It is impossible to dispute the genuine value of a well-built and reliable resource model. Developed on a solid geological foundation, with a comprehensive understanding of estimation domains and reinforced by robust geostatistics, the resource model serves as an excellent foundation for long-term mine planning. Moreover, it serves as the starting point for the reconciliation process. At the end of the day, the corporate financial teams typically focus on a single number, F3, which is closely tied to the performance of the resource model. This performance is just as crucial as the operational practices and hard work put in by the metallurgists at the plant. But what if your resource model isn't willing to cooperate and reflect the similar in-situ reality that your grade control model shows (assuming you are in the meat of your mineralization and having good representative volume for the comparison)? This scenario applies, of course, if your grade control model makes sense (see the Recon #4 Post), although generally, it is less sensitive and more difficult to mess up given the quantity of data. I'm not inclined to delve into the intricacies of all the best practices applied to resource modelling (Erik Ronald already has an excellent series of posts on the topic). Instead, let's focus on the key reasons why your resource model may not behave and consistently disrupt your reconciliation results. Datasets 1) Start by ensuring that your datasets convey the same picture. Easily achieve this by compositing your exploration dataset at the blasthole length and conducting pair analysis for closely spaced samples (perhaps 3-10 meters, depending on your drilling grids). A QQ plot will likely reveal any biases in your data. 2) For a more sophisticated approach, consider twinning some exploration holes with blastholes. Collect samples of the same size; though more labor-intensive, this method may yield more representative results. The same principle applies to the RC grade control method. 3) If bias is detected, unravelling the reasons behind it and eliminating them becomes a paramount task for any mining geologist. Both datasets can suffer from deficiencies in the drilling process (e.g., flushing fine gold with water), sample collection (when was the last time you checked the quality of the sample cone cutting in the pit?), sample preparation (ranging from sloppy splitting to consistent contamination), and analytical methods (have you considered differences in digestion methods? is you mine lab as reliable as the certified one?). Domains 1) Once you've confirmed that the data is clear, the next major culprit is the interpretation or the quality of your resource estimation domains. Depending on the deposit type and style of mineralization, they may be either too broad or too restrictive, developed with only geology in mind or as sterile grade shells, the possibilities are numerous. 2) However, while interpretations can vary (two geos – three opinions 😊), a simple visualization of your estimation domains versus the grade control data can provide insights into how your long-term interpretation aligns with dense drilling (it should!). Naturally, domains developed from sparse drilling may not precisely match the grade trends from blastholes, but they also shouldn't differ significantly. If they do, understanding the reasons and considering possible remodelling should help address the issue. Estimation 1) If your data is OK and the narrative conveyed by your domains aligns with that revealed by blastholes, you likely already have a close alignment in your Resource/Grade Control Model metrics. However, if consistent differences persist, it may be time to scrutinize the details of the estimation assumptions. 2) You can do this simply by visually comparing sections of the block models (typically, the resource model provides a story at a coarse scale while the GC model offers a full HD) or by putting effort into overlapping the models and calculating grade differences (usually, the block size for the GC model is smaller, requiring upscaling to account for the change-of-support effect). 3) Once you identify the areas causing the most issues, the focus shifts to uncovering the underlying reasons. There could be various factors at play: overly aggressive (or conservative) capping, insufficient drilling density leading to high-grade bleeding (consider adjusting capping and/or using restricted search), or a search that is either too smooth or too variable. Once the reason is pinpointed, tweaking the estimation parameters may be justified to better reflect the observed patterns in the GC model. These practical tips provide guidance on addressing persistent biases between the models. Feel free to share your experiences or challenges in the comments. Source: https://www.linkedin.com/posts/aleksandr-mitrofanov-phd-pgeo-9473ba6a_mining-mininggeology-resourcegeology-activity-7157433534163402752-uDWJ?utm_source=share&utm_medium=member_desktop