunscan,LIDAR/laser scanner this may be too expensive for you.Also the best option for space borne data is to use optical systems because non-optical systems (radar imagery) data for your area of study in Nigeria is not available but with high temporal resolution data of TERRA MODIS NDVI (a bimonthly composite of daily images) and the spatial resolution is 250m and i am sure that any cocoa plantation in Nigeria is more than 250m. I still maintain that you remove the word shaded from the title to avoid confusion, it will be better explained in the content this is because scientifically crops does not grow "normally" under shade .

Thanks Mr. Yemi Ogunbadewa for your suggestions, they are wonderful and tailor-made. However, are you sure that we might not have a plantation that its less than 6.25ha (250m X 250m)? This especially in South western Nigeria, where farm holding are small and fragmented due to inheritance. Why not landsat7 etm+, which has 30m spatial resolution and 26days of temporal resolution with the combination of infrared and red bands? However, Landsat may be problematic and inaccessible for some region due to SL error, but this could be overcome. For SAVI see the Idrisi manual for equation and methods.

Another approach to fill the data gaps between LandSat images is a data fusion approach using both MODIS and LandSat. I don't know what your computer capabilities are, but here are some papers describing the technique:

http://www.sciencedirect.com/science/article/pii/S0034425708001065

http://www.sciencedirect.com/science/article/pii/S003442570900087X

http://www.sciencedirect.com/science/article/pii/S0034425709001709

http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=1661809&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D1661809

http://spie.org/x648.html?product_id=981596

http://spie.org/x648.html?product_id=965055

Sorry I am not an expert in image processing so I may not be able to address specific questions regarding this technique, but I suggest contacting the corresponding author(s).

Anthony Nguy-Roberton, thanks so much. I am really grateful for the efforts. To all, thanks and thanks.

I am affraid that LandSat Satellite System had been finished last year. There are no more images available.

LandSat 5 failed, but LandSat 7 is still operational. While there is a scan line error which causes loss of data, using the data fusion technique it may be possible to still 'recreate' the data. Although, I must admit there will be larger data gaps compared to if you had a fully functional LandSat satellite which will likely reduce the accuracy of your results.

If is in shadow is very difficult, because remote sensor will be sensing the top canopy